The future of ideas: the fate of the commons in a connected world

Authors Lawrence Lessig

Plaintext

less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page i

THE FUTURE OF IDEAS
This work is licensed under a Creative Commons
Attribution-Noncommercial License (US/v3.0).

Noncommercial uses are thus permitted without any
further permission from the copyright owner.

Permissions beyond the scope of this license are
administered by Random House. Information on how
to request permission may be found at:

http://www.randomhouse.com/about/
permissions.html

The book maybe downloaded in electronic form
(freely) at:

http://the-future-of-ideas.com

For more permission about Creative Commons licenses,
go to:

http://creativecommons.org
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page iii

the future of ideas
T H E FAT E O F T H E C O M M O N S

IN A CONNECTED WORLD

///

Lawrence Lessig

f
R A N D O M H O U S E

N e w Y o r k
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page iv

All rights reserved under International and Pan-American
Copyright Conventions. Published in the United States by
Random House, Inc., New York, and simultaneously in
Canada by Random House of Canada Limited, Toronto.

Random House and colophon are registered trademarks of Random House, Inc.

library of congress cataloging-in-publication data
Lessig, Lawrence.
The future of ideas : the fate of the commons in a connected world / Lawrence Lessig.
p. cm.
Includes index.
ISBN 0-375-50578-4
1. Intellectual property. 2. Copyright and electronic data processing.
3. Internet—Law and legislation. 4. Information society. I. Title.
K1401 .L47 2001
346.04'8'0285—dc21 2001031968

Random House website address: www.atrandom.com

Printed in the United States of America on acid-free paper

2 4 6 8 9 7 5 3

First Edition

Book design by Jo Anne Metsch
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page v

To B e t t i n a ,

m y t e a c h e r

o f t h e m o s t i m p o r t a n t

l e s s o n .
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page vi
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page vii

P r e f a c e

I n 1 9 9 9 , in a book entitled The Control Revolution, journalist and legal
scholar Andrew Shapiro described two futures that the Internet might
take.1 The first was the familiar story of increased individual freedom, as the
network gave us greater control over our lives, and over the institutions, in-
cluding government, that regulate our lives. The second was a less familiar
warning—of the rebirth of technologies of control, as institutions “dis-
intermediated” by the Internet learned how to alter the network to reestab-
lish their control.
Shapiro saw good and bad in both futures. Too much dis-intermediation,
he warned, would interfere with collective governance; some balance was
needed. But likewise, efforts to rearchitect the Net to reenable control
threatened to undermine its potential for individual freedom and growth.
Shapiro did not predict which future would be ours. Indeed, his argu-
ment was that bits of each future were possible, and that we must choose a
balance between them. His account was subtle, but optimistic. If there was
a bias to the struggle, he, like most of us then, believed the bias would favor
freedom.
This book picks up where Shapiro left off. Its message is neither subtle
nor optimistic. In the chapters that follow, I argue that we are far enough
along to see the future we have chosen. In that future, the counter-
revolution prevails. The forces that the original Internet threatened to trans-
form are well on their way to transforming the Internet. Through changes in
the architecture that defined the original network, as well as changes in the
legal environment within which that network lives, the future that promised
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page viii

v i i i P R E F A C E

great freedom and innovation will not be ours. The future that threatened
the reemergence of almost perfect control will.
I don’t mean the control of George Orwell’s 1984. The struggle that I de-
scribe here is not between free speech and censorship, or between democ-
racy and totalitarianism. The freedom that is my focus here is the creativity
and innovation that marked the early Internet. This is the freedom that
fueled the greatest technological revolution that our culture has seen since
the Industrial Revolution. This is the freedom that promised a world of cre-
ativity different from the past.
This freedom has been lost. With scarcely anyone even noticing, the net-
work that gave birth to the innovation of the 1990s has been remade from
under us; the legal environment surrounding that network has been impor-
tantly changed, too. And the result of these two changes together will be an
environment of innovation fundamentally different from what it was, or
promised to be.
Or so it will be unless we do something now. Unless we learn something
important about the source of that creativity and innovation, and then pro-
tect that source, the Internet will be changed.
With dot.busts all around, it is not difficult to argue that this is the winter
of the Internet’s life. The question for us is whether the spring will be as
silent.

a book like this does not emerge from a library. It has instead been written
through hundreds of conversations over many years. I am a law professor,
but my argument spans computer design to economics. It is no doubt fool-
ish for anyone to try to pull together such a range of material, but I could
never have dared to be so foolish without the patient tutoring of many dif-
ferent people. Among these, I am most grateful to my colleagues at the
Electronic Frontier Foundation, including John Gilmore and John Perry
Barlow; and the Center for Public Domain, especially Laurie Racine and
Bob Young. Jeff Chester of the Center for Media Education and Mark
Cooper of the Consumers Union taught me a great deal about media
policy and the passion of this struggle. There is a long list of technical ex-
perts who have struggled to show me how the network works. Among these
I am most grateful to Hal Abelson, Scott Bradner, Ben Edelman, Dewayne
Hendricks, Joseph Reagle, David P. Reed, and Jerome Saltzer. Dewayne
Hendricks and David P. Reed helped me understand spectrum and, more
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page ix

P R E F A C E i x

important, the potential spectrum offered. Peter Huber helped me under-
stand telephone companies and the very different potential they offered.
I am grateful as well to an extraordinary collection of law professors, who
have built in the field of cyberlaw an amazing community. James Boyle’s
book Shamans, Software, and Spleens was my first introduction to the issues
that I address here; James Boyle the person has been a steady, invaluable
guide since. Jack Balkin, Yochai Benkler, Mark Lemley, Jessica Litman,
David Post, and Pam Samuelson have all taught me far more than I could
return to them.
My work on this book began at Harvard Law School’s Berkman Center.
The theme was born in the passionate rants of its extraordinary director,
Charlie Nesson. Through our work as colleagues, and on the Microsoft case
as well, Jonathan Zittrain helped me see how platforms matter. He has been
a constant, if neglected, friend throughout the development of the argu-
ment here.
I am also especially grateful to the hundreds of readers of The Industry
Standard who have reacted to the snippets of this book that I have woven
into columns for that magazine. While the furor of many of those readers is
sometimes hard to suffer, the insights and wisdom of many have been criti-
cal in re-forming the views I express here.
Finally, there is a collection of people who figure throughout the story of
this book, but who were more central to its writing than the text might reveal.
These are the figures who are truly fighting for a cause. Some of them are
quite well known—Richard Stallman, for example. Others are well known
among lawyers, at least—Dennis Karjala, Jessica Litman, Marc Rotenberg,
Pam Samuelson. But others inspire more through their simple and quiet
perseverance. Eric Eldred, whom you will meet in the course of these
pages, is the best example of this type. These ideas would never have been
put into words without the inspiration from people like him.

early versions of this book were read by a number of people. I
am grateful to those who offered critical (and sometimes especially critical)
comments—in particular Bruce Ackerman, Yochai Benkler, David Bollier,
Scott Hemphill, Dewayne Hendricks, Tom Maddox, Charles Nesson, Rich-
ard A. Posner, Barbara van Schewick, Timothy Wu, and Robert Young. My
research was aided by an army of students, including Amy Ash, Scott Ash-
ton, Aaron Bukofzer, Sky Canaves, Brian Gustafson, Drew Harris, Scott
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page x

x P R E F A C E

Hemphill, Matt Kahn, Matt Rice, Hilary Stockton, and Jonathan Sanders.
Pauline Reich, Hilary Stockton, and Richard Taketa contributed examples
to the text. Chris Guzelian was especially helpful in bringing the book to
closure, through both his research and a careful and talented final edit. Bet-
tina Neuefeind, however, remains the world’s greatest editor.
I am particularly grateful to Elisa Garza Kammeyer for her work through-
out this last year, first as a researcher and finally as an assistant. She will
prove to be the one truly famous person mentioned in this book, though
that is a story that will take many years to unfold.
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page xi

/ / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

C o n t e n t s

Preface vii

1: “Free” 3

PA R T I : D O T. C O M M O N S 17

2: Building Blocks: “Commons” and “Layers” 19
The Commons 19
Layers 23
3: Commons on the Wires 26
4: Commons Among the Wired 49
5: Commons, Wire-less 73
6: Commons Lessons 85

PA R T I I : D O T. C O N T R A S T 101

7: Creativity in Real Space 103
Creativity in the Dark Ages 104
The Arts 104
CONTENT 105
PHYSICAL 110
CODE 111
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page xii

x i i C O N T E N T S

Commerce 112
CODING 112
MARKETS 114
8: Innovation from the Internet 120
New Products from the Net 122
HTML Books 122
MP3 123
Film 124
Lyric Servers and Culture Databases 124
New Markets 126
New Means of Distribution 126
My.MP3 127
Napster 130
New Demand 132
New Participation: P2P 134

PA R T I I I : D O T. C O N T R O L 143

9: Old vs. New 145
10: Controlling the Wires (and Hence the Code Layer) 147
The End-to-End in Telephones 149
Fat Pipe 151
AT&T Cable 153
11: Controlling the Wired (and Hence the Content Layer) 177
Increasing Control 180
Copyright Bots 180
CPHack 184
DeCSS 187
iCraveTV 190
MP3 192
Napster 194
Eldred 196
Consequences of Control 199
12: Controlling Wire-less (and Hence the Physical Layer) 218
13: What’s Happening Here? 234
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page xiii

C O N T E N T S x i i i

14: Alt. Commons 240
The Physical Layer 240
Free Spectrum 241
Free Highways 244
The Code Layer 246
Neutral Platforms 246
The Content Layer 249
Copyright 250
F I V E - Y E A R R E N E WA B L E TERMS 251
S O F T WA R E C O P Y R I G H T 252
P R O T E C T I N G I N N O VAT I O N 253
PROTECTING MUSIC 254
R E B U I L D I N G T H E C R E AT I V E C O M M O N S 255
LIMITS ON CODE 256
LIMITS ON CONTRACT 257
L I M I T C O M M E R C I A L E X P L O I TAT I O N 258
Patents 259
M O R AT O R I U M 259
DAMAGES 260
REFORM 260
15: What Orrin Understands 262
Notes 269

Index 335
less_0375505784_4p_fm_r1.qxd 9/21/01 13:49 Page xiv
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 1

THE FUTURE OF IDEAS
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 2
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 3

“ F r e e ”

D avis guggenheim is a film director. He has produced a range of
movies, some commercial, some not. His passion, like his father’s before,
is documentaries, and his most recent, and perhaps best, film, The First
Year, is about public school teachers in their first year of teaching—a Hoop
Dreams for public education.
In the process of making a film, a director must “clear rights.” A film
based on a copyrighted novel must get the permission of the copyright
holder. A song in the opening credits requires the permission of the artist
performing the song. These are ordinary and reasonable limits on the cre-
ative process, made necessary by a system of copyright law. Without such a
system, we would not have anything close to the creativity that directors
such as Guggenheim have produced.
But what about the stuff that appears in the film incidentally? Posters on
a wall in a dorm room, a can of Coke held by the “cigarette smoking man,”
an advertisement on a truck in the background? These too are creative
works. Does a director need permission to have these in his or her film?
“Ten years ago,” Guggenheim explains, “if incidental artwork . . . was
recognized by a common person,” then you would have to clear its copy-
right. Today, things are very different. Now “if any piece of artwork is recog-
nizable by anybody . . . then you have to clear the rights of that and pay” to
use the work. “[A]lmost every piece of artwork, any piece of furniture, or
sculpture, has to be cleared before you can use it.”1
Okay, so picture just what this means: As Guggenheim describes it,
“[B]efore you shoot, you have this set of people on the payroll who are sub-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 4

4 L a w r e n c e L e s s i g

mitting everything you’re using to the lawyers.” The lawyers check the list
and then say what can be used and what cannot. “If you cannot find the
original of a piece of artwork . . . you cannot use it.” Even if you can find it,
often permission will be denied. The lawyers thus decide what’s allowed in
the film. They decide what can be in the story.
The lawyers insist upon this control because the legal system has taught
them how costly less control can be. The film Twelve Monkeys was stopped
by a court twenty-eight days after its release because an artist claimed a chair
in the movie resembled a sketch of a piece of furniture that he had de-
signed. The movie Batman Forever was threatened because the Batmobile
drove through an allegedly copyrighted courtyard and the original architect
demanded money before the film could be released. In 1998, a judge
stopped the release of The Devil’s Advocate for two days because a sculptor
claimed his art was used in the background.2 Such events teach the lawyers
that they must control the filmmakers.3 They convince studios that creative
control is ultimately a legal matter.
This control creates burdens, and not just expense. “The cost for me,”
Guggenheim says, “is creativity. . . . Suddenly the world that you’re trying to
create is completely generic and void of the elements that you would nor-
mally create. . . . It’s my job to conceptualize and to create a world, and to
bring people into the world that I see. That’s why they pay me as a director.
And if I see this person having a certain lifestyle, having this certain art on
the wall, and living a certain way, it is essential to . . . the vision I am trying
to portray. Now I somehow have to justify using it. And that is wrong.”

this is not a book about filmmaking. Whatever problems filmmakers
have, they are tiny in the order of things. But I begin with this example be-
cause it points to a much more fundamental puzzle, and one that will be
with us throughout this book: What could ever lead anyone to create such a
silly and extreme rule? Why would we burden the creative process—not
just film, but generally, and not just the arts, but innovation more broadly—
with rules that seem to have no connection to innovation and creativity?
Copyright law, law professor Jessica Litman has written, is filled with
rules that ordinary people would respond to by saying, “There can’t really be
a law that says that. That would be silly.”4 Yet in fact there is such a law,
and it does say just that, and it is, as the ordinary person rightly thinks, silly.
So why? What is the mentality that gets us to this place where highly edu-
cated, extremely highly paid lawyers run around negotiating for the rights to
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 5

T H E F U T U R E O F I D E A S 5

have a poster in the background of a film about a frat party? Or scrambling
to get editors to remove an unsigned billboard? What leads us to build a
legal world where the advice a successful director can give to a young artist
is this:

I would say to an 18-year-old artist, you’re totally free to do whatever you
want. But—and then I would give him a long list of all the things that he
couldn’t include in his movie because they would not be cleared, legally
cleared. That he would have to pay for them. [So freedom? Here’s the free-
dom]: You’re totally free to make a movie in an empty room, with your two
friends.5

a time is marked not so much by ideas that are argued about as by ideas
that are taken for granted. The character of an era hangs upon what needs
no defense. Power runs with ideas that only the crazy would draw into
doubt. The “taken for granted” is the test of sanity; “what everyone knows”
is the line between us and them.
This means that sometimes a society gets stuck. Sometimes these un-
questioned ideas interfere, as the cost of questioning becomes too great. In
these times, the hardest task for social or political activists is to find a way to
get people to wonder again about what we all believe is true. The challenge
is to sow doubt.
And so it is with us. All around us are the consequences of the most
significant technological, and hence cultural, revolution in generations.
This revolution has produced the most powerful and diverse spur to inno-
vation of any in modern times. Yet a set of ideas about a central aspect of
this prosperity—“property”—confuses us. This confusion is leading us to
change the environment in ways that will change the prosperity. Believing
we know what makes prosperity work, ignoring the nature of the actual pros-
perity all around, we change the rules within which the Internet revolution
lives. These changes will end the revolution.
That’s a large claim for so thin a book, so to convince you to carry on, I
should qualify it a bit. I don’t mean “the Internet” will end. “The Internet”
is with us forever, even if the character of “the Internet” will change. And I
don’t pretend that I can prove the demise that I warn of here. There is too
much that is contingent, and not yet done, and too little good data to make
any convincing predictions.
But I do mean to convince you of a blind spot in our culture, and of the
harm that this blind spot creates. In the understanding of this revolution
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 6

6 L a w r e n c e L e s s i g

and of the creativity it has induced, we systematically miss the role of a cru-
cially important part. We therefore don’t even notice as this part disappears
or, more important, is removed. Blind to its effect, we don’t watch for its
demise.
This blindness will harm the environment of innovation. Not just the in-
novation of Internet entrepreneurs (though that is an extremely important
part of what I mean), but also the innovation of authors or artists more gen-
erally. This blindness will lead to changes in the Internet that will under-
mine its potential for building something new—a potential realized in the
original Internet, but increasingly compromised as that original Net is
changed.
The struggle against these changes is not the traditional struggle between
Left and Right or between conservative and liberal. To question assump-
tions about the scope of “property” is not to question property. I am fanati-
cally pro-market, in the market’s proper sphere. I don’t doubt the important
and valuable role played by property in most, maybe just about all, contexts.
This is not an argument about commerce versus something else. The inno-
vation that I defend is commercial and noncommercial alike; the argu-
ments I draw upon to defend it are as strongly tied to the Right as to the Left.
Instead, the real struggle at stake now is between old and new. The story
on the following pages is about how an environment designed to enable the
new is being transformed to protect the old—transformed by courts, by
legislators, and by the very coders who built the original Net.
Old versus new. That battle is nothing new. As Machiavelli wrote in The
Prince:

Innovation makes enemies of all those who prospered under the old
regime, and only lukewarm support is forthcoming from those who would
prosper under the new. Their support is indifferent partly from fear and
partly because they are generally incredulous, never really trusting new
things unless they have tested them by experience.6

And so it is today with us: those who prospered under the old regime are
threatened by the Internet; this is the story of how they react. Those who
would prosper under the new regime have not risen to defend it against the
old; whether they will is the question this book asks. The answer so far is
clear: They will not.

* * *
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 7

T H E F U T U R E O F I D E A S 7

there are two futures in front of us, the one we are taking and the one
we could have. The one we are taking is easy to describe. Take the Net, mix
it with the fanciest TV, add a simple way to buy things, and that’s pretty
much it. It is a future much like the present. Though I don’t (yet) believe
this view of America Online (AOL), it is the most cynical image of Time
Warner’s marriage to AOL: the forging of an estate of large-scale networks
with power over users to an estate dedicated to almost perfect control over
content. That content will not be “broadcast” to millions at the same time;
it will be fed to users as users demand it, packaged in advertising precisely
tailored to the user. But the service will still be essentially one-way, and the
freedom to feed back, to feed creativity to others, will be just about as con-
strained as it is today. These constraints are not the constraints of econom-
ics as it exists today—not the high costs of production or the extraordinarily
high costs of distribution. These constraints instead will be burdens created
by law—by intellectual property as well as other government-granted exclu-
sive rights. The promise of many-to-many communication that defined the
early Internet will be replaced by a reality of many, many ways to buy things
and many, many ways to select among what is offered. What gets offered
will be just what fits within the current model of the concentrated systems
of distribution: cable television on speed, addicting a much more manage-
able, malleable, and sellable public.
The future that we could have is much harder to describe. It is harder be-
cause the very premise of the Internet is that no one can predict how it will
develop. The architects who crafted the first protocols of the Net had no
sense of a world where grandparents would use computers to keep in touch
with their grandkids. They had no idea of a technology where every song
imaginable is available within thirty seconds’ reach. The World Wide Web
(WWW) was the fantasy of a few MIT computer scientists. The perpetual
tracking of preferences that allows a computer in Washington State to sug-
gest an artist I might like because of a book I just purchased was an idea that
no one had made famous before the Internet made it real.
Yet there are elements of this future that we can fairly imagine. They are
the consequences of falling costs, and hence falling barriers to creativity.
The most dramatic are the changes in the costs of distribution; but just as
important are the changes in the costs of production. Both are the conse-
quences of going digital: digital technologies create and replicate reality
much more efficiently than nondigital technology does. This will mean a
world of change.
These changes could have an effect in every sphere of social life. Begin
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 8

8 L a w r e n c e L e s s i g

with the creative sphere, and let’s start with creativity off-line, long before
the law tried to regulate it through “copyright.”
There was a time (it was the time of the framing of our Constitution)
when creativity was essentially unregulated. As we’ll see in chapter 11, the
law of copyright effectively regulated publishers only. Its scope was just
“maps, charts, and books.” That meant every other aspect of creative life was
free. Music could be performed in public without a license from a lawyer; a
novel could be turned into a play even if the novel was copyrighted. A story
could be adapted into a different story; many were, as the very act of cre-
ativity was understood to be the act of taking something and re-forming it
into something (ever so slightly) new. The public domain was vast and
rich—the works of Shakespeare had just fallen from the control of publish-
ers in England; they would not have been protected in the United States
even if they had not.7
It’s not clear who got to participate in this creativity. No doubt social
norms meant that the right did not reach blindly across the sexes or races.
But the spirit of the times was storytelling, as a society defined itself by the
stories it told, and the law had no role in deciding who got to tell what sto-
ries. An old man fortunate enough to read might learn of the struggles with
pirates in the Gulf of Tripoli. He would retell this story to others in the town
square. A local troupe of actors might stage the struggle for patrons of a local
pub. If compelling, the troupe might move to the town next over and retell
the story.
It makes no sense to say that that world was “more creative” than ours. My
point is not about quantity, or even quality, and my argument does not imag-
ine a “golden age.” The point instead is about the nature of the constraints
on this practice of creativity: no doubt there were technical constraints on
it; no doubt these were important and real. But except for important subject
matter constraints imposed by the law, the law had essentially no role in say-
ing how one person could take and remake the work of someone else. This
act of creativity was free, or at least free of the law.
Skip ahead to just a few years in front of 2001 and think about the poten-
tial for creativity then. Digital technology has radically reduced the cost of
digital creations. As we will see more clearly below, the cost of filmmaking
is a fraction of what it was just a decade ago. The same is true for the pro-
duction of music or any digital art. Using what we might call a “music
processor,” students in a high school music class can compose symphonies
that are played back to the composer. Imagine the cost of that just ten years
ago (both to educate the composer about how to write music and to hire the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 9

T H E F U T U R E O F I D E A S 9

equipment to play it back). Digital tools dramatically change the horizon of
opportunity for those who could create something new.8
And not just for those who would create something “totally new,” if such
an idea is even possible. Think about the ads from Apple Computer urging
that “consumers” do more than simply consume:

Rip, mix, burn,

Apple instructs.

After all, it’s your music.

Apple, of course, wants to sell computers. Yet its ad touches an ideal that
runs very deep in our history. For the technology that they (and of course
others) sell could enable this generation to do with our culture what gen-
erations have done from the very beginning of human society: to take what
is our culture; to “rip” it—meaning to copy it; to “mix” it—meaning to re-
form it however the user wants; and finally, and most important, to “burn”
it—to publish it in a way that others can see and hear.9 Digital technology
could enable an extraordinary range of ordinary people to become part of a
creative process. To move from the life of a “consumer” (just think about
what that word means—passive, couch potato, fed) of music—and not just
music, but film, and art, and commerce—to a life where one can individu-
ally and collectively participate in making something new.
Now obviously, in some form, this ability predates digital technology. Rap
music is a genre that is built upon “ripping” (and, relatedly, “sampling”) the
music of others, mixing that music with lyrics or other music, and then
burning that remixing onto records or tapes that get sold to others.10 Jazz
was no different a generation before. Music in particular, but not just music,
has always been about using what went before in a way that empowers cre-
ators to do something new.11
But now we have the potential to expand the reach of this creativity to an
extraordinary range of culture and commerce. Technology could enable
a whole generation to create—remixed films, new forms of music, digital
art, a new kind of storytelling, writing, a new technology for poetry, criti-
cism, political activism—and then, through the infrastructure of the Inter-
net, share that creativity with others.
This is the art through which free culture is built. And not just through
art. The future that I am describing is as important to commerce as to any
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 10

1 0 L a w r e n c e L e s s i g

other field of creativity. Though most distinguish innovation from creativity,
or creativity from commerce, I do not. The network that I am describing en-
ables both forms of creativity. It would leave the network open to the widest
range of commercial innovation; it would keep the barriers to this creativity
as low as possible.
Already we can see something of this potential. The open and neutral
platform of the Internet has spurred hundreds of companies to develop new
ways for individuals to interact. E-mail was the start; but most of the mes-
sages that now build contact are the flashes of chat in groups or between
individuals—as spouses (and others) live at separate places of work with a
single window open to each other through an instant messenger. Groups
form easily to discuss any issue imaginable; public debate is enabled by
removing perhaps the most significant cost of human interaction—
synchronicity. I can add to your conversation tonight; you can follow it up
tomorrow; someone else, the day after.
And this is just the beginning, as the technology will only get better.
Thousands could experiment on this common platform for a better way;
millions of dot.com dollars will flow down the tube; but then a handful of
truly extraordinary innovations comes from these experiments. A wristwatch
for kids that squeezes knowingly as a mother touches hers, thirty miles away.
A Walkman where lovers can whisper to each other between songs, though
separated by an ocean. A technology to signal two people that both are avail-
able to talk on the phone—now. A technology to enable a community to de-
cide local issues through deliberation in virtual juries. The potential can
only be glimpsed. And contrary to the technology doomsayers, this is a po-
tential for making human life more, not less, human.
But just at the cusp of this future, at the same time that we are being
pushed to the world where anyone can “rip, mix, [and] burn,” a counter-
movement is raging all around. To ordinary people, this slogan from Apple
seems benign enough; to lawyers in the content industry, it is high treason.
To the lawyers who prosecute the laws of copyright, the very idea that the
music on “your” CD is “your music” is absurd. “Read the license,” they’re
likely to demand. “Read the law,” they’ll say, piling on. This culture that you
sing to yourself, or that swims all around you, this music that you pay
for many times over—when you hear it on commercial radio, when you
buy a CD, when you pay a surplus at a large restaurant so that it can play
the same music on its speakers, when you purchase a movie ticket where
the song is the theme—this music is not yours. You have no “right” to rip it,
or to mix it, or especially to burn it. You may have, the lawyers will insist,
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 11

T H E F U T U R E O F I D E A S 1 1

permission to do these things. But don’t confuse Hollywood’s grace with
your rights. These parts of our culture, these lawyers will tell you, are the
property of the few. The law of copyright makes them so, even though (as I
will show in the chapters that follow) the law of copyright was never meant
to create any such power.
Indeed, the best evidence of this conflict is again Apple itself. For the very
same machines that Apple sells to “rip, mix, [and] burn” music are pro-
grammed to make it impossible for ordinary users to “rip, mix, [and] burn”
Hollywood’s movies. Try to “rip, mix, [and] burn” Disney’s 102 Dalmatians
and it’s your computer that will get ripped, not the content. Software, or
code, protects this content, and Apple’s machine protects this code. It may
be your music, but it’s not your film. Film you can rip, mix, and burn only
as Hollywood allows. It controls that creativity—it, and the law that backs
it up.
This struggle is just a token of a much broader battle, for the model that
governs film is slowly being pushed to every kind of content. The changes
we see affect every front of human creativity. They affect commercial as well
as noncommercial activities, the arts as well as the sciences. They are as
much about growth and jobs as they are about music and film. And how
we decide these questions will determine much about the kind of society
we will become. It will determine what the “free” means in our self-
congratulatory claim that we are now, and will always be, a “free society.”
This is a struggle about an ideal—about what rules should govern the
freedom to innovate. I would call it a “moral question,” but that sounds too
personal, or private. One might call it a political question, but most of us
work hard to ignore the absurdities of ordinary politics. It is instead best de-
scribed as a constitutional question: it is about the fundamental values that
define this society and whether we will allow those values to change. Are
we, in the digital age, to be a free society? And what precisely would that
idea mean?

to answer these questions, we must put them into context. That’s what I
will do in the balance of this chapter. Step back from the conflict about
music or innovation, and think about resources in a society more generally.
How are resources, in this vague, general sense, ordered? Who decides who
gets access to what?
Every society has resources that are free and resources that are controlled.
Free resources are those available for the taking. Controlled resources are
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 12

1 2 L a w r e n c e L e s s i g

those for which the permission of someone is needed before the resource
can be used. Einstein’s theory of relativity is a free resource. You can take it
and use it without the permission of anyone. Einstein’s last residence in
Princeton, New Jersey, is a controlled resource. To sleep at 112 Mercer
Street requires the permission of the Institute for Advanced Study.
Over the past hundred years, much of the heat in political argument
has been about which system for controlling resources—the state or the
market—works best. The Cold War was a battle of just this sort. The social-
ist East placed its faith in the government to allocate and regulate resources;
the free-market West placed its faith in the market for allocating or regu-
lating resources. The struggle was between the state and the market. The
question was which system works best.
That war is over. For most resources, most of the time, the market trumps
the state. There are exceptions, of course, and dissenters still. But if the
twentieth century taught us one lesson, it is the dominance of private over
state ordering. Markets work better than Tammany Hall in deciding who
should get what, when. Or as Nobel Prize–winning economist Ronald
Coase put it, whatever problems there are with the market, the problems
with government are far more profound.
This, however, is a new century; our questions will be different. The issue
for us will not be which system of exclusive control—the government or the
market—should govern a given resource. The question for us comes before:
not whether the market or the state but, for any given resource, whether that
resource should be controlled or free.
“Free.”
So deep is the rhetoric of control within our culture that whenever one
says a resource is “free,” most believe that a price is being quoted—free, that
is, as in zero cost. But “free” has a much more fundamental meaning—in
French, libre rather than gratis, or for us non–French speakers, and as the
philosopher of our age and founder of the Free Software Foundation
Richard Stallman puts it, “free, not in the sense of free beer, but free in the
sense of free speech.”12 A resource is “free” if (1) one can use it without the
permission of anyone else; or (2) the permission one needs is granted neu-
trally. So understood, the question for our generation will be not whether
the market or the state should control a resource, but whether that resource
should remain free.13
This is not a new question, though we’ve been well trained to ignore
it. Free resources have always been central to innovation, creativity, and
democracy. The roads are free in the sense I mean; they give value to
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 13

T H E F U T U R E O F I D E A S 1 3

the businesses around them. Central Park is free in the sense I mean; it
gives value to the city that it centers. A jazz musician draws freely upon the
chord sequence of a popular song to create a new improvisation, which, if
popular, will itself be used by others. Scientists plotting an orbit of a space-
craft draw freely upon the equations developed by Kepler and Newton and
modified by Einstein. Inventor Mitch Kapor drew freely upon the idea of a
spreadsheet—VisiCalc—to build the first killer application for the IBM
PC—Lotus 1-2-3. In all of these cases, the availability of a resource that re-
mains outside the exclusive control of someone else—whether a govern-
ment or a private individual—has been central to progress in science and
the arts. It will also remain central to progress in the future.
Yet lurking in the background of our collective thought is a hunch that
free resources are somehow inferior. That nothing is valuable that isn’t re-
stricted. That we shouldn’t want, as Groucho Marx might put it, any re-
source that would willingly have us. As Yale professor Carol Rose writes, our
view is that “the whole world is best managed when divided among private
owners,”14 so we proceed as quickly as we can to divide all resources among
private owners so as to better manage the world.
This is the taken-for-granted idea that I spoke of at the start: that control
is good, and hence more control is better; that progress always comes from
dividing resources among private owners; that the more dividing we do, the
better off we will be; that the free is an exception, or an imperfection, which
depends upon altruism, or carelessness, or a commitment to communism.
Free resources, however, have nothing to do with communism. (The So-
viet Union was not a place with either free speech or free beer.) Neither are
the resources that I am talking about the product of altruism. I am not ar-
guing that there is such a thing as a “free lunch.” There is no manna from
heaven. Resources cost money to produce. They must be paid for if they are
to be produced.
But how a resource is produced says nothing about how access to that re-
source is granted. Production is different from consumption. And while the
ordinary and sensible rule for most goods is the “pay me this for that” model
of the local convenience store, a second’s reflection reveals that there is a
wide range of resources that we make available in a completely different
way.
Think of music on the radio, which you consume without paying any-
thing. Or the roads that you drive upon, which are paid for independently
of their use. Or the history that we hear about without ever paying the re-
searcher. These too are resources. They too cost money to produce. But we
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 14

1 4 L a w r e n c e L e s s i g

organize access to these resources differently from the way we organize ac-
cess to chewing gum. To get access to these, you don’t have to pay up front.
Sometimes you don’t have to pay at all. And when you do have to pay, the
price is set neutrally or without regard to the user, inside or outside the com-
pany. And for good reason, too. Access to chewing gum may rightly be con-
trolled all the way down; but access to roads, and history, and control of our
government must always, and sensibly, remain “free.”

the argument of this book is that always and everywhere, free resources
have been crucial to innovation and creativity; that without them, cre-
ativity is crippled. Thus, and especially in the digital age, the central ques-
tion becomes not whether government or the market should control a
resource, but whether a resource should be controlled at all. Just because
control is possible, it doesn’t follow that it is justified. Instead, in a free soci-
ety, the burden of justification should fall on him who would defend sys-
tems of control.
No simple answer will satisfy this demand. The choice is not between all
or none. Obviously many resources must be controlled if they are to be pro-
duced or sustained. I should have the right to control access to my house
and my car. You shouldn’t be allowed to rifle through my desk. Microsoft
should have the right to control access to its source code. Hollywood should
have the right to charge admission to its movies. If one couldn’t control ac-
cess to these resources, or resources called “mine,” one would have little in-
centive to work to produce these resources, including those called mine.
But likewise, and obviously, many resources should be free. The right to
criticize a government official is a resource that is not, and should not be,
controlled. I shouldn’t need the permission of the Einstein estate before I
test his theory against newly discovered data. These resources and others
gain value by being kept free rather than controlled. A mature society real-
izes that value by protecting such resources from both private and public
control.
We need to learn this lesson again. The opportunity for this learning is
the Internet. No modern phenomenon better demonstrates the importance
of free resources to innovation and creativity than the Internet. To those
who argue that control is necessary if innovation is to occur, and that more
control will yield more innovation, the Internet is the simplest and most di-
rect reply. For as I will show in the chapters that follow, the defining feature
of the Internet is that it leaves resources free. The Internet has provided for
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 15

T H E F U T U R E O F I D E A S 1 5

much of the world the greatest demonstration of the power of freedom—
and its lesson is one we must learn if its benefits are to be preserved.
Yet at just the time that the Internet is reminding us about the extraordi-
nary value of freedom, the Internet is being changed to take that freedom
away. Just as we are beginning to see the power that free resources produce,
changes in the architecture of the Internet—both legal and technical—are
sapping the Internet of this power. Fueled by a bias in favor of control,
pushed by those whose financial interests favor control, our social and po-
litical institutions are ratifying changes in the Internet that will reestablish
control and, in turn, reduce innovation on the Internet and in society gen-
erally.
I am dead against the changes we are seeing, but it is too much to believe
I could convince you that the full range is wrong. My aim is much more
limited. My hope is to show you the other side of what has become a taken-
for-granted idea—the view that control of some sort is always better. If you
stay with me to the end, then I want you to leave this book simply with a
question about whether control is best. I don’t have the data to prove any-
thing more than this limited hope. But we do have a history to show that
there is something important here to understand.

this showing moves in three steps. In the part that follows, I introduce
more formally what I mean by “free.” I relate that concept to the notion of
“the commons” and then introduce three contexts where resources in the
Internet are held in common. These commons are related to the innovation
the Internet has produced. My aim in this first part is to show just how.
I then consider in part II a parallel environment for innovation and cre-
ativity in “real space”—the space not tied directly to the Internet, though in-
creasingly affected by it. This is the space where records are now made,
books are still written, and film is primarily shot. This space does not present
the commons the Internet is—and for good reason, too. The character of
production in real space does not permit the freedom that the Internet does.
The constraint on creativity it yields there is a necessary, if unfortunate, fea-
ture of that space.
This context of creativity has been changed by the Internet. In the bal-
ance of part II, I offer examples of how. These examples will show how many
of the constraints that affected real-space creativity have been removed by
the architecture, and original legal context, of the Internet. These limita-
tions, perhaps justified before, are justified no more.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 16

1 6 L a w r e n c e L e s s i g

Or at least, were justified no more. For the argument of the third and
final part of this book is that the environment of the Internet is now chang-
ing. Features of the architecture—both legal and technical—that originally
created this environment of free creativity are now being changed. They are
being changed in ways that will reintroduce the very barriers that the Inter-
net originally removed.
These barriers, however, don’t have the neutral justification that the
constraints of real-space economics do.15 If there are constraints here, it
is simply because we are building them in. And as I will argue, there are
strong reasons why many are trying to rebuild these constraints: they will en-
able these existing and powerful interests to protect themselves from the
competitive threat the Internet represents. The old, in other words, is bend-
ing the Net to protect itself against the new.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 17

P A R T

///

DOT.COMMONS
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 18
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 19

B u i l d i n g B l o c k s :

“ C o m m o n s ” a n d “ L a y e r s ”

T his book is fundamentally about the Internet and its effect on innova-
tion, both commercial and non-. “Internet” and “society” are familiar
enough notions. But at the core of my argument are two fairly obscure ideas
that we must begin by making a bit more clear. The first of these is the idea
of a “commons”; the second is the notion of “layers.” The commons is an
old idea; layers, in the sense made familiar by network theorists, are rela-
tively new. But the two together organize the argument that follows. They
are building blocks to an end that will help reveal the Internet’s effect on
society.

THE COMMONS

if you’ve used the word commons before, you’re likely to think of a park,
as in the Boston Common. If you’ve studied economics or political science,
your mind will race to tragedy (as in “the tragedy of the commons”). Both
senses are related to what I mean, but neither alone is enough.1
The Oxford English Dictionary (mankind’s first large-scale collaborative
open source text project)2 equates the “commons” to a resource held “in
common.” That it defines as “in joint use or possession; to be held or en-
joyed equally by a number of persons.”3 In this sense, a resource held “in
common” is “free” (as I’ve defined that term) to those “persons.” In most
cases, the commons is a resource to which anyone within the relevant com-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 20

2 0 L a w r e n c e L e s s i g

munity has a right without obtaining the permission of anyone else. In some
cases, permission is needed but is granted in a neutral way.
Think about some examples:

• The public streets are commons. Anyone is free to access the streets
without first getting the permission of someone else. We don’t auction
rights of access, selling the right to use a particular bit of highway dur-
ing a particular bit of time. (Of course there are exceptions.) Nor do we
insist on particular licenses before we allow people to use the streets or
highways. Instead the highways are open and free—in the sense I mean
a commons to be free.
• Parks and beaches are increasingly commons. Anyone is free to access
these spaces without getting the permission of someone else. Access
is not auctioned off to the highest bidder, and the right to control ac-
cess is not handed off to some private or governmental entity. The
resource—as Carol Rose calls it, “the recreational resource”—is made
available to anyone.
• Einstein’s theory of relativity is a commons. It is a resource—a way of
understanding the nature of the universe—that is open and free for any-
one to take. Access to this resource is not auctioned off to the highest
bidder; the right to use the theory is not allocated to a single organiza-
tion.
• Writings in the public domain are a commons. They are a resource that
is open and free for anyone to take without the permission of anyone
else. An 1890 edition of Shakespeare is free for anyone to take and copy.
Your right to use and redistribute that 1890 text is without restraint.

Each of these resources is held in common. Each is “free” for others to
take. Some are free in the sense that no price is paid (you can use most roads
without paying a toll; as we will see, it would be unconstitutional in the
United States to require anyone to pay to use Einstein’s theory of relativity).
Some are free even though a price must be paid (a park is “free” in the sense
that I mean even if an access fee is required—as long as the fee is neutrally
and consistently applied).4 In both cases, the essential feature is reasonable,
and that access to the resource is not conditioned upon the permission of
someone else. The essence, in other words, is that no one exercises the core
of a property right with respect to these resources—the exclusive right to
choose whether the resource is made available to others.5
Economists will object, however, that my list conflates two very different
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 21

T H E F U T U R E O F I D E A S 2 1

cases. Einstein’s theory of relativity is different from the streets or public
beaches. Einstein’s theory is fully “nonrivalrous”; the streets and beaches
are not. If you use the theory of relativity, there is as much left over afterward
as there was before. Your consumption, in other words, does not rival my
own. But roads and beaches are very different. If everyone tries to use the
roads at the very same time (something that apparently happens out here in
California often), then their use certainly rivals my own. Traffic jams; pub-
lic beaches crowd. Your SUV, or your loud radio, reduces my ability to
enjoy the roads or beach.
The economists are right. This list of resources held in “the commons”
does conflate rivalrous with nonrivalrous resources. But our tradition is not
as tidy as the economists’ analytics. We have always described as “com-
mons” both rivalrous and nonrivalrous resources. The Boston Common is
a commons, though its resource is rivalrous (my use of it competes with
your use of it). Language is a commons, though its resource is nonrivalrous
(my use of it does not inhibit yours).6 What has determined “the commons,”
then, is not the simple test of rivalrousness. What has determined the com-
mons is the character of the resource and how it relates to a community. In
theory, any resource might be held in common (whether it would survive is
another question). But in practice, the question a society must ask is which
resources should be, and for those resources, how.
Here the distinction that the economists draw begins to help. Economists
distinguish rivalrous and nonrivalrous resources because the issues or prob-
lems raised by each kind are different.
If a resource is nonrivalrous, then the problem is whether there is enough
incentive to produce it, not whether there is too much demand to consume
it. A nonrivalrous resource can’t be exhausted. Once it is produced, it can’t
be undone. Thus the issue for nonrivalrous resources is whether the Edith
Whartons of the world have enough incentive to create. The problem with
nonrivalrous resources is to assure that I reap enough benefit to induce me
to sow.
A rivalrous resource presents more problems. If a resource is rivalrous,
then we must worry both about whether there is sufficient incentive to cre-
ate it (if it is the sort of resource that humans produce) and about whether
consumption by some will leave enough to others. With a rivalrous re-
source, I must still worry that I will reap enough benefit to make it worth it
to sow. But I must worry as well that others not deplete the resource that I’ve
produced. If a rivalrous resource is open to all, there is a risk that it will be
depleted by the consumption of all.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 22

2 2 L a w r e n c e L e s s i g

This depletion of a rivalrous resource is the dynamic that biologist Gar-
rett Hardin famously termed “the tragedy of the commons.”7 “Picture a pas-
ture open to all,” Hardin writes, and consider the expected behavior of
“herdsmen” who roam that pasture. Each herdsman must decide whether
to add one more animal to his herd. In making a decision to do so, Hardin
writes, the herdsman reaps a benefit, while everyone else suffers. The herds-
man gets the benefit of one more animal, yet everyone suffers the cost,
because the pasture has one more consuming cow. And this defines the
problem: Whatever costs there are in adding another animal are costs that
others bear. The benefits, however, are enjoyed by a single herdsman.
Therefore each herdsman has an incentive to add more cattle than the pas-
ture as a whole can bear. As Hardin describes the consequence:

Therein is the tragedy. Each man is locked into a system that compels him
to increase his herd without limit—in a world that is limited. Ruin is the
destination toward which all men rush, each pursuing his own best interest
in a society that believes in the freedom of the commons. Freedom in a com-
mons brings ruin to all.8

This “tragedy” consumes talk about “the commons.” “Ruin” is taken for
granted as the destiny of those who believe in the “freedom of the com-
mons.” Hardheaded sorts thus scorn the rhetoric of undivided resources.
Only the romantic wastes time wondering about anything different from the
perfect control of property.
But obviously Hardin was not describing a law of nature that must apply
to every good left in the commons. There is, for example, no tragedy for
nonrivalrous goods left in the commons—no matter how many times you
read a poem, there’s as much left over as there was when you started. Nor is
there always a tragedy even for rivalrous goods. As researchers have shown,
in many different contexts, norms adequately limit the problem of over-
consumption.9 Communities work out how to regulate overconsumption.
How and why are certainly complex questions. But that some do is undeni-
able.10
We therefore can’t just jump from the observation that a resource is held
“in common” to the conclusion that “freedom in a commons brings ruin to
all.” Instead, we must think empirically and look at what works. Where
there is a benefit from leaving a resource free, we should see whether there
is a way to avoid overconsumption, or inadequate incentives, without its
falling under either state or private (market) control.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 23

T H E F U T U R E O F I D E A S 2 3

My central claim throughout is that there is a benefit to resources held in
common and that the Internet is the best evidence of that benefit. As we will
see, the Internet forms an innovation commons. It forms this commons not
just through norms, but also through a specific technical architecture. The
Net of these norms and this architecture is a space where creativity can
flourish. Yet so blind are we to the possible value of a commons that
we don’t even notice the commons that the Internet is. And, in turn, this
blindness leads us to ignore changes to the norms and architecture of
the Net that weaken this commons. There is a tragedy of the commons
that we will identify here; it is the tragedy of losing the innovation com-
mons that the Internet is, through the changes that are being rendered on
top.11

L AY E R S

the idea of the commons may be obscure, but the notion of “layers” is
more easily recognized. The layers that I mean here are the different layers
within a communications system that together make communications pos-
sible. The idea is taken from perhaps the best communications theorist of
our generation, NYU law professor Yochai Benkler.12 As he uses the idea, it
helps organize our thought about how any communications system func-
tions. But in organizing our thought, his work helps show something we
might otherwise miss.
Following the technique of network architects, Benkler suggests that we
understand a communications system by dividing it into three distinct “lay-
ers.”13 At the bottom is a “physical” layer, across which communication trav-
els. This is the computer, or wires, that link computers on the Internet. In
the middle is a “logical” or “code” layer—the code that makes the hardware
run. Here we might include the protocols that define the Internet and the
software upon which those protocols run. At the top is a “content” layer—
the actual stuff that gets said or transmitted across these wires. Here we in-
clude digital images, texts, on-line movies, and the like. These three layers
function together to define any particular communications system.
Each of these layers in principle could be controlled or could be free.
Each, that is, could be owned or each could be organized in a commons.
We could imagine a world where the physical layer was free but the logical
and content layers were not. Or we could imagine a world where the physi-
cal and code layers were controlled but the content layer was not. And so on.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 24

2 4 L a w r e n c e L e s s i g

Consider some examples to make the possibilities real.
Speakers’ Corner: Speakers’ Corner is a place in London’s Hyde Park
where people who want to speak publicly gather on Sundays to deliver their
speeches. It is a wonderfully English spectacle, ordinarily filled with both
orators and loons. But the system of communication is distinctive: the physi-
cal layer (the park) is a commons; the code layer (the language used) is a
commons, too; and the content layer is ordinarily unowned—what these
nuts say is their own creation. All three layers in this context are free; no one
can exercise control over the kinds of communications that might happen
here.
Madison Square Garden: Madison Square Garden is another place where
people give speeches or, more likely, play games. It is a huge stadium/
auditorium near the center of Manhattan, owned by Madison Square Gar-
den, L.P. Only those who pay get to use the auditorium; and the Garden is
not obligated to take all comers. The physical layer is therefore controlled.
But as with Speakers’ Corner, both the code layer (the language) and the
content layer (what gets uttered) are at least sometimes not controlled. They
too can remain free.
The telephone system: The telephone system before its breakup was a
single unitary system. The physical infrastructure of this system was owned
by AT&T and its affiliates; so too was its logical infrastructure—determining
how and who you could connect—controlled by AT&T. But what you said
on an AT&T phone (within limits, at least)14 was free: the content of the
telephone conversations was not controlled, even if the physical and code
layers underneath were.
Cable TV: Finally, think of cable TV. Here the physical layer is owned—
the wires that run the content into your house. The code layer is
owned—only the cable companies get to decide what runs into your house.
And the content layer is owned—the shows that get broadcast are copy-
righted shows. All three layers are within the control of the cable TV com-
pany; no communications layer, in Benkler’s sense, remains free.
These examples suggest the range of ways of organizing systems of
communications. No single mix is best, though the differences among the
four are important. To the extent that we want a decentralized system of
communications, unowned layers will help. To the extent that we want con-
trolled systems of communications, owned layers will help. But the point of
the scheme so far is not to make predictions. The point is simply to make
clear the range, and that trade-offs within this range exist.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 25

T H E F U T U R E O F I D E A S 2 5

Speakers’ Madison Telephone Cable TV
Corner Square System
Garden
Content Free Free Free Controlled
Code Free Free Controlled Controlled
Physical Free Controlled Controlled Controlled

Now, from the language I’ve used so far, you might think that the Inter-
net is a communications system free all the way down—free, that is, at every
one of Benkler’s layers. It is not. What is special about the Internet is the way
it mixes freedom with control at different layers. The physical layer of the
Internet is fundamentally controlled. The wires and the computers across
which the network runs are the property of either government or individu-
als. Similarly, at the content layer, much in the existing Internet is con-
trolled. Not everything served across the Net is free for the taking. Much is
properly and importantly protected by property law.
At the code layer, however, in ways that will become clearer below, the
Internet was free. So too was much of the content served across the network
free. The Internet thus mixed both free and controlled layers, not just layers
that were free.
Our aim is to understand how this mix produced the innovation that we
have seen so far and why the changes to this mix will kill what we have seen
so far.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 26

C o m m o n s o n t h e W i r e s

T he internet is a network of networks. In the main, these networks con-
nect over wires. All of these wires, and the machines linked by them, are
controlled by someone. The vast majority are owned by private parties—
owned, that is, by individuals and corporations that have chosen to link to
the Net. Some are owned by the government.
Yet this vast network of privately owned technology has built one of the
most important innovation commons that we have ever known. Built on a
platform that is controlled, the protocols of the Internet have erected a free
space of innovation. These private networks have created an open resource
that any can draw upon and that many have. Understanding how, and in
what sense, is the aim of this chapter.

paul baran was a researcher at the Rand Corporation from 1959 to 1968.
His project in the early 1960s was communications reliability. The fear
slowly dawning upon the leaders of the world’s largest nuclear arsenal was
that the communications system controlling that arsenal was vulnerable
to the smallest of attacks. An accident, or a single nuclear explosion, could
disable the ability of the commander in chief to command. Chaos—or
worse—would be unavoidable.
Baran’s task was to explore a more secure telecommunications system.
His first step was to understand the system then in place. So he asked the
then provider of telecommunications in America, American Telephone &
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 27

T H E F U T U R E O F I D E A S 2 7

Telegraph, to see the plans for the AT&T network to determine whether the
communications system was secure.
AT&T balked. Though Baran had the proper security clearance, and
though the Defense Department supported his request, AT&T refused
Baran’s inquiry. They had studied the matter, AT&T reported. The system
was secure.
This was “the Bell system.” It is hard for us today to appreciate the power
of such a company. This was not just a large company, or even a large com-
pany with a very large market share. This was a partner with the govern-
ment, ruling telecommunications in America. It was therefore, in its own
view of itself, the governor of communications. States and the Federal Com-
munications Commission (FCC) might regulate it, but the information
and cooperation to make that regulation possible came from AT&T. It had
been, since interconnection began in earnest in 1912, America’s tele-
communications master.1
Things with telephones were not always this way. Indeed, the early his-
tory of telecommunications is essentially unrecognizable to us. Though the
Bell companies held the first patents on telephone technology, once those
patents expired, a vigorous competition emerged to bring telephone service
to Americans. AT&T concentrated on businesses. “Independents” focused
on residences. The competition produced a rapid expansion of coverage.
“From 1900 to 1915, at least 45% of the U.S. cities with populations over
5,000 had competing, non-interconnected telephone exchanges. During
the peak of the independent movement’s strength, between 1902 and 1910,
that percentage was more than 55%.”2
Today we would not recognize the phone system that this early competi-
tion produced. Though the reach of the telephone network was great—in
1920, 38.7 percent of farms and 30 percent of residences had a telephone3—
the networks did not interconnect. There was no guarantee that if your
grandmother across town had a telephone, you, using yours, could call her.
Thus when you purchased telephone service, your decision in part de-
pended upon whom you wanted to call and what service they were likely to
have.
The world was then with telephones as the world was with personal com-
puters ten years ago,4 or as the world with instant messaging is today.
Though there was a dominant system (AT&T for phones; Microsoft/Intel
for computers; AOL’s AIM for instant messaging), there was vigorous com-
petition among other systems (the “independents” for phones; Apple’s Mac-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 28

2 8 L a w r e n c e L e s s i g

intosh or IBM’s OS/2 for computers; Yahoo! or MSN for messaging). This
competition effectively pushed the dominant system to become better and
different. Just as the windows of Macintosh pushed Microsoft to Windows,
so too the rural service of the “independents” pushed AT&T to extend its
reach to farmers.
After a while, however, AT&T grew weary of this competition. The view
grew within the company that security would come only by merging with
the competitors. From 1908 to 1913, the Bell system adopted a number of
strategies to destroy the independents, including selective interconnection
and acquisition of competitors. If it could not gain customers through direct
competition, it would gain customers by purchasing competitors.5
Initially, this consolidation inspired skepticism among regulators and the
public. AT&T was attacked as a monster seeking monopoly. But by the early
1920s, antitrust enforcement in the United States was waning. The spirit of
the time favored consolidation and rationalization; competition was viewed
as “ruinous.” Thus AT&T was slowly able to secure agreements with the
government that essentially permitted it to extend its reach while protecting
it against antitrust review.
Paradoxically, AT&T’s most effective weapon in this expansion was to
offer competitors the ability to interconnect. Though our intuition is
likely to tell us that it was the failure to interconnect that hampered com-
petition, in fact, as economist Milton Mueller has effectively argued, it was
a lack of interconnection that spurred competition.6 As each independent
interconnected to the AT&T system, any distinctive advantage it could offer
as an independent disappeared. Consumers had no further interest in sub-
scribing to it over AT&T; hence the drive to AT&T as universal provider
was only increased. The network advantage of AT&T would grow relative to
other independents; hence the power of AT&T’s increasing monopoly was
enhanced.
Independents at the time understood this dynamic. Associations of inde-
pendents vigorously attacked the “traitors” among them that chose to inter-
connect with AT&T.7 But these competitors were increasingly seen as
inconveniences, by both regulators and the public. The idea of a world of
“universal service”—meaning not a telephone in every house, but a system
where every phone could reach every other phone—was too seductive.8 So
in 1913 the government entered into an agreement with AT&T that would
secure its monopoly in telecommunications in America, even though it was
sold as a solution to telecommunications monopoly.
Named after Bell vice president Nicholas C. Kingsbury, the “Kingsbury
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 29

T H E F U T U R E O F I D E A S 2 9

Commitment” required that Bell “stop acquiring independent phone com-
panies and [. . .] connect the remaining independents to Bell’s long-
distance network.”9 Bell also had to divest its telegraph arm, Western Union.
This stopped the company from its increasingly ravenous practice of acqui-
sition, but it “did nothing to promote competition in either telephony or
telegraphy.”10 The commitment did not force local exchanges to be more
competitive. It did not require interconnection with other long-distance car-
riers. The solution, “in short, was not the steamy unsettling cohabitation
that marks competition, but rather a sort of competitive apartheid, char-
acterized by segregation and quarantine.”11 As a major treatise on tele-
communications describes it:

The Kingsbury Commitment could be viewed as a solution only by a gov-
ernment bookkeeper who counted several separate monopolies as an ad-
vance over a single monopoly, even absent any trace of competition
among them.12

Monopolies are not all bad, and no doubt this monopoly did lots of
good. AT&T produced an extraordinary telephone system, linking 85 per-
cent of American homes at the peak of its monopoly power in 1965.13 It
spent billions of dollars to support telecommunications research. Bell Labs
invented fiber optic technology, the transistor, and scads of other major
technological advances. Its scientists earned at least half a dozen Nobel
Prizes in physics.
And it attracted a certain kind of person. As Paul Baran described it:

They were not motivated by making a lot of money. They were in the
business to provide a service: Loyalty to the organization and help to the
country providing the world’s best communication. And that was their mo-
tivation and their belief. It was a religion, a pure religion. . . . In their
mind, they were doing the right thing.14

These were not fat monopolists seeking to rob the nation of a quick buck.
These were “soldiers of communications,” for whom control and hierarchy
were key. As one publication in 1941 put it:

Because each of them has a part in this speeding of the spoken word, the
thousands of men and women who are engaged in the telephone service
in America are ever conscious of the fact that theirs is a high calling.15
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 30

3 0 L a w r e n c e L e s s i g

AT&T in turn succeeded during its monopoly reign in attracting the very
best telecommunications researchers. Baran attributes its success to “an ab-
solutely brilliant compensation system,”16 but the reason may well be that
AT&T was the only show in town. As Baran describes, “[F]or years and
years, that was the only place in the country that was doing work in tele-
communications.”17 One could research different telecommunications sys-
tems, and one could in principle even develop other telecommunications
systems. But there was nothing one could do with one’s innovation unless
AT&T bought it.
For much of the twentieth century, it was essentially illegal even to
experiment with the telephone system. It was a crime to attach a device to
the telephone system that AT&T didn’t build or expressly authorize. In
1956, for example, a company built a device called a “Hush-a-Phone.” The
Hush-a-Phone was a simple piece of plastic that attached to the mouthpiece
of a telephone. Its design was to block noise in a room so that someone on
the other end of the line could better hear what was being said. The device
had no connection to the technology of the phone, save the technology of
the plastic receiver. All it did was block noise, the way a user might block
noise by cupping his hand over the phone.18
When the Hush-a-Phone was released on the market, AT&T objected.
This was a “foreign attachment.” Regulations forbade any foreign attach-
ments without AT&T’s permission. AT&T had not given Hush-a-Phone any
such permission. The FCC agreed with AT&T. Hush-a-Phone was history.
Hush-a-Phone is an extreme case.19 The real purpose of the foreign at-
tachments rule was, at least as AT&T saw it, to protect the system from dirty
technology. A bad telephone or a misbehaving computer attached to the
telephone system could, AT&T warned, bring down the system for the
whole region. Telephones were lifelines, and they had to be protected from
the experiments of an inquisitive nation. Rules such as the foreign attach-
ments rules were intended to achieve this protection.
Whatever their intent, however, these rules had an effect on innovation
in telecommunications. Their effect was to channel innovation through
Bell Labs. Progress would be as Bell Labs determined it. Experiments
would be pursued as Bell Labs thought best. Thus telecommunications
would evolve as Bell Labs thought best.

baran u nde r st o o d this. As a researcher at a Defense Department–
supported lab, he knew how the “military” thought, and AT&T was mili-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 31

T H E F U T U R E O F I D E A S 3 1

tary. Thus he had reason to be skeptical about the claims that the existing
system would withstand a nuclear attack. He didn’t believe AT&T really
understood the threat. And if it did, he believed it simply didn’t want anyone
else understanding its weakness.
So he pushed AT&T to let him examine the system. It pushed back. And
so, from sources unnamed, Baran secured a copy of AT&T’s plans—the
blueprints for the telecommunications system of the United States.
When he saw the plans, Baran knew AT&T was wrong. He was certain
that the system it had built would not withstand a nuclear attack. The net-
work was too concentrated; it had no effective redundancy. So he continued
to press his idea for a different telecommunications system. He had a differ-
ent design for telecommunications, and he wanted AT&T to help him
build it.
This different model was not the Internet, but it was close to the Internet.
Baran proposed a kind of packet-switching technology to replace the persis-
tent circuits around which the telephone system was built. Under AT&T’s
design, when you called someone in Paris, a circuit was opened between
you and Paris. In principle, you could trace the line of copper that linked
you to Paris; along that line of copper, all your conversation would travel.
Baran’s idea was fundamentally different. If you digitized a conversation—
translating it from waves to bits—and then chopped the resulting stream
into packets, these packets could flow independently across a network and
create the impression of a real-time connection on the other end. As long as
they flowed fast enough, and the computers at both ends were quick, the
conversation encoded in this packet form would seem just like a conversa-
tion along a single virtual wire across the ocean.
Baran was probably not the first person to come up with this idea—MIT
loyalists insist that that was Leonard Kleinrock.20 And he was also not the
only person working on the idea in the early 1960s. Independently, in En-
gland, Donald Davies was developing something very similar.21 But
whether the first, or the only, doesn’t really matter for our purposes here.
What is important is that Baran outlined a telecommunications system fun-
damentally different from the dominant design, and that different telecom-
munications system would have effected a radically different evolution of
telecommunications.

baran pushed to get AT&T to help build this alternative design. AT&T
said he didn’t understand telephones. Over the course of many months, he
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 32

3 2 L a w r e n c e L e s s i g

attended classes sponsored by AT&T so that he would get with its pro-
gram. But the more Baran saw, the more convinced he was. And in a final
push, the Defense Department offered simply to pay AT&T to build the sys-
tem. The government promised no risk; it wanted only cooperation. But
even here, AT&T balked. As recounted in John Naughton’s A Brief History
of the Future:22

[AT&T’s] views were once memorably summarised in an exasperated out-
burst from AT&T’s Jack Osterman after a long discussion with Baran.
‘First,’ he said, ‘it can’t possibly work, and if it did, damned if we are going
to allow the creation of a competitor to ourselves.’23

“Allow.” Here is the essence of the AT&T design, supported by the state-
sanctioned monopoly. In “defend[ing] the monopoly,”24 it reserved to itself
the right to decide what telecommunications would be “allowed.” As Baran
put it, AT&T “didn’t want anybody in their vicarage.”25 It controlled the
wires; nothing but its technology could be attached, and no other system of
telecommunications would be permitted. One company, through one re-
search lab, with its vision of how communications should occur, decided.
Innovation here, for this crucial aspect of modern economic life, was as this
single organization would decide.
Now again, the point is not that AT&T was evil. Indeed, quite the con-
trary. We get nowhere in understanding how systems of innovation work
when we personify organizations and imagine them responsible for social
goals. AT&T had an obligation to its stockholders; it had an obligation to
the government to assure consistent quality service. It was simply acting to
assure that it met both of these obligations—maximizing its profits for its
shareholders while meeting its obligations to the government.
But what’s good for AT&T is not necessarily good for America. What
AT&T was doing may well have made sense for it; its vision of tele-
communications may well have made sense for the interests it understood
itself to be serving. But AT&T’s vision of what a telecommunications service
should be is not necessarily what a telecommunications service should be.
There is a possible—and in this case actual—conflict between the interests
of a centralized controller of innovation and the interest in innovation
generally.
Here the conflict was plain. If the Defense Department built a tele-
communications system based on packets rather than circuits, then the effi-
ciency of that system could in theory be much greater. When you’re on a
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 33

T H E F U T U R E O F I D E A S 3 3

circuit-switched system, listening to your lover in Paris tell you about someone
new, there’s lots of downtime on the line—silence—that is just wasted band-
width. If instead the system were packets, then the data from the downtime
would be silence; it’s easier to send the information necessary to reproduce
silence than it is to hold open a line while silence happens. The system could
better utilize the wires if the architecture enabled the sharing of the wires.
The owner of a legacy system built on a different model could well de-
cide that this challenge was too dangerous. If a more efficient system came
on-line, there would be strong pressure from the government to allow the
exception; that exception would not be easy to limit; the corrosion of the ex-
isting model could be great. Monopoly control would be lost.
Thus it is completely understandable that a company like AT&T would
not want to give birth to this new competitor, even if this new competitor
would be better for communications as a whole. The natural desire of any
company is to find ways to protect its market. And the chosen desire of a
competitive market is to limit the ways in which a company can protect its
market—but for most of the century, this chosen desire was not tele-
communications policy. For most of the century, in this context and others
that we will consider later on, the chosen desire of policy makers was to
back up the desire of companies to architect and support systems that pro-
tected them against competition in the market. Competition was a bother;
the vision of a telecommunications system was limited; and our tele-
communications architecture—including, as we will see, broadcasting and
radio—was architected to maximize the power and control of the few.26

at a certain point, Baran understood. When the project was pushed into
the Defense Communications Agency (DCA), Baran realized the project
would be bungled. As he told author John Naughton:

I felt that [DCA] could be almost guaranteed to botch the job since they
had no understanding of digital technology, nor for leading edge high
technology development. Further, they lacked enthusiasm. Sometimes, if
a manager doesn’t have the staff but has the drive and smarts to assemble
the right team, one could justify taking a chance. But lacking skills, com-
petence and motivation meant backing a sure loser.27

So Baran had the project pulled. There were not “the people at the time
who could successfully undertake this project, [and they] would likely
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 34

3 4 L a w r e n c e L e s s i g

screw up the program. An expensive failure would make it difficult for a
more competent agency to later undertake the project.”28 Thus, this archi-
tecture of control—centralizing innovation and protecting an existing
model of doing business—would not be questioned by Baran’s work. At
least not then.

the internet is not the telephone network. It is a network of networks
that sometimes run on the telephone lines. These networks and the wires
that link them are privately owned, like the wires of the old AT&T. Yet at the
core of this network is a different principle from the principle that guided
AT&T. Like the principle Baran confronted, this principle affects what is al-
lowed and what is not. And like the principle that Baran confronted, this
principle has an effect on innovation.
First described by network architects Jerome Saltzer, David Clark, and
David P. Reed in 1981, this principle—called the “end-to-end argument”
(e2e)—guides network designers in developing protocols and applications
for the network.29 End-to-end says to keep intelligence in a network at the
ends, or in the applications, leaving the network itself to be relatively
simple.
There are many principles in the Internet’s design. This one is key. But it
will take some explaining to show why.
Network designers commonly distinguish computers at the “end” or
“edge” of a network from computers within that network. The computers at
the end of a network are the machines you use to access the network. (The
machine you use to dial into the Internet, or your cell phone connecting to
a wireless Web, is a computer at the edge of the network.) The computers
“within” the network are the machines that establish the links to other
computers—and thereby form the network itself. (The machines run by
your Internet service provider, for example, could be computers within the
network.)
The end-to-end argument says that rather than locating intelligence
within the network, intelligence should be placed at the ends: compu-
ters within the network should perform only very simple functions that are
needed by lots of different applications, while functions that are needed by
only some applications should be performed at the edge. Thus, complexity
and intelligence in the network are pushed away from the network itself.
Simple networks, smart applications. As a recent National Research Coun-
cil (NRC) report describes it:
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 35

T H E F U T U R E O F I D E A S 3 5

Aimed at simplicity and flexibility, [the end-to-end] argument says that the
network should provide a very basic level of service—data transport—and
that the intelligence—the information processing needed to provide
applications—should be located in or close to the devices attached to the
edge [or ends] of the network.30

The reason for this design was flexibility, inspired by a certain humility.
As Reed describes it, “we wanted to make sure that we didn’t somehow build
in a feature of the underlying network technology . . . that would restrict our
using some new underlying transport technology that turned out to be good
in the future. . . . That was really the key to why we picked this very, very
simple thing called the Internet protocol.”31
It might be a bit hard to see how a principle of network design could mat-
ter much to issues of public policy. Lawyers and policy types don’t spend
much time understanding such principles; network architects don’t waste
their time thinking about the confusions of public policy.
But architecture matters.32 And arguably no principle of network archi-
tecture has been more important to the success of the Internet than this sin-
gle principle of network design—e2e. How a system is designed will affect
the freedoms and control the system enables. And how the Internet was de-
signed intimately affected the freedoms and controls that it has enabled.
The code of cyberspace—its architecture and the software and hardware
that implement that architecture—regulates life in cyberspace generally. Its
code is its law. Or, in the words of Electronic Frontier Foundation (EFF)
cofounder Mitch Kapor, “Architecture is politics.”33
To the extent that people have thought about Kapor’s slogan, they’ve
done so in the context of individual rights and network architecture. Most
think about how “architecture” or “software” or, more simply, “code” en-
ables or restricts the things we think of as human rights—speech, or privacy,
or the rights of access.
That was my purpose in Code and Other Laws of Cyberspace. There I
argued that it was the architecture of cyberspace that constituted its free-
dom, and that, as this architecture was changed, that freedom was erased.
Code, in other words, is a law of cyberspace and, as the title suggests, in my
view, its most significant law.
But in this book, my focus is different. The question I want to press here
is the relationship between architecture and innovation—both commercial
innovation and cultural innovation. My claim is that here, too, code mat-
ters. That to understand the source of the flourishing of innovation on the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 36

3 6 L a w r e n c e L e s s i g

Internet, one must understand something about its original design. And
then, even more important, to understand as well that changes to this origi-
nal architecture are likely to affect the reach of innovation here.

so which code matters? Which parts of the architecture?34
The Internet is not a novel or a symphony. No one authored a beginning,
middle, and end. At any particular point in its history, it certainly has a
structure, or architecture, that is implemented through a set of protocols
and conventions. But this architecture was never fully planned; no one de-
signed it from the bottom up. It is more like the architecture of an old Eu-
ropean city, with a central section that is clear and well worn, but with
additions that are many and sometimes confused.
At various points in the history of the Net’s development, there have been
efforts at restating its principles. Something called “RFC 1958,” published
in 1996, is perhaps the best formal effort. The Internet was built upon “re-
quests for comments,” or RFCs. Researchers—essentially grad students—
charged with the task of developing the protocols that would eventually
build the Internet developed these protocols through these humble requests
for comments. RFC 1 was written by Steve Crocker and outlined an under-
standing about the protocols for host (“IMP”) software. Some RFCs specify
particular Internet protocols; some wax philosophical. RFC 1958 is clearly
in the latter camp—an “informational” document about the “Architectural
Principles of the Internet.”35
According to RFC 1958, though “[m]any members of the Internet com-
munity would argue that there is no architecture,” this document reports
that “the community” generally “believes” this about the Internet: “that the
goal is connectivity, the tool is the Internet protocol and the intelligence is
end-to-end rather than hidden in the network.”36 “The network’s job is to
transmit datagrams as efficiently and flexibly as possible. Everything else
should be done at the fringes.”37
This design has important consequences for innovation—indeed, we can
count three:

• First, because applications run on computers at the edge of the net-
work, innovators with new applications need only connect their com-
puters to the network to let their applications run. No change to the
computers within the network is required. If you are a developer, for ex-
ample, who wants to use the Internet to make telephone calls, you need
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 37

T H E F U T U R E O F I D E A S 3 7

only develop that application and get users to adopt it for the Internet
to be capable of making “telephone” calls. You can write the applica-
tion and send it to the person on the other end of the network. Both of
you install it and start talking. That’s it.
• Second, because the design is not optimized for any particular existing
application, the network is open to innovation not originally imagined.
All the Internet protocol (IP) does is figure a way to package and route
data; it doesn’t route or process certain kinds of data better than others.
That creates a problem for some applications (as we’ll see below), but it
creates an opportunity for a wide range of other applications too. It
means that the network is open to adopting applications not originally
foreseen by the designers.
• Third, because the design effects a neutral platform—neutral in the
sense that the network owner can’t discriminate against some packets
while favoring others—the network can’t discriminate against a new in-
novator’s design. If a new application threatens a dominant application,
there’s nothing the network can do about that. The network will remain
neutral regardless of the application.

The significance of each of these consequences to innovation generally
will become apparent as we work through the particulars that follow. For
now, all that’s important is that you see this design as a choice. Whether or
not the framers of the network understood what would grow from what they
built, they built it with a certain philosophy in mind. The network itself
would not control how it would grow. Applications would. That was the key
to end-to-end design. As the inventor of the World Wide Web, Tim Berners-
Lee, describes it:

Philosophically, if the Web was to be a universal resource, it had to be able
to grow in an unlimited way. Technically, if there was any centralized
point of control, it would rapidly become a bottleneck that restricted the
Web’s growth, and the Web would never scale up. Its being “out of con-
trol” was very important.38

network architects Saltzer, Clark, and Reed were not the only peo-
ple to notice the value of an end-to-end design. Quite independently, if
later, the idea became apparent within AT&T itself. In the early 1990s,
while trying to implement an improvement in the voice quality of the
AT&T network (competition was beginning to have an effect: the effort was
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 38

3 8 L a w r e n c e L e s s i g

in response to the claim by Sprint that on its network you could hear a pin
drop), Bell Labs researcher David Isenberg became increasingly frustrated
with the “smart” network that AT&T was: at every layer in the distributional
chain, the AT&T network had been optimized for voice telephony. But this
optimization meant that any effort to change a layer in the AT&T distribu-
tional chain would disable other layers. Tweaking one part threw other parts
into disarray. The system was in no sense “modularized,” so change became
impossibly difficult.
This led Isenberg to a treasonous thought: what if the problem was in the
fundamental design of the network itself? What if the whole idea of a smart
network was a mistake? What if a better design would be a “stupid network,”
with intelligence built into the devices, and the network itself kept as simple
as possible?39
Isenberg had arrived through frustration at Saltzer, Clark, and Reed’s fun-
damental insight: A simple, or, as Isenberg described it, stupid network
would facilitate the greatest degree of innovation. A smart, or intelligent,
network would perhaps be optimized for certain users, but its own sophisti-
cation would inhibit different or new uses not initially understood. By
“build[ing] in assumptions about what the business proposition of the net-
work is, you constrain what’s possible.”40 The AT&T network was burdened
by the intelligence built into it. A simpler design could beat the sophisti-
cated design, at least along the dimension of innovation and change.
When Isenberg started to discuss his seditious thoughts, his employer,
AT&T, was not happy. In the early summer of 1997, he was permitted to
post a reply to an article that sang the virtues of smart networks. But soon
after his article was posted, it was republished in many different places on
the Net. Finally, in August 1997, Harry Newton published the article in his
Computer Telephony magazine—without AT&T’s permission. Isenberg be-
came the enemy from within the AT&T network. He was told not to accept
invitations from others to discuss his ideas. This control, understandably,
became intolerable. As he told me, “[T]he AT&T pension became portable
on January 1, 1998. I quit on January 2, 1998.”
However disliked high up within the fortress, Isenberg’s ideas began to
catch on both outside and inside. The virtues of “stupid networks” became
increasingly obvious, as the power of this simple network, the Internet, be-
came undeniable. Isenberg’s idea echoed the end-to-end principle: the two
were the same, and both showed why the Internet would flourish.
* * *
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 39

T H E F U T U R E O F I D E A S 3 9

the internet isn’t the only network to follow an end-to-end design,
though it is the first large-scale computer network to choose that principle
at its birth. The electricity grid is an end-to-end grid; as long as my equip-
ment complies with the rules for the grid, I get to plug it in.41 Conceivably,
things could be different. In principle, we might imagine that every device
you plug into a grid would register itself with the network before it would
run. Before you connected, you would have to get permission for that
device. The owner of the network could then choose which devices to pro-
hibit.
Likewise, the roads are end-to-end systems. Any car gets to enter the high-
way grid (put tolls to one side). As long as the car is properly inspected, and
the driver properly licensed, whether and when to use the highway is no
business of the highway. Again, we could imagine a different architecture:
each car might first register with the grid before it got on the highway (the
way airlines file flight plans before they fly).
But these systems don’t require this sort of registration, likely because,
when they were built, such registration was simply impracticable. The elec-
tronics of a power grid couldn’t handle the registration of different devices;
roads were built stupid because smart roads were impossible. Things are dif-
ferent now; smart grids, and smart roads, are certainly possible. Control is
now feasible. So we should ask, would control be better?
In at least some cases, it certainly would be better. But from the perspec-
tive of innovation, in some cases it would not. In particular, when the future
is uncertain—or more precisely, when future uses of a technology cannot
be predicted—then leaving the technology uncontrolled is a better way of
helping it find the right sort of innovation. Plasticity—the ability of a system
to evolve easily in a number of ways—is optimal in a world of uncertainty.
This strategy is an attitude. It says to the world, I don’t know what func-
tions this system, or network, will perform. It is based in the idea of uncer-
tainty. When we don’t know which way a system will develop, we build the
system to allow the broadest range of development.
This was a key motivation of the original Internet architects. They were
extremely talented; no one was more expert. But with talent comes hu-
mility. And the original network architects knew more than anything that
they didn’t know what this network would be used for.
As David Reed describes, “[T]here were a lot of experiments in those
days,” and “we . . . realized that [there] was very little in common [other]
than the way they used the network. There were sort of interesting ways that
they used the network differently from application to application. So we felt
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 40

4 0 L a w r e n c e L e s s i g

that we couldn’t presume anything about how networks would be used by
applications. Or we wanted to presume as little as possible. . . . We basically
said, ‘Stop. You’re all right’ as opposed to running a bake-off.”42 These de-
signers knew only that they wanted to assure that it could develop however
users wanted.
Thus, end-to-end disables central control over how the network develops.
As Berners-Lee puts it, “There’s a freedom about the Internet: as long as we
accept the rules of sending packets around, we can send packets containing
anything to anywhere.”43 New applications “can be brought to the Internet
without the need for any changes to the underlying network.”44 The “archi-
tecture” of the network is designed to be “neutral with respect to applica-
tions and content.”45 By placing intelligence in the ends, the network has no
intelligence to tell which functions or content are permitted or not. As RFC
1958 puts it, the job of the network is simply to “transmit datagrams.” As the
NRC has recently concluded:

Underlying the end-to-end argument is the idea that it is the system or ap-
plication, not the network itself, that is in the best position to implement
appropriate protection.46

In chapter 2, I introduced the idea of a commons. We can now see how
the end-to-end principle renders the Internet an innovation commons,
where innovators can develop and deploy new applications or content with-
out the permission of anyone else. Because of e2e, no one need register an
application with “the Internet” before it will run; no permission to use the
bandwidth is required. Instead, e2e means the network is designed to assure
that the network cannot decide which innovations will run. The system is
built—constituted—to remain open to whatever innovation comes along.
This design has a critical effect on innovation. It has been, in the words
of the NRC, a “key to the explosion of new services and software applica-
tions” on the Net.47 Because of e2e, innovators know that they need not get
the permission of anyone—neither AT&T nor the Internet itself—before
they build a new application for the Internet. If an innovator has what he or
she believes is a great idea for an application, he or she can build it without
authorization from the network itself and with the assurance that the net-
work can’t discriminate against it.
At this point, you may be wondering, So what? It may be interesting (at
least I hope you think this) to learn that the Internet has this feature; it is at
least plausible that this feature induces a certain kind of innovation. But
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 41

T H E F U T U R E O F I D E A S 4 1

why do we need to worry about this feature of the Internet? If this is what
makes the Internet run, then as long as we have the Internet, won’t we have
this feature? If e2e is in the Internet’s nature, why do we need to worry about
e2e?
But this raises the fundamental point: The design the Internet has now
need not be its design tomorrow. Or more precisely, any design it has just
now can be supplemented with other controls or other technology. And if
that is true, then this feature of e2e that I am suggesting is central to the
network now can be removed from the network as the network is changed.
The code that defines the network at one time need not be the code that de-
fines it later on. And as that code changes, the values the network protects
will change as well.

the consequences of this commitment to e2e are many. The birth of
the World Wide Web is just one. If you’re free from geekhood, you are likely
not to distinguish the WWW from the Internet. But in fact, they are quite
distinct. The World Wide Web is a set of protocols for displaying hyper-
linked documents linked across the Internet. These protocols were devel-
oped in the late 1980s by researchers at the European particle physics
lab CERN—in particular by Tim Berners-Lee. These protocols specify
how a “Web server” serves content on the WWW. They also specify how
“browsers”—such as Netscape Navigator or Microsoft’s Internet Explorer—
retrieve content on the World Wide Web. But these protocols themselves
simply run on top of the protocols that define the Internet. These Internet
protocols, referred to as TCP/IP, are the foundation upon which the proto-
cols that make the World Wide Web function—HTTP (hypertext transfer
protocol) and HTML (hypertext markup language)—run.48
The emergence of the World Wide Web is a perfect illustration of how in-
novation works on the Internet and of how important a neutral network is to
that innovation. Tim Berners-Lee came up with the idea of the World Wide
Web after increasing frustration over the fact that computers at CERN
couldn’t easily talk to each other. Documents built on one system were not
easily shared with other systems; content stored on individual computers
was not easily published to the networks generally. As Berners-Lee writes:

Incompatibility between computers had always been a huge pain in every-
one’s side, at CERN and anywhere else. . . . The real world of high-energy
physics was one of incompatible networks, disk formats, and character-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 42

4 2 L a w r e n c e L e s s i g

encoding schemes, which made any attempt to transfer information be-
tween computers generally impossible. The computers simply could not
communicate with each other.49

Berners-Lee thus began to think about a system to enable linking among
documents—through a process called “hypertext”—and to build this link-
ing on top of the protocols of the Internet. His ideal was a space where any
document in principle could be linked to any other and where any docu-
ment published was available to anyone.
The components of this vision were nothing new. Hypertext—links from
one document to another—had been born with Vannevar Bush,50 and
made famous by Bill Atkinson’s HyperCard on the Apple Macintosh. The
world where documents could all link to each other was the vision of Robert
Fano in an early article in the Proceedings of the IEEE.51 But Berners-Lee
put these ideas together using the underlying protocol of the Internet.
Hyperlinked documents would thus be available to anyone with access to
the Internet, and any document published according to the protocols of the
World Wide Web would be available to all.
The idea strikes us today as genius. Its success makes us believe the idea
must have been obvious. But what is amazing about the story of the birth of
the World Wide Web is how hard it was for Tim Berners-Lee to convince
anyone of the merit in the plan. When Berners-Lee tried to sell the plan at
CERN, management was unimpressed. As Berners-Lee writes:

What we hoped for was that someone would say, “Wow! This is going to be
the cornerstone of high-energy physics communications! It will bind the
entire community together in the next ten years. Here are four program-
mers to work on the project and here’s your liaison with Management In-
formation Systems. Anything else you need, you just tell us.” But it didn’t
happen.52

When he went to a meeting of hypertext fans, he could get few to under-
stand the “ah-ha” of hypertext on the Net. For years he wandered from ex-
pert to expert, finding none who understood the potential here. And it was
only after he started building the Web out, and started informing ordinary
people on a hypertext mailing list about the protocols he was developing,
that the Net started to grow.
The experts didn’t get it. Someone should put that on a bumper sticker
and spread it around. Those controlling the resources of the CERN com-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 43

T H E F U T U R E O F I D E A S 4 3

puter lab wouldn’t support the technology that would give the world the
Web. Only those innovators outside of the control of these managers saw
something of the potential for the Web’s growth.
Berners-Lee feared that competing protocols for using the Internet would
wipe away interest in the WWW. One protocol built about the same time
was called Gopher. Gopher enabled the easy display of a menu of options
from a site. When you went to a Gopher-enabled site, you would see a list
of links that you could then click on to perform some function. Gopher was
extremely popular as an Internet application—running on the Internet
protocols—and use of Gopher took off in the early 1990s.53
But for the purposes that Berners-Lee imagined, Gopher was extremely
limited. It would not enable the easy construction of interlinked docu-
ments. It was closer to a universal menuing system than a system for linking
ideas. Berners-Lee was afraid that this inferior standard would nonetheless
stick before the new and better WWW became well known.
His fear, however, was not realized, both because of something Berners-
Lee did and because of something the creators of Gopher did—and both
are lessons for us.
Berners-Lee was no bully. He was not building a protocol that everyone
had to follow. He had a protocol for displaying content on the World Wide
Web—the HTML language that Web pages are built in. But he decided not
to limit the content that one could get through a WWW browser to just Web
pages. Instead he designed the transfer protocol—HTTP—so that a wide
range of protocols could be accessed through the WWW—including the
Gopher protocol, a protocol for transferring files (FTP), and a protocol for
accessing newsgroups on the Internet (NNTP). The Web would be neutral
among these different protocols—it would in this sense interconnect.54
That made it easy to use the Web, even if one wanted to get access to Go-
pher content. But the second doing was much more important to the death
of Gopher as a standard.
As Berners-Lee describes it, high off its success in populating the world
with Gopher, the University of Minnesota—owner of the right to Gopher—
suggested it might exercise its rights to charge for the use of the Gopher pro-
tocol.55 Even the suggestion of this terrified developers across the world.
(It was, Berners-Lee writes, “an act of treason.”56) Would developers be
hijacked by the university once they depended upon their system? How
much would they lose if the platform eventually turned against the devel-
opers?
Berners-Lee responded to this by convincing CERN to release the right
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 44

4 4 L a w r e n c e L e s s i g

to the Web to the public. At first he wanted to release the protocol under the
GPL, or General Public License (the “GNU General Public License,”
which we will see much more of in chapter 4). But when negotiations over
that bogged down, he convinced CERN simply to release the rights into the
public domain. Anyone had the right to take and use the protocols of the
WWW and build anything upon them that they wanted.57
The birth of the Web is an example of the innovation that the end-to-end
architecture of the original Internet enabled. Though no one quite got it—
this the most dramatic aspect of the Internet’s power—a few people were
able to develop and deploy the protocols of the World Wide Web. They
could deploy it because they didn’t need to convince the owners of the net-
work that this was a good idea or the owners of computer operating systems
that this was a good idea. As Berners-Lee put it, “I had designed the Web so
there should be no centralized place where someone would have to ‘regis-
ter’ a new server, or get approval of its contents.”58 It would be a “good idea”
if people used it, and people were free to use it because the Internet’s design
made it free.

thus two networks—the network built by AT&T and the network we call
the Internet—create two different environments for innovation. One net-
work centralizes creativity; the other decentralizes it. One network is built
to keep control of innovation; the other constitutionally renounces the right
to control. One network closes itself except where permission is granted; the
other dedicates itself to a commons.
How did we get from the one to the other? What moved the world gov-
erning our telecommunications system from the centralized to the decen-
tralized?
This is one of the great forgotten stories of the Internet’s birth. Everyone
knows that the government funded the research that led to the protocols
that govern the Internet.59 It is part of the Internet’s lore that it was the gov-
ernment that pushed network designers to design machines that could talk
to each other.60 The government in general, and the Defense Department
in particular, had grown tired of spending millions for “autistic computing
machines.”61 It therefore wanted some system for linking the systems.
Yet we are practically trained to ignore another form of governmental in-
tervention that also made the Internet possible. This is the regulation that
assured that the platform upon which the Internet was built would not turn
against it.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 45

T H E F U T U R E O F I D E A S 4 5

The physical platform on which the Internet took off came prewired. It
was the telephone wires that linked homes to homes. But the legal right to
use the telephone wires to link to the Internet did not come preordained.
That right had to be earned, and it was regulation that earned it. Nothing
guaranteed that modems would be permitted on telephone lines. Even
today, countries in Asia regulate the use of modems on telephone lines.62
What was needed before the revolution could begin was permission to con-
nect the Net to this net.
And what made that permission possible? What made it possible for a
different use to be made of the telephone wires from that which AT&T had
originally imagined?
Here a second kind of regulation enters the story. Beginning in force in
1968, when it permitted foreign attachments to telephone wires, continuing
through the 1970s, when it increasingly forced the Bells to lease lines to
competitors, regardless of their purpose, and ending in the early 1980s with
the breakup of AT&T, the government increasingly intervened to assure
that this most powerful telecommunications company would not interfere
with the emergence of competing data-communications companies.
This intervention took many forms. In part it was a set of restrictions on
AT&T’s permissible businesses.63 In part it was a requirement that it keep its
lines open to competitors.64 In part it was the general fear that any effort to
bias communications more in its favor would result in a strong reaction
from the government.65
But whatever the mix, and whichever factor was most significant, the
consequence of this strategy was to leave open the field for innovation
in telecommunications. AT&T did not control how its wires would be
used, because the government restricted that control. By restricting that
control, the government in effect created a commons on AT&T’s wires.
In a way analogous to the technical requirements of end-to-end, then,
these regulations had the effect of leaving the network open and hence of
keeping the use of the network neutral. Once the telephone system was
used to establish a circuit, the system was kept free for that circuit to send
whatever data across it the user wished. The network thus functioned as a
resource left open for others to use.
This is end-to-end operating at a different layer in the network design. It
is end-to-end not at the layer determining the connection between two
phones on the telephone system. That connection may well be formed by a
system that does not comply with the end-to-end rule.
But once the circuit is connected, then the environment created by the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 46

4 6 L a w r e n c e L e s s i g

mix of technical principles and legal rules operating upon the telecommu-
nications system paralleled an end-to-end design at the network layer. This
mix of design and control kept the telephone system open for innovation;
that innovation enabled the Internet.

are there costs to the e2e design? Do we lose something by failing to
control access to the resources—the bandwidth—of the network?
Certainly the Internet is not without its weaknesses. The capacity of the
Net at any one moment is not infinite, and though it grows more quickly
than the demand, it does at times get congested. It deals with this conges-
tion equally—packets get transported on a first-come, first-served basis.
Once packets leave one end, the network relays them on a best-efforts basis.
If nodes on the network become overwhelmed, then packets passing across
those nodes slow down.66
For certain applications, “best efforts” is not enough. Internet tele-
phony, for example, doesn’t do well when packets carrying voice get de-
layed. Any delay greater than 250 milliseconds essentially makes the system
unusable.67 And as content on the Net moves to real-time, bandwidth-
demanding technology, this inability to guarantee quality of service be-
comes increasingly costly.
To deal with this problem, technologists have begun to propose changes
to the architecture of the Net that might better enable some form of guar-
anteed service. These solutions generally pass under the title “Quality of
Service” (QoS) solutions. These modifications would enable the network to
treat different “classes” of data differently—video, for example, would get
different treatment from e-mail; voice would get different treatment from
the Web.
To enable this capacity to discriminate, the network would require more
functionality than the original design allowed. At a minimum, the network
would need to be able to decide what class of service a particular applica-
tion should get and then treat the service accordingly. This in turn would
make developing a new application more complex, as the programmer
would need to consider the behavior of the network and enable the appli-
cation to deal with that behavior.
The real danger, however, comes from the unintended consequences of
these additional features—the ability of the network to then sell the feature
that it will discriminate in favor of (and hence also against) certain kinds of
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 47

T H E F U T U R E O F I D E A S 4 7

content. As the marketing documents from major router manufacturers
evince, a critical feature of QoS solutions will be their ability to enable the
network owner to slow down a competitor’s offerings while speeding up its
own—like a television set with built-in static for ABC but a clear channel
for CBS.
These dangers could be minimized depending upon the particular QoS
technology chosen. Some QoS technologies, in other words, are more con-
sistent with the principle of end-to-end than are others.68 But proponents
of these changes often overlook another relatively obvious solution—
increasing capacity.69 That is, while these technologies will certainly add
QoS to the Internet, if QoS technologies like the “RSVP” technology do so
only at a significant cost, then perhaps increased capacity would be a
cheaper social cost solution.70
Put differently, a pricing system for allocating bandwidth solves certain
problems, but if it is implemented contrary to end-to-end, it may well do
more harm than good.
That is not to argue that it will do more harm than good. We don’t know
enough yet to know that. But it raises a fundamental issue that the scarcity
mentality is likely to overlook: The best response to scarcity may not be a
system of control. The best response may simply be to remove the scarcity.
This is the promise that conservative commentator George Gilder re-
ports. The future, Gilder argues, is a world with “infinite” bandwidth.71 Our
picture of the Net now—of slow connections and fast machines—will soon
flip. As copper is replaced with glass (as in fiber optics) and, more impor-
tant, as electronic switches are replaced by optical switches, the speed of the
network will approach the speed of light. The constraints that we know from
the wires we now use will end, Gilder argues. And the end of scarcity, he
argues, will transform all that we do.72
There is skepticism about Gilder’s claims about technology.73 So, too,
about his economics. The economist in all of us can’t quite believe that any
resource would fail to be constrained; the realist in all of us refuses to be-
lieve in Eden. But I’m willing to believe in the potential of essentially infi-
nite bandwidth. And I am happy to imagine the scarcity-centric economist
proven wrong.
The part I’m skeptical about is the happy progress toward a world where
network owners simply provide neutral fat (or glass) pipe. This is not the
trend now, and there is little to suggest it will be the trend later. As law pro-
fessor Tim Wu wrote to me about Gilder’s book:
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 48

4 8 L a w r e n c e L e s s i g

I think it is a “delta dollar sign” problem as we used to say in chemistry (to
describe reactions that were possible, but not profitable). Private actors
seem to only make money from infrastructure projects if built with the
ability to exclude. . . . [H]ere in the industry, all the projects that are “hot”
are networks with built-in techniques of exclusion and prioritization.74

Here is a tragedy of the commons. If the commons is the innovation com-
mons that the protocols of the Net embrace, e2e most important among
them, then the tragedy of that commons is the tendency of industry to add
technologies to the network that undermine it. But this is an issue for the
dark part of this book. For now, my aim is only brightness: to get you to see
the commons that has been built through a set of protocols that defined the
Internet that was.

the internet was born on a controlled physical layer; the code layer,
constituted by the TCP/IP, was nonetheless free. These protocols expressed
an end-to-end principle, and that principle effectively opened the space
created by the computers linked to the Net for innovation and change. This
open space was an important freedom, built upon a platform that was con-
trolled. The freedom built an innovation commons. That commons, as do
other commons, makes the controlled space more valuable.75
Freedom thus enhanced the social value of the controlled: this is a lesson
that will recur.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 49

C o m m o n s A m o n g t h e W i r e d

W IRED is a magazine that was first published in early 1993. Its title is un-
defined, but it aspires to signal those who are connected or, as one on-
line dictionary puts it, “with it” with respect to all things digital. To those
outside the world of “things digital,” the “wired” are those caffeine-chugging
techheads staring at C code as the clock chimes 0100 (military time). But to
those inside digital culture, “the wired” are those who understand the poten-
tial of this place called cyberspace and who are making that potential real.
The character of this group has changed. In the early 1990s, they were
more intrigued by fast code than fast cash. Today, it is more the opposite. Yet
if there is a group that can still be called “connected”—those who have built
and are building the Internet that we have come to know—then this chap-
ter is about them, about the commons among them, and about the innovation
this commons built.
This commons had three aspects. One is a commons of code—a com-
mons of software that built the Net and many of the applications that run on
the Net. A second is a commons of knowledge—a free exchange of ideas
and information about how the Net, and code that runs on the Net, runs.
And a third is the resulting commons of innovation built by the first two
together—the opportunity, kept open to anyone, to innovate and build
upon the platform of the network.
A certain culture made each of these commons possible, as did a certain
feature about the stuff these coders built—code. Something, that is, about
the norms that first defined this world, as well as something about the na-
ture of the code. My aim in this chapter is to explore both the character of
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 50

5 0 L a w r e n c e L e s s i g

this culture and the nature of this code, and how the two interact to produce
a layer of freedom at the content layer.
For the content layer is the layer at which the commons in this chapter
lives. The commons of the last chapter, built by end-to-end, is a commons
at the code layer of the network. The commons here lies on top, even
though built, like the code layer below it, in software. Code here is content,
and at the birth of the Net, much of this content was free.
As will become clearer in chapter 11, however, the content of code is not
fundamentally different from the content we are more familiar with—
music, or film, or (at least digital) texts. As I will argue, in the digital world,
all the stuff protected by copyright law is in one sense the same: It all de-
pends fundamentally upon a rich and diverse public domain. Free content,
in other words, is crucial to building and supporting new content. The free
content among the “wired” is just a particular example of a more general
point.

to introduce these commons, however, we need to think a bit more
about code. Our world is increasingly constituted by environments built in
code—in the instructions inscribed in either software or hardware. Yet our
intuitions about “code” are likely to be incomplete.
“Code” is written (primarily) by humans, though the code that humans
write is quite unlike the code that computers run. Humans write “source
code”; computers run “object code.” Source code is a fairly understandable
collection of logical languages designed to instruct the computer what it
should do. Object code is a string of ones and zeros impenetrable to the or-
dinary human. Source code, however, is too cumbersome for a computer to
run; it is therefore “compiled” before it is run, meaning translated from
human-readable to machine-understandable code.
Object code is therefore the lifeblood of the computer, but it is the source
code that links computers and humans. To understand how a program runs;
to be able to tinker with it and change it; to extend a program or link it to
another—to do any of these things with a program requires some access to
the source.1
Things were not always this way. When computers were first built, they
didn’t have “software.” Their functions were literally wired into the ma-
chines. This way of coding was obviously cumbersome. By the early 1960s,
it was essentially replaced.2 While some computer functions are still per-
formed by “hard-wired” code (for example, the code in the ROM chip that
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 51

T H E F U T U R E O F I D E A S 5 1

is executed when you boot up your computer), the meat of computers today
is software.
At first, no one much cared about controlling this code. In the beginning
of commercial computing, computer companies wrote software, but that
software was peculiar to each company’s machine. Each company had its
own operating system (OS, the underlying program upon which all other
programs are run). These operating systems were not compatible. A pro-
gram written for an IBM machine would not run on a Data General ma-
chine. Thus, the companies had very little reason to worry about a program
being “stolen.” Computer companies were in the business of selling com-
puters. If someone “stole” a program meant for a particular computer, they
could run it only if they had that computer.
This was a world of incompatible machines, and that troubled those who
depended upon many different kinds of machines to do their work. The
government, for example, spent millions on computers but grew frustrated
that these machines could not talk with one another. The same was true of
the company that would build perhaps the most important operating system
in the history of computing: AT&T.
For in this chapter, for at least this part of this chapter, AT&T is the hero.
AT&T purchased many computers to run its national network of phones.
Because of a consent decree with the government in 1956, however, it was
not permitted to build and sell these computers itself. It was therefore de-
pendent upon the computers that others built and frustrated, like the
government, by the fact that these other computers couldn’t talk to each
other.3
Researchers at Bell Labs, however, decided to do something about this.
In 1969, Ken Thompson and Dennis Ritchie began an operating system
that could be “ported” (read: translated) to every machine.4 This operating
system would therefore be a common platform upon which programs could
run. And because this platform would be common among many different
machines, a program written once could—with tiny changes—be run on
many different machines.
In the history of computing, this urge for a cross-platform-compatible lan-
guage was long-standing. ALGOL was an early example.5 So too was
COBOL, when the government announced that it would not purchase or
lease any computer equipment that could not run COBOL.6 But the birth
of Unix—the name given to AT&T’s ur–operating system—was the most
important. For not only did AT&T develop this foundational operating sys-
tem, it also gave it away. Because of the restrictions imposed by the 1956
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 52

5 2 L a w r e n c e L e s s i g

consent decree, AT&T was not allowed to sell a computer operating system.
Thus, Thompson and Ritchie succeeded in convincing the company to
simply give the OS to anyone who wanted it.
The first takers of this free OS were universities.7 Computer science de-
partments could use the source code to teach their students about how op-
erating systems were written. The system could be critiqued, just as English
grad students can critique Shakespeare because they have the text of the
Shakespeare plays to read. And as this system became understood, fixes to
bugs in this system were contributed back to AT&T. The process produced
a vast and powerful network of people coming to speak the language of Unix
and of a generation growing up tutored by Unix.
In this way, for this period, Unix was a commons. The code of Unix was
a commons; the knowledge that this code generated was a commons; and
the opportunity to innovate with and on top of this code was a commons.
No one needed permission from AT&T to learn how its file system worked
or how the OS handled printing. Unix was a trove of knowledge that was
made available to many. Upon this treasure, many built.

over time, however, the openness of commercial code began to change.
As products became more numerous and users became more diverse, and as
the cross-platform compatibility of programs grew, the companies produc-
ing these products exercised more and more control over how the products
might be used. The code thus “forked”—developing in different and in-
compatible ways, increasingly proprietary. Users became less partners in the
process of developing and using computer systems and more consumers.
And suppliers of code were less eager to permit their code to be copied by
others.
One instance of this increase in control turned out to be quite important
in the history of computing. Richard Stallman was a researcher at MIT. He
was an early disciple of the norms of openness (as in “the open society”) or,
more generally, freedom. The whole world, of course, was not open. But
throughout much of the 1970s, the norm in computing was. Exceptions
were scorned.
The lab where Stallman worked had a printer connected to the network.
The clever coders in this lab had written a program to notify them when the
printer malfunctioned. A jam, for example, would generate a message to
users on the network, and someone close to the printer could then go cor-
rect the problem.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 53

T H E F U T U R E O F I D E A S 5 3

In 1984, the lab updated the software (a driver) that ran the printer. Stall-
man then asked the company supplying the printer for a copy of the source
code so that he could replicate the notification function in this new version
of the printer driver. The company refused. The code to the printer driver
was now closed, Stallman was informed. No one was allowed to tinker
with it.8
To Stallman, this was a moral offense. The knowledge built into that
driver had been produced by many people, not all of whom had been
employed by the company. There was something wrong, then, with the
company locking up that knowledge. And this wrong sowed the seed in
Stallman’s mind of a movement to resist this closing.9
In 1985, that movement was born. Stallman founded the Free Software
Foundation. Its aim was to encourage the development of software that car-
ried its source with it. The aim was to assure that the knowledge built into
software was not captured and kept from others. The objective was to sup-
port a commons for code.10
AT&T gave the movement an important boost, quite unintentionally. In
1984, after AT&T was broken up, the company was freed of the restrictions
on computing that it had been living under since 1956. Once freed of these
restrictions, AT&T decided to enter the computing business. One of its
most important assets in this business was Unix. Hence, AT&T decided to
exercise control over Unix. After 1984, Unix would no longer be free. Com-
panies, universities, and individuals wishing to use Unix would have to li-
cense the right from AT&T.
To many, this too was betrayal. A generation had devoted its professional
career to learning and building upon the Unix system. Now AT&T claimed
the exclusive right to the product of this learning. Although AT&T had
taken the suggestions that had been made, although Unix had been im-
proved in response, the company now wanted to trade on these improve-
ments by making the code exclusive and unfree.
The reactions against AT&T’s take-back were sharp. Berkeley had a
version of Unix that it had been distributing; after AT&T’s change, the
Berkeley release had to undergo a massive retooling to extract the AT&T
code so that it could release a version of Unix (BSD Unix) that was free of
AT&T’s restrictions.
But Stallman responded in a more productive and ambitious way. He
wanted to build a free version of Unix that would, by design, always be free.
So the Free Software Foundation launched project GNU—a recursive
acronym meaning “GNU’s not Unix.” The GNU project was first to develop
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 54

5 4 L a w r e n c e L e s s i g

the suite of tools necessary to build an operating system and then to use
those tools to build the GNU OS.
Throughout the 1980s, Stallman worked to do just that. Slowly he added
tools to the project. Beginning with an extraordinary editor, Emacs,
and then, even more important, with a compiler, the GNU C Compiler
(GCC),11 the project slowly pulled together the tools an operating system
would need.
But as the 1980s came to an end, Stallman’s project began to slow down.
Stallman had developed a problem with his hands. For a while he lost the
ability to type. As he turned to the final step in the project—building a ker-
nel (the heart of an OS) for the GNU operating system—his pace had been
cut dramatically. He had mixed all of the ingredients needed for an operat-
ing system to function, but he was missing the core.12

in finland, a young student studying computer science wanted the
chance to experiment with an operating system. Unix was the gold standard;
Linus Torvalds had little gold.13
Instead, Torvalds started playing with Minix, an educational version of an
OS released by computer science professor Andrew Tannenbaum. Minix
ran on a PC but was designed as a teaching tool. It was therefore incom-
plete. So in 1990, Torvalds began building an alternative to Minix, which
he released to the Internet in 1991. That code was released subject to a li-
cense called the General Public License (GPL). (We’ll see more of this
later.) It was therefore free for anyone to take and use, as long as they didn’t
bottle up what they took.
People rapidly realized, however, that with a little bit of work linking the
parts of an OS that Stallman had built to the core of the OS that Torvalds
had released, Stallman’s objective of an open and free Unix-flavored OS
could be realized. Quite quickly, then, Linux—or GNU/Linux for those
who want to keep the contributions in view—was born.14 GNU/Linux was
a platform that came with its source; anyone could take and build upon this
platform. Because it came with its source, anyone could tinker with it to
make it better. Many did, and in a very short period of time, GNU/Linux be-
came quite good.15

so goo d , in fact, that GNU/Linux is now the fastest-growing operating
system in the world. It is supplied by a host of companies, commercial and
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 55

T H E F U T U R E O F I D E A S 5 5

non-. It has been ported to every major chip—the “most widely ported op-
erating system available for PCs,” Torvalds reports.16 There is a version of
Linux on the PPC chip (used by Macintosh computers). There is a version
on the chip used by Sun. It is quickly becoming the Unix that Thompson
and Ritchie imagined—a platform to which the world can “write once, run
anywhere.”17 Like Mother Nature herself, GNU/Linux is quickly becoming
universal and free.

in the terms that I have described, GNU/Linux is an “open code” proj-
ect. It is software that carries its source code with it, and it requires that its
source be kept available for others. The source code can be viewed and
modified by a user; parts can be taken and used by other coders. It there-
fore builds a commons of (1) code, (2) knowledge, and (3) innovation upon
that code.
But Linux is just one example of a large number of open code projects
that populated the network at its birth. It is neither the first, nor the only, nor
even perhaps the most important open code project. To keep the impor-
tance of the open code movement in view, we should remember these other
projects as well.
For example, more successful (in market share, at least) and just as im-
portant is the project that built the Apache server. A server is that part of the
Net infrastructure that “serves” content. When you go to a Web page and
view the contents of that page, the computer delivering the content is a
“server.” In the first days of the Web, the expectation was that companies
like Netscape Corporation would build and sell servers. Netscape did, as did
others, including Microsoft. But in addition to these commercial ventures,
there was a version of a Web server that was made available for free. This
was the HTTP’d server produced by the National Center for Supercomput-
ing Applications (NCSA).
The NCSA server was adequate but buggy. Because it was government
funded, however, its source code was free. The only requirement the NCSA
imposed on users, or modifiers, of the NCSA code was that they give NCSA
credit.
A group of early adopters of the HTTP’d protocol began to share
“patches” to the server—improvements as well as bug fixes that would make
the code run more efficiently. At a certain point, the group decided to form
a collective that would build a new server on top of the NCSA server. This
server—called the Apache server both because of the pride associated with
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 56

5 6 L a w r e n c e L e s s i g

the Apache name and because, as a series of patches, it was “a patchy”
server—was then released to the world as a free, open source server.
Apache quickly took off. As its market share grew, users fed more patches
back. Developers worked hard to integrate these patches. The quality of the
server quickly improved. Apache soon became the number one server in the
world. To this day, two-thirds of the servers on the World Wide Web are
Apache servers. As writer Glyn Moody describes it, “[A]lthough the debate
still rages fiercely about whether open source software such as GNU/Linux
can ever hope to best Microsoft, Apache has already done it.”18
The coders who built the Apache servers were not paid by any company
called “Apache.” Many of the developers worked part-time, paid by the
companies they worked for. One of the leaders of the project, Net wizard
Brian Behlendorf, says this “essential volunteerism” was crucial for the proj-
ect. The work done for the project had to come from people who were mo-
tivated to help the project, not from people just paid to code from nine to
five.
Linux and Apache are the two most prominent open code projects. But
there are others still. Perl, developed by Larry Wall, is a programming lan-
guage that enables high-power manipulation of text. It is the glue that
makes most Web sites run. It too was developed as an open source project,
and it is by far the dominant language of its class, ported to every important
operating system. And deeper in the guts of the Internet’s code are other sys-
tems that are even more crucial to the Net. The Berkeley Internet Name
Domain (BIND) system (which connects names, such as lessig.org, to IP
addresses, such as 123.45.67.890) was developed originally as an open code
project at the University of California at Berkeley; at version 4.9.2, Paul
Vixie became its principal architect.19 Some believe that BIND is “the sin-
gle most mission-critical Internet application.”20 Whether it is or not, it was
built through open code. Likewise was “sendmail,” which processes mail
routing, an open code project developed by Eric Allman.21 It now runs on
75 percent of servers in the world.22
These projects together constitute the soul of the Internet. Together with
the public domain protocols that define the Internet (governed by no li-
cense at all but free for all to take, referred to collectively as TCP/IP, in-
cluding the core protocols for the World Wide Web), this free code built the
Internet. This is not a single program or a single operating system. The core
of the Internet was this collection of code built outside the proprietary
model.
For the property obsessed, or those who believe that progress comes only
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 57

T H E F U T U R E O F I D E A S 5 7

from strong and powerful property rights, pause on this point and read it
again: The most important space for innovation in our time was built upon
a platform that was free. As Alan Cox, second only to Linus Torvalds in the
Linux chain, puts it in an essay in response to Microsoft’s attack on open
code values:

[M]ost of the great leaps of the computer age have happened despite,
rather than because of, [intellectual property rights (IPR)]. [B]efore the
Internet the proprietary network protocols divided customers, locked them
into providers and forced them to exchange much of their data by tape.
The power of the network was not unlocked by IPR. It was unlocked by
free and open innovation shared amongst all.23

Not strong, perfect control by proprietary vendors, but open and free pro-
tocols, as well as open and free software that ran on top of those protocols:
these produced the Net.

this free code builds a commons. This commons in turn lowers the cost
of innovation. New projects get to draw upon this common code; every proj-
ect need not reinvent the wheel. The resource thus fuels a wide range of in-
novation that otherwise could not exist.
Free code also builds a commons in knowledge. This commons is made
possible by the nature of information. My learning how a Web page is built
does not reduce the knowledge of how a Web page is built. Knowledge, as
we’ve seen, is nonrivalrous; your knowing something does not lessen the
amount that I can know.
There is something particular about how free code builds this knowledge
commons. Code is performative; what it says, it does. Hence one learns
about code not just by reading the code, but also by implementing it.
Think about the code that builds the World Wide Web. Web pages are
written (primarily) in a markup language called HTML. That language is a
set of tags that mark text or graphics to be displayed on a Web page. Every
major Web browser has a function that reveals the source of the Web page
being viewed. If you see a page and want to see how it was built, you simply
“reveal source” on the page, and the Web page turns into the set of codes
that generated the page.
This feature of the World Wide Web meant that it was extremely easy for
coders to learn how to build Web pages. Most of the early learning was sim-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 58

5 8 L a w r e n c e L e s s i g

ply copying a page and modifying it as the coder wished. Even if the code
building a Web page was copyrighted, others could learn from it. Again, be-
cause the code here performs as well as expresses, the feedback for teaching
was strong.
This feature of the World Wide Web was chosen. There was no necessity
that the source code for a Web page be viewable. Other languages—such as
Apple’s AppleScript language—enable authors to easily hide the code that
makes the script run. Hiding could well have been the default on the World
Wide Web.
But had it been the default, then the knowledge commons of the World
Wide Web would have been vastly smaller. And had it been smaller, the
growth and innovation of the World Wide Web would have been much less.
Designing the Web to be open meant that the Web would grow more quickly.
Finally, and most important to the argument of this book, free code at
the content layer builds a commons in innovation, just as end-to-end at the
code layer does. By keeping its teachings open, and hence by assuring that
others can build upon these teachings differently, free code assures that in-
novation cannot be chilled. Like the simple, neutral network of e2e, free
code hasn’t the power to discriminate against new innovation.
This innovation commons, however, unlike end-to-end, is protected by
law. The principles of end-to-end are protected (if at all) through norms.
Software engineers decide whether to implement software that complies
with end-to-end. But law protects the innovation commons that open code
built.
My aim in the balance of this chapter is to show just how.

the law that protects the innovation commons built by open code is a
combination of contract law and copyright. We will consider copyright law
in greater depth in chapter 11, but for now suffice it to say that a copyright
attaches to essentially any creative work at the time that creativity is fixed in
a tangible form. Your e-mails are copyrighted at the time you write them,
your love letters when you pen them.
The law creates this “exclusive right”—aka monopoly right—to help
solve a problem that exists with creative information.24 As we saw in chap-
ter 2, information is naturally nonrivalrous. If you use it, I still have as much
left as before. It is also naturally nonexcludable. In Jefferson’s poetry, “[H]e
who receives an idea from me, receives instruction himself without lessen-
ing mine; as he who lites his taper at mine, receives light without darkening
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 59

T H E F U T U R E O F I D E A S 5 9

me.”25 But these two features create an incentive problem for authors. If
there were no way for an author to control the copies of his or her created
work, then there would be no easy way for the author to profit from that
work. For some authors, that would mean less time spent producing. And
that in turn would mean less creative work.
Copyright tries to remedy this problem by giving authors a limited
right—an exclusive right to control some uses of their work for a lim-
ited time. This right is designed to remedy the problem of incentives.
But obviously it does the author no good if the author must exercise his or
her exclusive right alone. Writers aren’t publishers. Thus, through contracts
(written against the background of copyright law), the right to use copy-
righted work is granted. That granting is a contract called a “license.”
So far I’ve been speaking of “open code,” or “free code,” but these terms
describe two very different kinds of licenses. The original open code license
was developed by the Free Software Foundation. This is the “General Pub-
lic License,” or GPL.
The GPL sets the terms under which one has the right to use free soft-
ware.26 Among software licenses generally, the GPL is unique. While most
licenses serve to limit the copies a licensee may make, the GPL serves to
limit the restrictions on copying that a licensee can make. Anyone can use
or modify a GPL work, as long as, in the words of the license preamble,
“you . . . give the recipients all the rights that you have. You must make sure
that they, too, receive or can get the source code.”27 As Richard Stallman de-
scribes it:

A program is free software, for you, a particular user, if: You have the free-
dom to run the program, for any purpose. You have the freedom to modify
the program to suit your needs (To make this freedom effective in practice,
you must have access to the source code, since making changes in a pro-
gram without the source code is exceedingly difficult.). You have the free-
dom to redistribute copies, either gratis or for a fee. You have the freedom
to distribute modified versions of the program, so that the community can
benefit from your improvements.28

The consequence of this license is that copyrighted work licensed under
the GPL is always available to others to use or modify as they wish, as is the
code that derives from GPL-protected code. This combination of copyright
law and contract law essentially renders GPL code, as well as code deriva-
tive of GPL code, “free.”
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 60

6 0 L a w r e n c e L e s s i g

The GPL is not the only license used within the open code movement. A
second class of licenses is ordinarily called “open source.” Open source li-
censes are less restrictive (or less freedom enhancing, depending upon your
perspective) than free software licenses. While free software licenses require
that derivative work be free, open source licenses have no such requirement
(where, again, “free” means that the source code must be made available to
other users, not necessarily free of charge).29 Microsoft is free to use the
Apache server (licensed under an open source license) and incorporate it
into a proprietary product; Microsoft is not free to take Linux and incorpo-
rate it into Windows.
This is an important difference, no doubt, but not so much for our pur-
poses here. The GPL is enough of a model to understand the argument that
follows, so the argument that follows presumes the GPL as the license.
Where there is a reason to distinguish, I will be careful to make sure you see
the differences in these two kinds of (legal) code.30 But for now, let’s assume
the legal code governing this free code is the code of the GPL.

as i said, the Linux operating system is licensed under the GPL. That
means that the source code of Linux travels with the code itself. Anyone can
tinker with the code; anyone can see how it works.
Any code that derives from this open code must itself be licensed under
GPL. This code, and the knowledge it teaches, is therefore open and free.
Proponents of open code sell it on the promise that users are not hostage
to open code projects. That claim is true, but in two very different senses—
one small, one big. The smaller claim is just that users, or coders adopting
open code, can tinker with that code. They can control the code, changing
it as they wish. As Peter Wayner puts it, the “free source world . . . [is] a re-
turn to the good old days when you could take apart what was yours.”31 You
can add functionality or fix functionality that is imperfect as supplied.
Other operating systems are not open or free. Apple, for example, licenses
and sells Macintosh computers. Macintosh computers come with the
Mac OS. Bundled in the package of programs shipped with the Mac OS 9
is a program called File Synchronization. The aim of File Synchronization
is to allow a user to keep two sets of files synchronized. If you have laptop
and desktop computers, then the objective of File Synchronization is to
allow you to make sure that the files on both systems are kept up to date.
But File Synchronization has a problem. If you try to synchronize files
whose icons have been customized (icons are the pictures on the desktop
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 61

T H E F U T U R E O F I D E A S 6 1

that you double-click on to open a file), then the program gets an “out of
memory” error. The program thus has a bug; that bug makes it impossible
to use the program on systems where the icons have been modified. But, of
course, modifying the icons is one of the freedoms the Mac OS builds into
its system. So those who take advantage of this freedom in the Mac OS sys-
tem can’t take advantage of the free synchronization program that Apple
bundles with the computer.
All code has bugs; there’s nothing special about the Mac OS in this re-
spect. But what is significant about this story is how long this simple bug sur-
vived. Though this program, File Synchronization, was released in 1998,
the bug remained, even in 2000. Fixing this memory allocation error in this
small part of the Mac OS was not important to Apple Computer. So the bug
remained unsquashed.
Had at least this part of the Mac OS been open, however—had the
source code been available—then no doubt someone out there would have
found it worthwhile to correct the mistake that the code had created. This
coder could have taken the code that the Mac OS supplied and fixed the
problem that the Mac OS created. And whether the coder sold the fix to
others or not, the fix would have been made long before two years had
passed.32 Users of the File Synchronization program would not have been
held hostage to the flaws in that program.

the freedom to tinker is thus an important freedom. But there is a
much more fundamental sense in which open code assures that users are
not held hostage, and it is this feature of open code that links it to the prin-
ciple of end-to-end discussed in the previous chapter. Just as end-to-end’s
openness assures a neutral network that runs your code and doesn’t turn
against your innovation, open code’s openness assures that the foundation
of the computing environment is neutral and can’t turn against the innova-
tor. An open code platform keeps a platform honest. And honest, neutral
platforms build trust in developers.
We can see something of the importance of this by passing quickly over
the most important lawsuit affecting cyberspace so far—United States v.
Microsoft.33
The government’s claim in United States v. Microsoft was essentially this:
While Microsoft had built an important platform upon which developers
across the world had constructed code, Microsoft had adopted a practice
that chilled certain kinds of innovation. When an innovator had a tech-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 62

6 2 L a w r e n c e L e s s i g

nology that threatened Microsoft’s platform, Microsoft, the government
claimed, adopted a strategy to kill that innovation. The platform, in other
words, turned against some kinds of innovation, while no doubt protecting
others. And that pattern, the government alleged, stifled innovation in just
the way that AT&T’s control over the telephone network had stifled innova-
tion inconsistent with AT&T’s business model.34 Unlike a network that re-
mained committed to end-to-end, Microsoft, the government argued, built
discrimination into its platform, and this discrimination harmed innovation.
To understand the government’s claim requires a bit of context. About
the time that Torvalds released to the world the kernel of the Linux OS,
Microsoft was releasing to the world its latest version of a program called
Windows. This latest version was receiving an extraordinary response in the
user community. Windows 3.0 was finally a stable and powerful tool simu-
lating a “graphical user interface” (GUI) operating system. It was not itself
an operating system; it was simply a program that ran on the underlying op-
erating system DOS. But it was efficient enough that a huge portion of the
DOS market wanted to buy or upgrade to this version of Windows.
Before Windows, users of PCs with the Intel chip architecture—“IBM-
compatible” PCs, as they were called—were stuck with command line
interface.35 When you booted up the system, it presented to you a simple
command line—for example, “C:>”—and it expected you, the user, to then
enter a command to launch a program. It was a Unix-inspired system,
initially intended for the talented who played with Unix.
Microsoft was the originator of the “disk operating system” (DOS).36
When IBM decided to release a personal computer, it licensed the operat-
ing system for that computer from Microsoft. Microsoft, however, guided by
perhaps the smartest businessman of our generation, Bill Gates, kept the
rights to DOS. IBM would develop after a bit its own version of DOS—
called PC-DOS. But Microsoft was always free to build and sell MS-DOS.
MS-DOS quickly became the market leader because Microsoft was bet-
ter at responding to consumer demand. Its coders were better at solving cod-
ing problems. And by the late 1980s, MS-DOS was by far the dominant
operating system in a world dominated by PC-compatible computers.
But there was significant pressure on Microsoft. First, Microsoft was not
the only DOS producer. A company called Digital Research developed
a different DOS—DR-DOS—that was increasingly viewed as superior
to Microsoft’s DOS. In 1990, BYTE magazine said of DR-DOS 5.0 that it
“cures many (but not all) of MS-DOS’s shortcomings” and was a vastly su-
perior DOS.37 Other reviews said much the same.38
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 63

T H E F U T U R E O F I D E A S 6 3

Second, graphical user interfaces were threatening Microsoft. As con-
sumers adopted the PC, they were increasingly frustrated with the com-
mand line interface and started to look at other PC products. The most
attractive of these other PC products was the Apple Macintosh, introduced
by Apple in 1984. The Macintosh offered a simple GUI interface that made
it easy to use a PC.
Gates admired the Mac. His company first wrote Word and Excel for the
Mac. And in 1985, Gates tried to license the Macintosh OS to port it to the
Intel chip architecture.39 In what may have been the worst decision in
corporate computer history, Apple declined the offer. Gates therefore
turned his crew to the task of building Windows.
It is hard for us today to recall the significance of the threat that Gates felt
then. Microsoft was a tiny company compared with Apple. In 1984, Apple’s
annual sales were $1.5 billion; Microsoft’s were $98 million.40 In hindsight,
the inevitable success of Windows seems assured, but at the time, many at
Microsoft certainly felt the threat of a world of competitors, some building
better OSs (DR-DOS) and some building better user interfaces or, more
precisely, GUIs, pronounced “gooeys” (Apple). Gates therefore pushed the
firm to develop something better of each.
By 1991, Gates realized that the future would be a world with one inte-
grated GUI-OS. Windows 95 would be that OS, but it would take some
time to get to Windows 95. Windows 95 would integrate the GUI of Win-
dows 3.0 with DOS, giving users a powerful and integrated system compa-
rable to that of the Apple Macintosh.
The key, however, was to make sure that while making the switch,
Microsoft didn’t lose its customer base. Most were using MS-DOS; every-
one wanted Windows. So the key in the next four years, the company be-
lieved, was to hold the field.
This is where things got legal, and the account that follows is noth-
ing more than the allegations made by the antitrust enforcers (and others)
about how Microsoft responded to the threat it faced. But the allegations
are substantial, the parallel to later allegations is clear, and they will help
make the story of the commons clear if we map this (alleged) story of con-
trol.
To hold the field, Microsoft had to assure that no competitor would suc-
ceed in stealing its operating system customer base. The threat of defection
was strong, given the increasingly strong competition of other DOSs. But
the tool to assure no great defection was the powerful and popular program
Windows 3.0. Everyone wanted Windows 3.0, and it was more important
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 64

6 4 L a w r e n c e L e s s i g

for a computer manufacturer to be selling Windows 3.0 than it was to be
selling any particular DOS.
Microsoft thus used, the government alleged, its power over Windows to
avoid competition with DOS. Both through pricing strategies that made it
economically infeasible for an equipment manufacturer to sell any version
of DOS other than MS-DOS and, according to the European Union’s anti-
trust enforcement wing, by directly tying the sale of Windows 3.0 to the sale
of MS-DOS, Microsoft made it impossible for competitors to gain any
foothold in the Microsoft base. Thus, Microsoft could hold the field of its
own users until it could enable the migration of its users over to Win-
dows 95.
Microsoft was behaving, the government alleged, strategically.41 It was
limiting the options of its customers so as to protect its own market position.
And this form of competition, the government alleged, was anticompetitive.
Rather than allowing competition in operating systems to flourish, Micro-
soft was threatening competitors who were building products threatening it.
In 1994, the government notified Microsoft that it was about to file suit
against it. The claim was illegal pricing and tying behavior in securing the
field for Microsoft products. After extensive negotiations, Microsoft signed a
consent decree, promising the government that it would not engage in a
string of listed behaviors. The government accepted Microsoft’s promises,
and the case with Microsoft was settled.
A consent decree is just a contract. The government promises not to sue
if Microsoft promises not to engage in certain behaviors. Nothing in the
contract concedes that the behavior challenged was illegal; nothing in
the contract binds the government never to sue Microsoft for other anti-
competitive behavior. The decree is just a simple way for the government to
stop the harm it believes a company is doing, without a long and expensive
trial proving that the behavior is, in fact, illegal. If the company that signed
a consent decree breaches the agreement, then the government is able to
get a court to enjoin the breaching behavior.
After the government settled the case, it announced to the world that it
had made the world free for operating system competition. Assistant Attor-
ney General Anne Bingaman promised there would now be a great deal of
competition among operating systems.42 The field had been opened up,
Bingaman said; competitors would flock to the field. Microsoft would face
important competitive pressures; its monopoly position would be erased.
Microsoft had a different view of the decree. It believed it would have no
real effect. Said Bill Gates to the press, “None of the people who run those
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 65

T H E F U T U R E O F I D E A S 6 5

divisions are going to change what they do or think or forecast. Nothing.
There’s one guy in charge of [hardware company] licenses. He’ll read the
agreement.’ ”43
It turned out that Microsoft was the better predictor. Soon after the an-
nouncement of the decree, DR-DOS folded. IBM announced it was
pulling its development of its competing operating system, OS/2. All the
competitors for the desktop PC-compatible operating system essentially dis-
appeared. The consent decree was law, but reality was unaffected. The de-
cree was too late in coming, and by the time it came, the reality it regulated
had changed.
In 1997, however, the government was back at it. Once again, the gov-
ernment claimed, Microsoft was mucking about with the market. Once
again, the government argued, it was behaving strategically to disable com-
petition. This time, however, the strategy was not to protect DOS. This time
the aim was to protect Windows 95 against a nascent form of competition,
the Internet.
Before 1995, Microsoft had not yet come to understand what the Internet
would become. Bill Gates didn’t take it seriously. But in 1995, Gates got
religion. In a series of meetings, he increasingly made it clear that he be-
lieved the Internet was the next great revolution and the revolution that was
Microsoft’s greatest threat.
When Microsoft woke up to the revolution, there was an image that kept
Gates awake at night. The story went something like this: Microsoft’s power
came from the fact that applications developers wrote their applications to
the Microsoft platform. To the extent they continued to write to the Micro-
soft platform, Microsoft’s power would remain strong. But the Internet
presented a different opportunity for developers. Through a browser tech-
nology like Netscape, tied to an application programming language like
Java, developers would increasingly find it valuable to write programs to the
network directly. This would reduce dependence on Windows, which
would in turn reduce the power of the Microsoft monopoly. A partial sub-
stitute for Windows then threatened to undermine the power of Windows.44
Whether you believe this threat or not, there is fairly clear evidence that
Bill Gates believed this threat.45 Hence Microsoft once again shifted into
defensive mode. As with DOS in the early 1990s, it needed to hold the field
with users of Windows 95 until it could develop a network-based applica-
tion suite that would compete with the threat posed by Java/Netscape.
It held the field with a familiar strategy. As with the MS-DOS/
Windows 3.0 strategy, the aim here was to assure that a significant portion of
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 66

6 6 L a w r e n c e L e s s i g

the base did not move away from Windows 95 before Windows 95 could be
migrated into a network-based world. This time the strategy would be im-
plemented not through a fancy application that everyone would want. This
time the tie would be effected by linking a Microsoft browser to the under-
lying operating system, Windows 95. Every computer manufacturer that
wanted to sell a PC (with an Intel architecture) had to sell it with Win-
dows 95. Microsoft took advantage of that fact by making sure the buyers
also got Internet Explorer.46
How it made sure the two went together changed over time. At first, the
program was a simple addition bundled within the operating system—just
as File Synchronization on the Macintosh is bundled with the OS. But
eventually the program was “integrated” into the operating system, making
it easy for programs to move to the Internet and hard for another browser to
be located within the Windows 95 system.
The government charged that this bundling behavior was in effect the
same sort of tie as the one attacked between Windows 3.0 and DOS. It also
claimed the objective was the same: Microsoft was using its power over the
operating system to behave strategically against an innovation that threat-
ened it.

the story of this battle between the government and Microsoft is not yet
(at this writing) over. But the significance of the case to the argument that
follows has little to do with what the appellate courts eventually say. For
whether the claim is proven or not, its essence has a parallel to the lessons
from chapter 3: Microsoft was accused of strategic behavior designed to pro-
tect its monopoly position. It was accused, that is, of using its power over the
operating system to kill innovation that threatened this power. As Wall
Street Journal writer David Bank puts it:

Any product that was popular represented a potential threat to the Win-
dows platform because it could become a platform itself. Integration was
Microsoft’s weapon for disabling the threats to Windows. . . . Anyone who
competed against Microsoft’s platform lost.47

Again, that’s not to say that Microsoft chilled innovation generally. Obvi-
ously, and again, the Windows platform has been an extraordinary boon for
innovation. Thousands of programs have been written for it; hundreds of
thousands of coders have used its structure to their own advantage. But
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 67

T H E F U T U R E O F I D E A S 6 7

sometimes an innovation challenges Microsoft, either by challenging the
monopoly Microsoft has been said to hold or by making its business model
less attractive in the future. And here, the government charged, Microsoft
has been quick to respond. If your business model threatens it, then Micro-
soft will respond by killing your business. Gates was not, the government al-
leged, willing to play on a level playing field. (Indeed, as David Bank quotes
Gates “yelling” at executive Paul Maritz, “You’re putting us on a level play-
ing field! You’re going to kill the company.”)48
Whether or not you believe that Microsoft engaged in the behavior that
the government charged (and whether or not the Court of Appeals and the
Supreme Court ultimately rule that such conduct violates the antitrust
laws), the important point is this: Microsoft could engage in the behavior al-
leged by the government only because Microsoft controlled its own code.
The source code for the Microsoft operating system is closed; Microsoft
does not reveal the source to the public generally. Thus, Microsoft can
change and direct its source code in ways that advance its own strategic vi-
sion. It is capable, that is, of behaving strategically, by changing its code to
challenge competitors, because its code is closed. It can “control the pace
of innovation” because only it can muck about with its code.49
Yet this is just the power that open code doesn’t have. An open code proj-
ect can’t bundle a product the users don’t want; users, because source code
is there, are always free to unbundle. An open source project can’t under-
mine a competing system; the competing system is always free to take the
open source system and fight back. The source code for open source proj-
ects is therefore a check on the power of the project; it is a limit on the
power of the project to behave strategically against anything written to the
platform.
This “check” is realized in the perpetual possibility of an open code proj-
ect to “fork.” Forking occurs when a project led in one direction splits and
develops in two or more directions. This right to split is guaranteed both by
the code (because the source code is available) and by the law (because the
license for open code projects guarantees that people are free simply to take
the code and develop it in different directions). From the same code base,
developers are free to develop different versions. These different versions
can take on a life of their own. And conceivably, a fork could divide the user
community into different sects. A project thriving because it had x thousand
users could then collapse if not enough users support the project.
This threat of forking is not idle. Important open source projects, such as
the BSD Unix clone, have forked in the past.50 There is nothing in the cur-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 68

6 8 L a w r e n c e L e s s i g

rent licenses of open code projects that would undermine this threat in the
future.51 Instead, the possibility of forking keeps pressure on the guardians
of an open code project to develop the project in a way that the broadest
range of users wants. As author Peter Wayner puts it, “[I]t prevents one per-
son or clique from thwarting another group.”52
This is democracy brought to code. An open code system can’t get too far
from the will of the users without creating an important incentive among
some users to push the project a different way. And this in turn means the
platform cannot act strategically against its own. The threat that created a
federal case in United States v. Microsoft is not a threat if an operating sys-
tem is built on open code. And the absence of strategic behavior in turn in-
spires others to build for this code.
Now my claim is not that this neutrality is the only factor that affects
whether coders build to one platform rather than another. If 99 percent of
the world is on a closed platform, and 1 percent is on an open platform,
then regardless of the benefits of the 1 percent platform, there are naturally
strong pressures to code for the 99 percent.
Yet this factor is not well enough understood, and its effect on innovation
is systematically ignored. The effect here is similar to the effect we’ve seen
in other contexts, and our aim, here and in other contexts, should be to bet-
ter appreciate its role in building the Internet we have seen.

the users of an open code project are not therefore hostages. That’s the
lesson argued so far. They are not hostages to bad code—the right to tinker
is assured. And they are not hostages to strategic code—open code can’t be-
have strategically. These two features together constitute the innovation
commons that the Internet creates. They capture the public value that open
code supports.
But there is a challenge with open code projects that many believe is in-
surmountable. This is the challenge to assure that there are sufficient in-
centives to build open code. Open code creates a commons; but the
problem with this sort of commons is not the problem of overgrazing. (In-
deed, as “accidental revolutionary” Eric Raymond puts it, open code creates
an “inverse commons.” “Grazing” does not reduce the code that is avail-
able. Instead, “in this inverse commons, the grass grows taller when it’s
grazed on.”)53
The problem instead is to assure a sufficient incentive to supply new or
improved code—a provisioning problem, in other words. In a world where
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 69

T H E F U T U R E O F I D E A S 6 9

software is sold like chewing gum, and where great value is believed to re-
side in the power to control who can copy this code, it is hard for many to
see how there would be enough incentive to build code that is given away
to anyone who wants it.
Here, however, we must work as empiricists, not ideologues. For we just
have to look around to see the extraordinary amount of open code being
written, despite the inability to control its copying. As Richard Stallman has
said, “We do develop a lot of free software. If a theory says we can’t, you have
to look for the flaws in the theory.”54 The fact of this coding means that
coders must have very different reasons for participating in open code proj-
ects. This reality means that the ability to control the code is not necessary
for individuals to have an incentive to code.
Instead, there are plenty of examples of businesses that find it worthwhile
to support open code projects without an assurance of perfect control. IBM
is the most famous example. In 1998, IBM decided to dump its own server
product and embrace the open source Apache server. It was free to do this
because the open source Apache server was free for the taking. And IBM
was inspired to do this because of the quality of that server.
This began an important relationship between IBM and open code. After
it adopted the Apache server, IBM then embraced the Linux operating sys-
tem. Rather than support ten different operating systems for its ten different
systems, IBM found a great benefit in standardizing on a single system. In
2000, the market value of Linux-related IBM servers was $30 million. By
2004, IBM expects this value to increase to $3.4 billion.55 And because the
operating system across IBM’s different computer systems will be the same,
the ease with which code from one project can be carried to another will
increase.
But IBM has not simply leeched off the wealth of open code projects. It
has committed its own resources—in excess of $1 billion, it reports—to sup-
porting the development of Linux and Apache. The company is therefore
committing its own money for something it might otherwise get for free.
Why? What incentive could a company have to pay for what it could get
for free?56
IBM’s behavior is at first hard to understand, but not if you focus a bit on
the nature of the project it supports.
IBM makes money by selling equipment. It sells more equipment if the
software that runs that equipment is better. Thus, the free software it sup-
ports simply adds to the value of the equipment it sells.
More important, IBM adds services on top of the free software it supports.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 70

7 0 L a w r e n c e L e s s i g

The Apache server software has been folded into a suite of software that
IBM sells—WebSphere. WebSphere does a range of Internet-related func-
tions that Apache doesn’t. The value of WebSphere therefore increases as
the power of Apache improves.
Thus, IBM’s willingness to improve Apache and Linux is not in itself
hard to understand. The puzzle is why it gives its improvements back to the
public. Why doesn’t it simply take the Apache server and fold it into Web-
Sphere but then keep its improvements to itself? Why, in other words,
doesn’t IBM simply take the resources of Apache and then defect from the
open source movement?
The incentive not to defect comes from something special about the
character of software development. Software is not static; it needs to evolve.
If IBM were to fork Apache, taking a version of the Apache server pri-
vate, it would face an even greater cost in keeping that forked version up to
date as the functionality of the server was improved. The contributions of
others to the Apache design would have to be folded into IBM’s proprietary
design. Tracking and implementing those changes would be extremely
costly.
It therefore makes more sense—from a purely commercial perspective—
for IBM (at least) to support the open coding that builds Linux and Apache,
even though it can’t capture the full value of the code it contributes.
Or from a different perspective, IBM loses more from hiding its improve-
ments than it gains.57
If this behavior still seems bizarre, then you need to put it into a broader
context. The reigning view about software speaks as if a rational company
would never write code unless it has perfect control over what it produces.
But perfect control is rarely assured in any free market, not with code or any-
thing else.
This is a hard fact for lawyers to understand (protected as they are by
exclusionary rules such as the bar exam), but most of production in our so-
ciety occurs without any guarantee of government protection. Starbucks
didn’t get a government monopoly before it risked a great deal of capital
to open coffee shops around the world. All it was assured was that people
would have to pay for the coffee they sold; the idea of a high-quality coffee
shop was free for others to take. Similarly, chip fabricators around the world
invest billions in chip production plants, with no assurance from the
government that another competitor won’t open a competing plant right
next door.
In each of these cases, and in the vast majority of cases in a free economy,
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 71

T H E F U T U R E O F I D E A S 7 1

one person’s great idea is open for others to take. Burger King and McDon-
ald’s; Peet’s Coffee and Starbucks; Peapod and Webvan. No doubt the first
movers would like it if others couldn’t use their idea or if others wouldn’t
notice their idea until long after a market is set. But it is in the nature of the
limits on patent rights, and in the nature of transparency in the market, that
innovators in the ordinary market can’t keep their good ideas to themselves.
Some protection for ideas, and a bit more for expression, is provided by the
legal system. But this protection is incomplete or leaky. Perfect control is
never its character.
Innovators nonetheless innovate. And they innovate because the return
to them from deploying their new idea is high, even if others get the bene-
fit of the new idea as well. Innovators don’t simply sit on their hands until a
guaranteed return is offered; real capitalists invest and innovate with the un-
derstanding that competitors will be free to take their ideas and use them
against the innovators.
Thus, rather than puzzling about why anyone would code for free sys-
tems, we might as well puzzle about why anyone would innovate without a
government-granted monopoly to protect them. Indeed, history will teach
that, at an earlier time, this was very much the view. Mercantilists believed
that exclusive rights were needed before any investment made sense; the
English monarchy at an earlier time protected many ordinary investments
through a state-backed monopoly.
Free markets, however, function on a very different basis. We don’t grant
every merchant a guaranteed market; we don’t reward every new marketing
plan with a twenty-year monopoly; we don’t grant exclusive rights to each
new way of doing business. In all these cases, because the market produces
enough incentive on its own, the fact that others can free-ride doesn’t kill in-
novation.
The same lesson is being relearned in the context of code. No doubt IBM
would be happier if it could control improvements to Apache; but the re-
turn from better server sales is enough to induce IBM to invest without get-
ting the benefit of perfect control.58 Likewise if IBM offers services that run
on its servers: fast-running, more reliable servers will make it easier to sell
the services that would run on top.
What’s true for Big Blue is being learned elsewhere as well. Mercantilism
among coders is dying. And as it dies, coders learn what free markets have
taught since Smith called them free: that innovation is best when ideas flow
freely.59
This is not to argue that software should be totally free, or that innovation
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 72

7 2 L a w r e n c e L e s s i g

should be completely unprotected. I am not an opponent of protection,
where protection is properly justified. My aim is not to argue against systems
of control generally. It is simply to resist a mistaken inference: that if some
control is good, then more control must be better.

my aim in this chapter has been twofold. The first part has been to intro-
duce the idea of open code and to demonstrate how it operates at the con-
tent layer to inspire a wide range of innovation. It does this both for the
reasons that technologists give—it is fast, cheap, and powerful—and for rea-
sons that are too often missed. By offering to the world a wide range of code
and hence coding resources, open code lowers the barriers to entry for in-
novators.60 By building a neutral platform, open code invites a different kind
of innovation. By protecting that neutral platform, both through licenses
and through distributed source code, the system assures developers that the
platform will remain neutral in the future.
This feature of open code, however, is not limited to code. The lesson of
open code extends to other content as well. As we will see when we consider
the law of copyright, this balance between free and controlled resources is
precisely the balance that the law must strike in intellectual property con-
texts generally. And while our intuition is that more control produces more
innovation, this commons among the wired suggests at least that the story is
more complex. Less control over code at the content layer has arguably pro-
duced more innovation and development of this code. Keeping this re-
source in a commons increases the value of the resource—both because
others can draw upon this resource and because it mitigates the number
of strategic games played by others. We will see something more of these
strategic games in chapter 11.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 73

C o m m o n s , W i r e - l e s s

“
T he radio spectrum” refers to that swath of electromagnetic radio
frequencies that are used today for everything from the transmitting of
AM radio to the broadcasting of television and cellular phones. Technically
it refers to the use of radio waves, for any purpose, between 3 kilohertz and
300 gigahertz.
This spectrum is regulated. The Titanic gave us that regulation. In the
aftermath of her sinking, navy analysts argued that had the radio spectrum
been better regulated, a ship less than twenty miles from the wreck could
have saved hundreds of passengers.1 The chaos in the spectrum confused
the ship, however, so it missed the calls of help from the sinking luxury
liner. The government used this confusion as a reason to begin to regulate
access and use of the spectrum.
By the fall of 1912, the push to extend this regulation was great. Congress
enacted the Radio Act of 1912, vesting in the Secretary of Commerce the
right to license the operation of a radio apparatus.2 In 1926, after a series of
court decisions limiting the power of the Department of Commerce, then
Secretary of Commerce Herbert C. Hoover said the authority was insuffi-
cient. Congress responded with the Radio Act of 1927, vesting in the Fed-
eral Radio Commission (FRC) control over the radio spectrum.3 The FRC
thus established a process by which the right to use a certain spectrum was
licensed. Any use without a license was a criminal offense.
Thus spectrum, after 1927, at least, was not a commons. To use the spec-
trum required the permission of someone else—the government. That per-
mission was granted according to the government’s view of what uses were
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 74

7 4 L a w r e n c e L e s s i g

best. There was no neutrality in the government’s decisions about who got
to use this “public” resource. This was a resource that was fundamentally
controlled, with the government as the controller.
This control had an increasingly profound effect upon radio program-
ming. Early radio programming was different from today’s. The spectrum
was not filled with commercial broadcasters and Rush Limbaugh. Indeed,
there was no such thing as a radio commercial. Radio at its start looked a lot
like the Internet at its start. Broadcasters on early radio included a wide
range of noncommercial, religious, and educational services. Commercial
radio was just a tiny fraction of the total.4
But once the government got involved, all this quickly changed. It is an
iron law of modern democracy that when you create a regulator, you create
a target for influence, and when you create a target for influence, those in
the best position to influence will train their efforts upon that target. Thus,
commercial broadcasters—NBC and CBS in particular—were effective in
getting the government to allocate spectrum according to their view of how
spectrum should be used.5 (This was helped by the broadcasters’ practice of
offering free airtime to members of Congress.)6 The period from 1927 to
1934 saw an extraordinary shift in the nature of radio use—from a diverse
collection of uses, some commercial, most not, to a single dominant use of
the radio spectrum—namely, commercial radio. As Thomas Hazlett writes,
“[B]y the mid-1930s, [NBC and CBS] would be responsible for an as-
tounding 97% of night-time broadcasting.”7
This transition was not without opposition. When radio stations started
advertising, they incited a massive and continuous campaign of opposition.
Herbert Hoover said of the trend, “It is inconceivable that we should allow
so great a possibility for service to be drowned in advertising chatter.”8 Poll
after poll indicated that the people hated the emerging commercial system.9
Over time, however, people got used to the commercials, and the oppo-
sition died. By the mid-1930s, Congress was ready for a new statute, the
Communications Act of 1934. The act charged a renamed agency (the Fed-
eral Communications Commission) with the duty to regulate “as public in-
terest, convenience or necessity” requires within certain spectrum-defining
areas.10 And it empowered the FCC to make decisions about how best to
use the spectrum in the public interest.
This extensive regulation of what before 1912 had been a purely unregu-
lated practice of wireless communication was upheld by the Supreme
Court in 1946. Regulation of the radio spectrum was necessary, Justice Felix
Frankfurter argued, because “[t]here is a fixed natural limitation upon the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 75

T H E F U T U R E O F I D E A S 7 5

number of stations that can operate without interfering with one another.”11
Justice Frank Murphy, though dissenting from the Court’s opinion, agreed
with the Court at least this far:

Owing to its physical characteristics[,] radio, unlike the other methods of
conveying information, must be regulated and rationed by the govern-
ment. Otherwise there would be chaos, and radio’s usefulness would be
largely destroyed.12

It was in the nature of things, the government argued and the Court
agreed, that only if spectrum were controlled by the government would
spectrum be usable. Spectrum could not be free.

about the time the Supreme Court came to this conclusion, an English
economist was concluding just the opposite. In a review of the FCC’s regu-
lation of spectrum, economist Ronald Coase concluded that there was no
justification for political regulation of access to spectrum.13 Spectrum was
no more “scarce” than land or trees were scarce. Scarcity is the nature of all
valuable resources; but not all valuable resources are allocated by the
government—at least, not in a free society.14
Rather than a regime of licensing, Coase argued, spectrum should be al-
located into property rights and sold to the highest bidder.15 A market for
spectrum would better and more efficiently allocate spectrum than a system
of government-granted licenses.
History has been kinder to Coase than to the regulators of the early FCC.
In 1991, he won a Nobel Prize for his work on transaction cost economics.
And long before the Nobel committee recognized his genius, many policy
makers in the United States came to believe that Coase’s system was better
than the FCC’s. A market in spectrum would more efficiently allocate spec-
trum than any system controlled by the government.
This is the debate I described at the start of the book. It is a debate
between two regimes for controlling access to a resource—in this case, spec-
trum. One regime (the FCC’s) relies upon the government; the other
(Coase’s) relies upon the market. Both presume that spectrum must be con-
trolled. They differ only in the controller. Both thus reject a model of spec-
trum as a commons.
Among these proponents of a market for spectrum, none is more vocal
and persuasive than American Enterprise Institute Fellow Thomas Haz-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 76

7 6 L a w r e n c e L e s s i g

lett.16 A system of government licenses, Hazlett argues, chills innovation. A
world where holders of rights in spectrum cannot sell those rights chills the
process by which new uses of spectrum develop. Far better, Hazlett argues,
if the holders of spectrum rights had the freedom to sell those rights to the
highest bidder. Then, Hazlett argues, more creative and innovative uses of
spectrum would be enabled.17
Hazlett has done an extraordinary service demonstrating the harm of
government-managed spectrum. He is certainly right that the current
regime stifles innovation in spectrum use. If the innovator must first get
permission from the government, then the innovator is much less likely to
try. Permission from the government is an expensive commodity. New ideas
rarely have this kind of support. Old ideas often have deep legislative con-
nections to defend them against the new.
But to demonstrate the harm in government control of a resource is not
yet to demonstrate the need for private control. Hazlett is right if control is
necessary. But is control necessary? Even if the market is a better system for
allocating control than the state, is the market in spectrum better than free
spectrum, if no ex ante allocation is required?

the answer is: Maybe not.18 Increasingly, there are strong technical ar-
guments for a different way of allocating spectrum—or, better, arguments
for a different way of not allocating spectrum. These “different ways” we can
abbreviate as “wideband technologies.” These technologies include “spread
spectrum” technologies as well as technologies that allow some spectrum
uses to be “overlayed” on top of others.19 Wideband technologies would
allow many different users to “share” spectrum without the government or
the market handing out rights to use the spectrum up front. Just as users of
the Internet “share” the resources of the Internet through protocols that co-
ordinate multiple, unplanned use, so too users of spectrum could “share”
the resources of spectrum through protocols that coordinate multiple, un-
planned use. Rather than controlled, spectrum would be, in this model,
“free.” Rather than permission to use it, the right to use it would be granted
to anyone who wanted it. Rather than property, spectrum would be a com-
mons.
This would not mean, as I will explain more fully below, that use of the
spectrum would not be regulated. The regulation would simply be differ-
ent. We speak of the “freeway” system to refer to highways. Highways are
“free” in the sense that I mean: they are a commons open to anyone to use.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 77

T H E F U T U R E O F I D E A S 7 7

But the devices that use the highway system are highly regulated. You can’t
take a go-cart onto Route 66; you can’t drive a tank down your local street.
Regulations control the devices that can be used on a highway. But regula-
tion does not control who gets to go where. Use remains, in our terms, free.
The same could exist for spectrum. But to see how, we need to think a bit
differently about what spectrum is and how it is used. As David Reed says of
policy makers, so is it for most of us: We are “grounded in theory or com-
mon sense [about spectrum] that does not match the phenomena we are
seeing every day.”20

to understand the possibility of free spectrum, consider for a moment
the way old versions of Ethernet worked. Ethernet is the protocol you most
likely use to connect your computer to your company’s local area network.
If you have a cable modem at home, it is the protocol used to connect your
computer to the cable modem. It is essentially a way for many devices on a
single network to “share” the resources of that network. But the critical fea-
ture of this sharing is that it occurs without any central authority deciding
who does what when.
How?
When a machine on an Ethernet network wants to talk with another
machine—when it wants, for example, to send content to a printer, or to
send an e-mail across the Internet through an e-mail server—the machine
requests from the network the right to transmit. It asks, in other words, to re-
serve a period of time on the network when it can transmit. It makes this
reservation only if it hears that the network at that moment is quiet. It be-
haves like a (good) neighbor sharing a telephone party line: first the neigh-
bor listens to make sure no one is on the line, and only then does she
proceed to call. Likewise with the old versions of Ethernet: the machine
would first determine that the network was not being used; if it wasn’t, it
would send a request to reserve the network.21
What if two machines sent that request at the very same time? If that hap-
pened, the network would record a “collision” on the network, and each
machine would register that its request had failed. Each machine would
need to request access to the network again. But rather than each machine
requesting access at the same time, each waits for a random amount of time
until it sends its request again. Ethernet technologies demonstrate that this
protocol for dealing with collisions is quite good at facilitating coordinated
use of a common network.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 78

7 8 L a w r e n c e L e s s i g

In this story, the Ethernet network is functioning as a commons. It is a re-
source that is made available generally to everyone connected to the net-
work. Of course, everyone on the network must request permission to use
the resource. But this permission can be content neutral. The network does
not have to ask what your application is before it reserves space on the net-
work.
More important, these protocols are a way for many different machines to
share this common resource, without the coordination of any top-down con-
troller. No one licenses the use of one machine over another; no system for
selling rights to use the Ethernet network is needed. Instead, many different
machines share access to this common resource and coordinate its use with-
out top-down control.
Ethernet is not radio spectrum, though it is “spectrum in a tube.”22 And
wideband technologies work differently from Ethernet protocols, though
the Ethernet protocols do at least show how bottom-up coordination is pos-
sible. This bottom-up coordination of spectrum in a tube should in turn
suggest the possibility of a different way of controlling spectrum in the air. It
should suggest, that is, the possibility that radio spectrum might be allocated
in this shared bottom-up way, rather than in the traditional top-down model
of coordination advocated by the licensors-of-property types.

how would such a system work?
The existing paradigm of radio spectrum broadcasting embraces the op-
posite of end-to-end principles. The ends in the broadcast medium—
receivers—are stupid, not smart. All the intelligence is in the broadcaster
itself. A receiver just listens for the strong signal separated by silence. When
another strong signal comes close to the signal it’s listening to, existing re-
ceivers get confused. They can’t decide which signal to focus upon, so they
wander in and out among them all.
A different paradigm for broadcasting imagines smart radios (smart re-
ceivers and transmitters) replacing the dumb. These receivers can distin-
guish the transmissions they are to focus on from background noise. They
distinguish the good from the bad either because each transmission, coming
as a packet of data, tells the system where to listen next or because there is a
fixed pattern of listening that the receivers are programmed to follow. In ei-
ther case, smart receivers make it possible for many receivers to effectively
share the same spectrum range. And through technologies that facilitate
coordination—again, analogous to the technologies of Ethernet—this sys-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 79

T H E F U T U R E O F I D E A S 7 9

tem would permit many receivers, and hence many broadcasters, to coordi-
nate use of the same radio spectrum.23
The idea for this way of allocating spectrum reaches back to World
War II, and to the work of actress Hedy Lamarr.24 Lamarr and her partner,
George Antheil, were exploring ways for submarines to communicate with-
out detection. They invented a system where a transmitter would hop along
the radio spectrum—transmitting for a moment at one frequency, and then
jumping at the next moment to another—while the receiver, knowing the
pattern the transmitter would take, would tune to the different frequencies
at precisely the right moment in time.
Lamarr’s technology was taken up by the Defense Department, though
her invention was never deployed. Instead, work on the technology was clas-
sified. In the mid-1980s, however, information about this research was de-
classified, and interest in this mode of using spectrum increased.25 The
deployment of the idea, of course, was now different. Digital processors
made it possible to jump across the spectrum much more quickly and effi-
ciently. And researchers increasingly saw that not only would this be a more
efficient way to use the radio spectrum, but communications using this
technology would be more secure. Rather than simply tuning in to a con-
versation on a cell phone (as many “scanners” do now), the conversation on
the cell phone would be spewed across many different channels. The re-
ceiver would be unable to keep up unless it was clued in to the pattern of
the transmission.
This is the Internet sans wires. The data being transmitted—for instance,
a song or a TV show—are carved up into packets of data; those packets are
sent across the radio spectrum along a broad swath of spectrum. They are
then collected at the other end and reassembled by the smart receiver. Col-
lisions or mistransmissions are retransmitted, as on the Internet. A vast array
of spectrum is in turn effectively shared, in just the way “spectrum in a
tube” (the wires of the Internet) is shared. No central controller is needed,
just as no controller on the Internet is needed. Anyone with an idea, and a
device that obeyed simple spectrum rules, could deploy that idea, just as
anyone with an idea for the Internet, and a computer that obeyed TCP/IP,
could deploy that idea to the whole of the Net.26

around the early 1980s, the rules governing spectrum became an obses-
sion with a retired West Point officer, David Hughes. Hughes had begun on-
line community life in Colorado by setting up one of the first on-line
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 80

8 0 L a w r e n c e L e s s i g

bulletin boards in the nation. His aim was to find a cheap way for commu-
nities to connect, and he was located in rural America, where the thought
of wires being used to connect was neither obvious nor useful.
So Hughes began exploring radios and soon came across the explod-
ing technologies of spread spectrum radio. Using cheap (and increas-
ingly cheaper) radio devices, Hughes began setting up spread spectrum
experiments—demonstrating the power of a technology that did not depend
upon spectrum being owned.
Though this work was technical, Hughes’s motivation was “community—
not politics, not business, not technology, not government—community in
all of its parts.”27 As he explained:

My work with radio has been based upon how you reach the smallest com-
munities, and across community. Not just to it, but within it. . . . It’s always
been to the end of the highest level of connectivity at the lowest cost for
every community on the face of the globe.28

Hughes began to push free access to spectrum. His work was designed to
demonstrate how open spectrum could connect communities much more
cheaply. At the core of his plan was a technology for sharing spectrum rather
than allocating it—in other words, a plan for making the physical layer of
spectrum free by treating the physical layer as if it were in a commons.
Hughes worked for a time with FCC technical adviser Dewayne Hen-
dricks. Hendricks too was eager to exploit this new technology. In the early
1980s, the FCC announced its plan to explore using spectrum as a com-
mons. Hendricks was eager to develop technologies to do just this. While
Hendricks was at the FCC, he pushed Chairman William Kennard’s pro-
gram to experiment with these alternative uses of spectrum. But when the
FCC slowed its progress, Hendricks decided to follow the path of Hughes,
leaving the government to build what many in Washington said could not
be built.
The problem was again the FCC. While the FCC had allocated a range
of spectrum to be “unlicensed”—meaning people could use this spectrum
without receiving a license—it was not encouraging this alternative use. So
Hendricks had the idea to go elsewhere to explore new ways to use the spec-
trum. The Kingdom of Tonga was receptive to this alternative model for
regulating spectrum use. Hendricks packed his bags.
In Tonga, Hendricks built a system to deliver high-speed Internet access
to all citizens in Tonga. This access would use the radio spectrum; the speed
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 81

T H E F U T U R E O F I D E A S 8 1

was two to five times faster than the fastest cable modem in the United
States.29 Once built, the system would deliver this content at just about zero
marginal cost.
Hendricks could build this system in Tonga because he was free of FCC
regulations. Tonga has its own rules for allocating spectrum; it chose to
make a sufficient amount free to enable this free Internet use. Rather than
fight with the skeptics over whether the system would work in theory, Hen-
dricks decided to prove it would work simply by building it.
Hendricks has not stopped with Tonga. Encouraged by the FCC’s push to
develop Internet infrastructure in Native American tribal lands, he has now
begun a program to give Native American tribes access to free spectrum.30
Within eight tribal lands, he is building a similar system to that in Tonga. Na-
tive Americans on those reservations will have access to superfast, supercheap
wireless Internet technologies—long before the rest of America does.
How can Hendricks do this, given the rules of the FCC? Hendricks’s plan
starts within the rules the FCC has set; when he runs against the rules, he
will shift to plan B: The Native American tribes argue that they are sover-
eign nations. The Supreme Court has agreed. Their claim is that they are
free to regulate spectrum on their lands as long as they don’t interfere with
spectrum off their lands. Hendricks’s system won’t interfere. And by the
time the lawyers resolve the battle, these Native American tribes will be con-
nected at a higher speed than the fastest cable modems in AT&T’s labs.
This is regulatory activism in its finest form.

hughes and hendricks are just two of a gaggle of innovators experi-
menting with these alternatives to allocated spectrum. Some of the most fa-
mous innovations are the “Bluetooth” protocols, which enable low-power
connections between mobile devices and PCs. Millions of devices now
embed the protocol, which uses one of the few “unlicensed” bands that the
FCC has allowed.31 Another example is Apple Computer’s AirPort tech-
nology, enabling wireless links to computer networks using a protocol
called 802.11b.32 (Real Madison Avenue whizzes, these protocol namers!)
This technology enables extremely fast wireless connections between com-
puters and a network.
But these are just the beginning.33 Consider the work of Charmed Tech-
nologies. Founded by MIT Media Lab alumnus Alex Lightman, Charmed
Technologies aims to develop wearable computing systems. These wearable
devices will link to the Internet and feed information in real time back to the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 82

8 2 L a w r e n c e L e s s i g

user or wearer. Think of Robocop, with the wearer as the robot—able to see
a person and have the computer identify him or look up an address merely
by pointing the viewer at a building.
Who knows whether such a system would catch on? Who at this point
can tell whether being perpetually connected is what people really want?
But the fact that we can’t tell means the opportunity to experiment is im-
portant. And the opportunity to experiment here depends upon access to
the resources needed to experiment—spectrum, in other words. Ideas like
Lightman’s require space to develop, without first having to prove to exist-
ing AT&Ts why each new idea is a good idea.
If Lightman’s idea depended solely on the Internet—if the last leap were
not across the air but simply a link to a wired computer—then he would
have this right to experiment. The right to connect is guaranteed by a
broadly competitive market for Internet service providers (ISPs). He could
make his service available on the Net, and anyone on the Net could get ac-
cess to it. But because his service depends upon a leap from a person to a
server across the air, he must depend on the right to access wireless spec-
trum. And that right is threatened.
For if there were a broad swath of unlicensed spectrum—spectrum that
anyone could use and many could link to—then many Alex Lightmans
could experiment with ways to link the Net to people and the Net to
things.34 These experiments would generally fail, but a few no doubt would
succeed. And it is these successes that could transform the Internet as it is.
If the same opportunity for innovation and creativity existed around wireless
technologies as existed initially around the Net, then the changes we should
expect are the same as the extraordinary changes the Net has built. Free ac-
cess to this free resource should produce the same sort of innovation that
free access to the controlled resource of telephone lines produced.
This free resource would thus enable wireless access for a wide range of
new services—some still unimagined, others the dream of Internet innova-
tors. And this free resource would compete with other providers of access to
the Internet, keeping competition strong in this critical part of our informa-
tion infrastructure.

by now you will have noticed something different about the argument of
this chapter. Unlike the commons I’ve described so far, a broad commons
in radio spectrum does not yet (generally) exist. And unlike the com-
mons I’ve described so far, with wireless there is not yet a wide range of in-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 83

T H E F U T U R E O F I D E A S 8 3

novation to point to and ponder. Instead, in this chapter my argument is
about a commons that could be, not one that already exists. My claim is that
there is enough evidence of a different way to order spectrum that we
should be exploring whether spectrum could be ordered as a commons.
How exactly would such a regime work? Well, again, to say that spectrum
should be in a commons is not to say that the government would leave spec-
trum “unregulated.” There would be a role for regulation even if spectrum
were “free.” But this regulation would look very different from the regula-
tion that now controls spectrum. The government (or the market) would
not be deciding who gets to use the spectrum. The government would sim-
ply be assuring that the technologies that use the spectrum are properly cer-
tified technologies.35 The FCC would need to certify that the devices were
properly configured. Just as the FCC does now with computers (to make
sure they don’t interfere with radio transmissions), it would do with radios
(to make sure of the same).
Thus, the spectrum-as-commons model does not assume no role for the
government. The role of the government, however, would be much less in-
vasive than under the current regulatory regime. The government does de-
cide who gets to drive on the highways; it doesn’t sell off a right to drive on
the highways; it simply makes sure that the devices that are used on the
highway are certified as safe.

as the technological potential to share spectrum becomes increasingly
clear, a wide range of scholars and technicians is now pushing the FCC to
adopt a very different mode for allocating spectrum.36 These advocates
cover a broad political spectrum. As I’ve suggested over and over in this
book, this diversity makes perfect sense. The advocates for free or open spec-
trum want to enable an extensive range of new technologies. They resist the
efforts by entrenched interests to use government-granted rights over spec-
trum as a way to protect their own interests. They resist, that is, both
government-granted and market-regulated licenses. Thus, when the gov-
ernment proposed auctioning off more of the radio spectrum, conservative
economist George Gilder responded not by praising markets, but by attack-
ing the political corruption implicit in these deals. Says Gilder:

Still more subversive of good policy, the very auction process entrenches
obsolescent technology and promotes the false idea that spectrum is the
basis of a natural monopoly.37
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 84

8 4 L a w r e n c e L e s s i g

Gilder favors innovation and change over state-supported monopolies.
His aim is to push policies that would open up the resources of spectrum to
the widest range of innovators. A spectrum commons would do just this. Just
as the Internet did, it would open up a resource for the common use of a
wide range of innovators. These many innovators would experiment with
ways of using the network that none of us could now imagine. They would
fuel a second and possibly far more important wave of innovation than the
initial wave of the Internet that we have seen so far.

liberating spectrum from the control of government is an important
first step to innovation in spectrum use. On this point there is broad agree-
ment, from those who push for a spectrum commons to those, like Hazlett,
who push for a fully propertized spectrum market. All agree that the only
thing that government-controlled spectrum has produced is an easy oppor-
tunity for the old to protect themselves against the new. Innovation moves
too slowly when it must constantly ask permission from politically con-
trolled agencies. The solution is to eliminate the need to ask permission, by
removing these controllers at least.
Liberating spectrum from the control of the market is a second and much
more controversial step. Hazlett and others insist that the rationing of a mar-
ket is necessary, both to avoid overuse and to provide a sufficient incentive
to improve spectrum efficiency. A spectrum commons will invite tragedy
too quickly.
For the moment, we can defer resolving the differences between these
two positions, to emphasize their common view: Both want a world where
the power of controllers to stifle innovation has been eliminated. Both
agree that government control over spectrum is simply a way for the old to
protect themselves against the new. Both therefore push for a radical change
in spectrum management policies, to free innovators from the need to
please politicians before they have the right to innovate.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 85

C o m m o n s L e s s o n s

C ommons may be rare. They may evoke tragedies. They may be hard to
sustain. And at times, they certainly may interfere with the efficient use
of important resources.
But commons also produce something of value. They are a resource for
decentralized innovation. They create the opportunity for individuals to
draw upon resources without connections, permission, or access granted by
others. They are environments that commit themselves to being open. Indi-
viduals and corporations draw upon the value created by this openness.
They transform that value into other value, which they then consume pri-
vately.
The Internet has been built on two kinds of commons; it has the poten-
tial to move to a third. The protocols of the Net embedded principles in the
Net that constructed an innovation commons at the code layer. Though
running on other people’s property, this commons invited anyone to inno-
vate and provide content for this space. It was a common market of innova-
tion, protected by an architecture that forbade discrimination.
Free or open source software provided a second commons at the content
layer. The open code components of the Net were perpetual options for in-
novation. The ideas and implementation of code that would build the Inter-
net were made freely available both technically and legally. Legally, to the
extent that licenses that protected open code required that it remain in the
commons. Technically, in the sense that the code that built core and
peripheral functions—including, importantly, the World Wide Web—was
made available to all.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 86

8 6 L a w r e n c e L e s s i g

Finally, free spectrum was the promise to produce a new commons at the
physical layer. Here again, access would be uncontrolled and the use of this
access would be determined by a wide range of innovators. Not solely by the
handful of innovators owning these essential facilities, but by a wide range
of innovators who might have a different view of how the facilities might be
used.
These three commons work together. They increase the value of con-
trolled resources by connecting them with free resources. The strands of
fiber being laid across the world are all controlled by individuals and corpo-
rations. They are, for the most part, private. But the value they have is a
function of the use to which they will be put. And that use is this commons
called the Internet. The commons contributes to its value, and it makes the
control that contributes to it possible.

no doub t my account is incomplete. I have spoken of how these com-
mons induce innovation; I have not pretended to measure how much or
how significantly. I have not surveyed the full range of factors that might be
said to affect innovation. My focus has been narrow and selective.
My excuse, however, is that the debate right now is not about the degree
to which free or common resources help. The attitude of the most influen-
tial in public policy is that the free, or common, resources provide little or
no benefit. There is for us a cultural blindness—an unwillingness to even
account for the role of the commons. As Yale law professor Carol Rose ar-
gues, and as I indicated at the start, though “our legal doctrine has strongly
suggested that some kinds of property should not be held exclusively in pri-
vate hands, but should be open to the public,”1 we live in a time when the
dominant view is that “the whole world is best managed when divided
among private owners.”2 The very idea that nonexclusive rights might be
more efficient than exclusive rights rarely enters the debate. The assump-
tion is control, and public policy is dedicated to maximizing control.
But there is another view: not that property is evil, or that markets are cor-
rupt, or that the government is the best regime for allocating resources, but
that free resources, or resources held in common, sometimes create more
wealth and opportunity for society than those same resources held privately.
Against the background of the commons we’ve seen in the context of the
Internet, my aim in this chapter is to explore some reasons why. What do we
gain by keeping resources free? What is lost when we allow certain re-
sources to be controlled?
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 87

T H E F U T U R E O F I D E A S 8 7

In this chapter, I draw together a few clues to answer this question. Draw-
ing upon a wide range of writing, I want to pull together different accounts
that suggest the value in keeping resources free. My aim is not proof; it is in-
stead simply to connect ideas that are often left apart.
We can begin with our own legal tradition and with the resources that our
tradition has left in the commons and the reasons why. Professor Rose has
identified two reasons why our tradition has kept a particular resource—
such as a public road, a right-of-way, a navigable waterway, or a town
square—in common. First, these resources are “physically capable of mo-
nopolization by private persons.”3 Monopoly means power, and the mo-
nopolist would be capable of exerting power over the community. Second,
the public has a superior claim to these resources because “the properties
themselves [are] most valuable when used by indefinite and unlimited
numbers of persons.”
The easiest example here is the case of a road. A road is kept in the com-
mons because the opportunity for “holdouts” would be too great if the road
were private. If a road became the common path along which all commerce
passed, if along that path other businesses were built and other services were
provided, then there would be a great value secured by this common road.
And selling that road might then risk a hijacking by the owner of the road.
The public gets great value out of the road, and the road has value because
of the “publicness” of the road. The risk this value creates is that a private
actor might take advantage. The property is thus “affected with a public
interest” in the sense that the road’s value comes from the public’s depen-
dence on it.4
Likewise with a town square. No doubt in any town there are many dif-
ferent places that might be a town square. But over time, one place is the
town square, and it may well become valuable just because it is associated
with custom and history within a given community. Keeping this resource
in the hands of a community is a way to assure that no single actor takes ad-
vantage of the value the community has created. The value of this particu-
lar square comes not from the actions of its owner, but from a tradition that
invests it with significance above others.5
In both these cases, the resource is kept in the commons because of the
risk of an unfair capture if the resource were private. But why “unfair”? Why
isn’t it completely fair that the “owner” of the property be able to extract all
of its value?
Here is the great insight in Carol Rose’s analysis. Where the resource has
a value because of its openness—where its value increases just because more
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 88

8 8 L a w r e n c e L e s s i g

use it; where “the more the merrier”—then it makes sense to attribute much
of the value of this resource to the “publicness” of the resource. Indeed, as
Rose argues, “the usual rationing function of pricing could be counter-
productive [in these cases]: participants need encouragement to join these
activities, where their participation produces beneficial ‘externalities’ for
other participants.”6
These are cases where “increasing participation enhances the value of the
activity rather than diminishing it.” Or, we might say more precisely, these
are cases where the value from increased participation outweighs any cost
from increased utilization. The value, in these cases, comes from the con-
vergence of many upon a common use, or standard, or practice. And in
these cases, keeping the resource in the commons is a way to assure that that
value is preserved for all.7
These arguments from tradition are thus grounded in both fairness and
efficiency, and economists have extended the arguments from efficiency.8
One extension in particular links back directly to the end-to-end argument.
The linking goes like this: Some resources have an understood purpose.
We know what we will do with a certain resource, or at least the range of
possible uses for that resource is small. But other resources don’t come with
their purpose preset.
Take telephone wires in the 1910s. Communications wires had been
strung in America since the early 1800s. When they were first strung, their
use was simple: telegraph. Given the technology at the time, there was little
more that the wire could be used for; it was single-purpose. When tele-
phones came along, there was a second possible use for the wire. That led
to a new shake-up in business models. But here again, given the technology,
the range of possible uses for these wires was not great.
Contrast this with computer networks. The most striking feature of the
early history of the Internet is the repeated assertion by those at its founding
that they simply didn’t know what the network would be used for. Here they
were building this large-scale computer network, with a large number of re-
sources devoted to it, but none of them had a clear idea of the uses to which
this network would be put. Many in the 1980s believed the Internet would
be a fair substitute for telephones (they of course were wrong); none had any
idea of the potential for many-to-many publishing that the World Wide Web
would produce.9
Where we have little understanding about how a resource will be
used, we have more reason to keep that resource in the commons.10 And
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 89

T H E F U T U R E O F I D E A S 8 9

where we have a clear vision of how a resource will be used, we have more
reason to shift that resource to a system of control.
The reason is straightforward. Where a resource has a clear use, then,
from a social perspective, our objective is simply to assure that that resource
is available for this highest and best use. We can use property systems to
achieve this end. By assigning a strong property right to the owners of such
resources, we can then rely upon them to maximize their own return from
this resource by seeking out those who can best use the resource at issue.
But if there is no clear option for using the resource—if we can’t tell up
front how best to use it—then there is more reason to leave it in common,
so that many can experiment with different uses.11 Not knowing how a re-
source will be used is a good reason for making it widely available.12
Scott Bradner and Mark Gaynor have captured this insight in a paper that
uses “real options theory” to value different network designs. Their conclu-
sion is that where uncertainty is highest, network designs that embrace end-
to-end maximize the value of the network; and where uncertainty is low,
then end-to-end is not a particular value.13
In this case, end-to-end is a stand-in for a commons. Here too is a re-
source that can be used in any number of unpredictable ways. As David
Reed describes the founding of the network design, “[T]he idea was we
didn’t want to decide. . . . We felt that we couldn’t presume anything about
how networks would be used by applications.”14 And given the unpre-
dictable character of the ways it might be used, there is something gained by
keeping the resource open.
There is a second line of work that suggests another efficiency-based rea-
son why open resources may have more value than closed resources. This
work derives from the theory of management, and it helps explain why con-
trol can sometimes systematically fail.
The idea here has been made familiar by Professor Clay Christensen of
the Harvard Business School in his book The Innovator’s Dilemma.15 The
dilemma describes a perfectly understandable series of decisions that leads
well-managed companies to miss the opportunities of disruptive technologi-
cal change. Leading companies within a particular market will outperform
others in perfecting the technology that defines their existing market. They
will consistently develop superior products for continuing the development
of their product line.
What these companies can’t do is identify and develop disruptive tech-
nologies. (As David Isenberg puts it, “[T]he milk of disruptive innovation
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 90

9 0 L a w r e n c e L e s s i g

doesn’t flow from cash-cows.”)16 And this is not because a company is irra-
tional or because it doesn’t understand the nature of the market. The blind-
ness that keeps the company fixed in a dying path is actually its clear
understanding of probable returns. It sees real revenue from existing cus-
tomers who need marginally better technology. It doesn’t see the revenue
from radically new technologies that depend upon unidentified or undevel-
oped markets. From its perspective, given its customers and reasonable ex-
pectations, these successful companies rationally fail.
Christensen offers the disk drive industry as an example. Disk drives have
increased in capacity while falling in physical size at a dramatic rate.17 Over-
all, we can see that this shrinking created an extraordinary new market for
computing power. In hindsight, it is clear that victory would go to the com-
pany that developed the smallest, most powerful drive.
But at each stage of that development, this obvious truth was missed by
the very best disk drive manufacturers. The progress that led to the market
we see now was not continuous; it was punctuated by disruptive changes in
disk drive size. At each of these moments of disruption, the change occurred
not because some genius had discovered a new technology that permitted
the drive to shrink in size. The technology of each smaller drive was famil-
iar and available to all. Instead, the disruptive changes occurred when an
outside firm saw a new market and was willing to bet the firm on the success
of this market. This new market was always more competitive than the old.
The size of this market was uncertain. So from the perspective of the domi-
nant player, moving into this new market seemed like a bad move. Its cus-
tomers wanted nothing like the technology of the new drives; and it didn’t
have a vision that showed it the potential of a radically different market.
This blindness of successful companies comes not from management’s
failing. This pattern of failure can be seen in the very best firms. This is
not the market’s acting irrationally; it is the product of a rational strategy,
given the market as it appears at any one time.
As David Reed says about AT&T: It was not willing to bet on data given
that “the known applications couldn’t justify it and they weren’t willing to
bet on the unknown applications.”18
Others have described a similar blindness. Jim Carlton tells the story of
Apple Computer’s failing to see the potential of a market where its OS was
licensed to Microsoft. Apple looked at the margins it was getting from its
relatively small but rich market for PCs, and it compared those margins to
those of other computer manufacturers. Apple’s position looked far supe-
rior, so, rather than licensing the OS, Apple kept it closed.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 91

T H E F U T U R E O F I D E A S 9 1

Carlton describes this as pathology.19 It was the product, he suggests, of
committee decision making. And trading upon what happened since 1985,
the reader is left with the view that mismanagement has accounted for
Apple’s failure.
But the Christensen story suggests how it was Apple’s success that caused
Apple’s failure. Its inability to see was not a function of its blindness. Its in-
ability to recognize the value in a radically different model of doing business
may well have been a rational decision, given the information available.
What Christensen teaches is why, systematically, the view of what is rational
from the perspective of a single actor may well prove irrational from the per-
spective of the market as a whole.
The Innovator’s Dilemma offers its own strategy for dealing with this
blindness. But we can see in the Internet a strategy for dealing with the very
same blindness. If firms will be focused on continuing progress, if they will
ignore new markets that fail to promise the same level of supracompetitive
returns, if they will miss disruptive technologies that in fact produce radical
new industries, then we have another reason, in theory, to keep at least some
critical resources for innovation within a commons. If the platform remains
neutral, then the rational company may continue to eke out profit from the
path it has chosen, but the competitor will always have the opportunity to
use the platform to bet on a radically different business model.
This again is the core insight about the importance of end-to-end. It is
a reason why concentrating control will not produce disruptive techno-
logy. Not necessarily because of evil monopolies, or bad management,
but rather because good business is focused on improving its lot, and dis-
ruptive technologists haven’t a lot to improve. The disrupters are hungry to
build a different market; the incumbent is happy to keep the markets as they
are.
This last point suggests a third line of work suggesting an efficiency-based
reason for preferring open rather than controlled resources. If the Chris-
tensen story is of the blundering giant, then this is the story of the malevo-
lent giant. Here the actor—a company or an individual holding some
monopoly privilege—fully understands how a new technology might in-
crease social value. But the giant also realizes that there is no way it can cap-
ture this increase in social value. Unable to capture the gain, and certain to
lose its own rents, the malevolent giant acts to resist the technological
change, as a way of preserving its own power.
Such cases are easy to describe in the abstract; proving they exist in reality
is much harder. Whatever its intent, the malevolent giant rarely has the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 92

9 2 L a w r e n c e L e s s i g

power to control a technology completely; and even where it does have the
power, other market forces may be adequate in checking the exercise of
the power.
We as a society should favor the disrupters. They will produce movement
toward a more efficient, prosperous economy. Christensen argues for man-
agement structures that would facilitate that; the Internet is an architectural
structure that does the same.
This link between innovation and architecture is the focus of the work of
two other Harvard Business School professors as well. Professors Carliss
Baldwin and Kim Clark have demonstrated the importance of modular de-
sign in facilitating design evolution and hence industry innovation. In the
first volume of an intended two-volume work, they demonstrate the funda-
mental shift in the design of the computer industry, as IBM increasingly
modularized the design of its systems, and as regulators increasingly forced
IBM to permit the modules to be provided by others. This change reduced
the market value of IBM, but that reduction was overwhelmed by the in-
crease in value in the rest of the industry. As they describe it, a “multiplica-
tion and decentralization of design options led to the emergence of a new
industry structure for the computer industry,” and this in turn radically in-
creased the value of the industry.20
Modularity liberates control resources, as the multiplication of interfaces
frees innovators to develop new and competing designs. It is another exam-
ple of how free resources enable innovation.
Efficiency is not the end of the reasons why free resources might prove
valuable. Instead, one final set of values also indicates the value in keeping
a resource in common. These are democratic values.
The democratic tradition is our strongest ground for resisting the system
of control. Why don’t we simply sell the right to govern to the highest bid-
der? (The cynical will say we already have in effect. Maybe, but I’m talking
formally.) Why don’t we have a system where we auction off the rights to
control the government as a permanent property right?
This is clearly not how we arrange governance today. The right to par-
ticipate in a democracy is kept in common. We don’t permit people to sell
their right to vote. We permit neither the government to control how that
resource is used nor the market to control how that resource is used. In-
stead, we keep that resource in common hands, whether perpetually (in
democracies that can be recalled at any moment) or periodically (in democ-
racies like that in the United States, where elections are held every few
years).
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 93

T H E F U T U R E O F I D E A S 9 3

Democracies thus forbid propertizing the right to control government.
Why? This is not a hard question to answer, though raising it as a question
will help us think through this problem elsewhere. We don’t sell the right to
vote because the currency—cash—is not the only or most important di-
mension of value in our society. There are people who devote themselves to
careers that don’t make them wealthy—schoolteachers and civil servants.
We don’t think they, by virtue of that choice, should have less power to con-
trol how their government is run. They’ve made choices that result in their
having less power in the marketplace; but the marketplace is not a proxy for
every domain of social power. As the philosopher Michael Walzer properly
observes, there are many spheres of social influence in our lives.21 And we
permit power in one sphere to dominate power in another in very few con-
texts. We don’t in the United States permit sex to be purchased; we don’t sell
wives for dowries; we don’t sell babies; and we don’t sell votes.22 These trans-
actions are blocked because allowing the market to control them would be
to allow one sphere total power over all others. This we have chosen not
to do.
A similar insight gives more reason for certain resources to remain in
common. Access to locations where protest happens—town halls, or town
squares, or, in the language of First Amendment law, public fora—remains
open to all if open to any, or remains open on equal terms. Here the market
is not permitted control.
And likewise, one might well argue, when the resource becomes founda-
tional to participation in a society, then we assure that it remains in the com-
mons. The right to vote is a foundational resource in our society; we don’t
allow it to be bought or sold. Access to the roads or highways is central to so-
cial freedom; we don’t auction off such access and thereby restrict the right
to travel. And some have argued that basic infrastructure—like phones or
emergency services—should be considered common resources that must be
made available to all. The specifics we can argue about, but the general
point should be clear: There are values that a commons could serve that are
lost if the resource is privatized.

in advocating the commons, I have not argued for a world with only a
commons. Not all resources can or should be organized in a commons. Not
all resources must be organized as a commons just because some are. There
are public streets as well as private drives, freeways as well as toll roads. The
Internet links seamlessly with networks that are completely private. A world
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 94

9 4 L a w r e n c e L e s s i g

with open wires radio spectrum is perfectly consistent with a world where
exclusive cable lines are reserved to those who pay. The open and the closed
always coexist and depend upon each other in this coexistence.
But there are reasons why some resources need to be controlled and oth-
ers do not. We’ve seen these reasons before, but we are in a better position
now to understand them. While some resources must be controlled, others
can be provided much more freely. The difference is in the nature of the re-
source, and therefore in the nature of how the resource is supplied.
This was the insight of many in the Enlightenment and, within our tra-
dition, Thomas Jefferson most forcefully. Listen to Jefferson writing to Isaac
McPherson in 1813 about the character of the patent power:

[1] If nature has made any one thing less susceptible than all others of ex-
clusive property, it is the action of the thinking power called an idea,
which an individual may exclusively possess as long as he keeps it to him-
self; but the moment it is divulged, it forces itself into the possession of
everyone, and the receiver cannot dispossess himself of it. [2] Its peculiar
character, too, is that no one possesses the less, because every other pos-
sesses the whole of it. He who receives an idea from me, receives instruc-
tion himself without lessening mine; as he who lites his taper at mine,
receives light without darkening me. [3] That ideas should freely spread
from one to another over the globe, for the moral and mutual instruction
of man, and improvement of his condition, seems to have been peculiarly
and benevolently designed by nature, when she made them, like fire, ex-
pansible over all space, without lessening their density at any point, and
like the air in which we breathe, move, and have our physical being, inca-
pable of confinement, or exclusive appropriation. [4] Inventions then can-
not, in nature, be a subject of property.23

I’ve added numbers in brackets to Jefferson’s text to make clear the dis-
tinct points he is making:
First, Jefferson is describing the nature of an “idea.” An idea is, in the
terms of the economist, imperfectly excludable. I can keep a secret from you
(and therefore exclude you from the secret), but once I tell you the secret, I
can’t take it back. We can’t (yet) erase what has entered our heads.
Second, he is describing the nonrivalrous character of resources like
ideas. Your consumption does not lessen mine, as your lighting a candle at
mine does not darken me.
These two points then suggest a third: that “nature” has made this world
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 95

T H E F U T U R E O F I D E A S 9 5

to guarantee that “ideas should freely spread from one to another over the
globe.” Enlightenment was in her plan.
Thus it follows that without government, in the state of nature, there
would be no such thing as a “patent” since patents are granted for “inven-
tions” and inventions, “in nature,” cannot be “a subject of property.”
What is striking about this passage is the glee with which Jefferson reports
this fact of nature. Here is the first patent commissioner showing just why
nature is against the work of the U.S. Patent Office. But the motive of his
glee is the betterment of man. This fact about nature means that of all the
resources, information can be the freest.
Yet obviously, Jefferson’s story is not true of all resources, or even all
resources in the commons. His is an account of a nonrivalrous resource. A
rivalrous resource would not permit your consumption without lessening
mine. And his argument cannot be taken to mean that there should be no
control that governs nonrivalrous resources. Nature may not protect them,
but neither does nature erect governments. Jefferson was not arguing
against patent protection; he was instead arguing against the idea that patent
protection was in some sense a natural right.
This distinction between resources helps us isolate the different reasons
why a resource might need to be controlled.

1. If the resource is rivalrous, then a system of control is needed to assure
that the resource is not depleted—which means the system must assure
the resource is both produced and not overused.

2. If the resource is nonrivalrous, then a system of control is needed sim-
ply to assure the resource is created—a provisioning problem, as Profes-
sor Elinor Ostrom describes it. Once it is created, there is no danger
that the resource will be depleted. By definition, a nonrivalrous re-
source cannot be used up.

What follows then is critical: The system of control that we erect for ri-
valrous resources (land, cars, computers) is not necessarily appropriate for
nonrivalrous resources (ideas, music, expression). Indeed, the same system
for both kinds of resources may do real harm. Thus a legal system, or a soci-
ety generally, must be careful to tailor the kind of control to the kind of re-
source. One size won’t fit all.
A second point also follows and is equally important: Even for resources
that are nonrivalrous, some form of control will often be required. For these
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 96

9 6 L a w r e n c e L e s s i g

resources, there is still the need to assure an adequate incentive to supply or
to provision the resource. Thus, even here, some control will often be
needed.
In both cases, the necessary control could be provided through a number
of techniques—through law, norms, the market, or, importantly, technology.
Laws against theft can protect the property interest of rivalrous resources;
norms against overuse can protect some shared resources; prices imposed by
the market can induce provisioning and reduce consumption; and tech-
nology can make it easier to control.
This range of techniques means that there are many different ways to pro-
vide the degree of control that any particular resource might need. The
commons that Carol Rose describes are governed not by the market or by
law imposed by state actors. They are instead governed by “custom” or
norms within the relevant community.

Custom thus suggests a means by which a “commons” may be managed—
a means different from exclusive ownership by either individuals or gov-
ernments. The intriguing aspect of customary rights is that they vest
property rights in groups that are indefinite and informal, yet nevertheless
capable of self-management. Custom might be the medium through
which such an informal group acts generally; thus, the community claim-
ing customary rights was really not an “unorganized” public at all.24

Commons in the Internet are regulated differently. “Custom” is not the
typical controller anymore. It was at a certain time—USENET, for exam-
ple, which facilitated a worldwide messaging board organized into separate
topics, was governed by a custom that forbade commercial advertising;
when that custom died, much of the value of USENET died. But in the
contexts we have considered, custom is not the ruler. Controls imposed
through technology instead govern many of these resources.
The cases we’ve seen so far are a mix of rivalrous and nonrivalrous re-
sources, and the techniques of control within each are mixed as well. The
wires that supported the network that was the original Internet are clearly ri-
valrous; so too may be the radio spectrum that Hendricks and Hughes want
to share (though maybe not).25 But digital copies of operating systems are
not rivalrous. One copy is as good as the next, and once we have a single
copy, there is no limit to the copies we might make. Likewise with music, or
video that is made available in digital form.
With the rivalrous resources we’ve seen, technology guards against deple-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 97

T H E F U T U R E O F I D E A S 9 7

tion. Protocols for sharing the resource assure that many can use it without
anyone depleting the rest for others.
With the nonrivalrous resources, technology can’t itself solve the prob-
lem of incentives. Here one kind of law (contract law, through self-imposed
licenses) serves to solve some of the provisioning problems, at least
with open source or free software. This is the function of the GPL and other
open source licenses: relying upon the particular character of code, they
create a strong incentive for coders to contribute back to the commons.
But where these incentives are not enough, the law (through the odd de-
vice of what we have come to call “intellectual property”)26 adds more. In-
tellectual property does this by giving the producers a limited exclusive right
over their intellectual property, so they can recover the costs of producing
that property and receive a sufficient return to give them the incentives to
produce that property. A “sufficient return,” however, is not perfect control,
and intellectual property law does not, therefore, give authors or inventors
perfect control. The basic premise, found in our Constitution, is that “nei-
ther the creator of a new work of authorship nor the general public ought to
be able to appropriate all the benefits that flow from the creation of a new,
original work.”27 Instead, some of that benefit ought to be reserved to the
public, in common.

society benefits from resources that are free; but unless some system
of control is implemented for resources that must be created, or for re-
sources, once created, whose use is rivalrous, then no benefit will be re-
ceived. The key is to balance the free against control, so that the benefits of
each can be achieved.
Yet this balance is not automatic. There is no guarantee that the control
will be enough and no promise that it won’t be too much or too little. The
aim of society must always be to draw the optimal balance, and our obliga-
tion over time is to assure that that drawing not become skewed. The
level of control at one time might be insufficient at a different time. And
the level of freedom assured at one time might become threatened as the
technologies of control change.

this point should be obvious, but let’s make sure.
Let’s imagine a fishing village that for generations has managed to fish in
equilibrium with the stocks. The fish, in this example, are held in a com-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 98

9 8 L a w r e n c e L e s s i g

mons; the community doesn’t allocate right of control. Fishermen have an
understanding about when a catch is too much; they have boats designed
with this understanding in mind.
Along comes a new technology for fishing, which if used by each fisher-
man would radically deplete the existing stocks. Now the community faces
a decision—how best to regulate the use of this technology to assure the re-
source is not depleted. If the community does nothing, the norms of the
community might still be sufficient to keep the catch in line. But if the
norms are not enough, then the community must deploy a new technology
of control.
This “new technology,” however, is not determined. The solutions could
be many. The community might issue a regulation that says how much each
fisherman can catch; it might create a property right in the resource and
allow individual fishermen to trade it. Or it might deploy some technology
that would limit the catch of each fisherman over a given period of time. All
of these are possible responses to the threat posed to the common resource
by the new technology. Each responds to a change that undermines the old
equilibrium.
The same story can happen the other way around. Consider the problem
of copyright on the Internet. As I’ve already explained, the aim of copyright
is to give an author an exclusive right sufficient to create an incentive to pro-
duce, but not so great a right as to undermine the public domain. The Con-
stitution limits this exclusive right—Congress may not, for example, give
copyright to ideas, nor may it deny a right to fair use. These limits are in ad-
dition to the express constitutional limits imposed by the clause granting
Congress the power to create these “exclusive rights”—namely, that the
rights be for a “limited term” and that they “promote the progress of Sci-
ence.”
When the Internet first became popular, there was great fear that the
technology for digital copying would render useless the rights granted by
law. If I could make perfect copies for free and distribute them for free, then
the legal restriction would become much less useful.
This led Congress to expand the rights protected by the Copyright Act, to
balance the change in technology that the Internet produced. But as many
have argued, this change may have been premature. For there are tech-
nologies that can be deployed to protect copyrighted work. And if deployed
successfully, these technologies may actually give copyright holders more
control than they would have had absent the Internet, thereby lessening the
need for law.28
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 99

T H E F U T U R E O F I D E A S 9 9

Here the technology is expanding control beyond the balance originally
set, where, as in the previous example, technology was expanding free use
beyond the balance originally set. In both cases, the point is the same: the
balance must reflect the technologies as they exist. And changes in tech-
nologies can significantly change this balance.
The point is more than theoretical. In essence, the changes in the en-
vironment of the Internet that we are observing now alter the balance
between control and freedom on the Net. The tilt of these changes is pro-
nounced: control is increasing. And while one cannot say in the abstract
that increased control is a mistake, it is clear that we are expanding this con-
trol with no sense of what is lost. The shift is not occurring with the idea of
a balance in mind. Instead, the shift proceeds as if control were the only
value.
The aim in the balance of this book is to make this transformation plain.
We are remaking cyberspace, and these remakings will undermine the in-
novation we have seen so far.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 100
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 101

P A R T

I I

///

DOT.COntrast
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 102
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 103

C r e a t i v i t y i n R e a l S p a c e

T here was a time before the Internet. Innovation and creativity were dif-
ferent then. I don’t mean that creators were different then or that the
process of creativity has changed. But the constraints on creativity and in-
novation were different. This difference can be expressed at each layer of
Yochai Benkler’s system. Because the physical, and code, and content layers
were controlled differently, the opportunities for innovation were different.
We all know about these differences in the constraints among these lay-
ers. They are all obvious, if a bit in the background. They flow directly from
the nature of real constraints within a scarcity-based economy. They are not
the product of conspiracy or the will of evil minds. They are importantly un-
avoidable, at least in real space.
My aim in this chapter is to remind you of these things that we all know.
I will rehearse the constraints on innovation that flow from the character of
these different layers of communication in real space, so that we can better
see how they have changed.
In real space. It is this qualification about which we must become self-
conscious. Our intuitions about property, and about how best to order soci-
ety, are intuitions built in a particular physical world. We have learned a
great deal about how best to order that world, given the physics, as it were,
of that particular world.
But the physics of cyberspace is different. The character of the constraints
is different. So while there may be good reason to carry structures that de-
fine real space into cyberspace, we should not assume that those structures
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 104

1 0 4 L a w r e n c e L e s s i g

will automatically map. The different physics of cyberspace means that the
rules that govern that space may be different as well.1
A different physics. I’m not talking about science fiction or about ideas
that you’ve never considered before. Indeed, we’ve already seen a careful
translation of real-space constraints into the physics of a very different
world—the world of ideas. Jefferson made that translation in his writing
about the nature of patent. My argument is nothing more (and certainly
much less) than Jefferson’s. The world we must consider is partway between
the world of ideas that he describes and the world of things that colors our
intuitions. Cyberspace is between these two worlds. It offers not quite the
freedom of the world of ideas, though it offers much more of that freedom
than the world of things.
In the balance of this chapter, I want to make explicit constraints in the
world of things, so that we can better see how these constraints have affected
our thought about the world of ideas, and hence also about cyberspace.
One final note. My argument is not that all constraints are corrupting of
something called “creativity.” Certain constraints obviously enable cre-
ativity. The constraints of the classical form gave us Mozart and Beethoven.
The aim is therefore not to find a world without constraint; it is to remove
the constraints that might otherwise inhibit innovation. Just because it is
good that sonnets forbid rambling paragraphs, it doesn’t follow that a tax on
books would inspire better writing.

C R E AT I V I T Y I N T H E D A R K AG E S

put yourself back in the dark ages, the time before the Internet took
off—say, the 1970s—and ask: What was the environment for creativity
then? What was required of a creator or innovator to bring his or her cre-
ativity to market? What limits were imposed? I want to consider this ques-
tion in two contexts—first the arts and then commerce.

The Arts

we can understand the environment for creativity in the arts with the
same three layers that Benkler describes when talking of a communications
system. Like a communications system, creativity in the arts is affected by
constraints at the physical, code, and content layers. To author, or to create,
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 105

T H E F U T U R E O F I D E A S 1 0 5

requires some amount of content to begin with, to which the author adds a
creative component, which, for a few, is then published and distributed.

content

the content an author must draw upon varies with the “writing.” Some
part is new—this is the part we think of as “creative.” But as many have ar-
gued, we’ve come to exaggerate the new and forget that a great deal in the
“creative” is actually old.2 The new builds on the old, and hence depends,
to a degree, on access to the old. Academics writing textbooks about poetry
need to be able to criticize and hence, to some degree, use the poetry they
write about. Playwrights often base their plays upon novels by others. Novel-
ists use familiar plots to tell their story. Historians use facts about the history
they retell. Filmmakers retell stories from our culture. Musicians write
within a genre that itself determines how much of the past content it needs
to be within that genre. (There is no such thing as jazz that does not take
from the past.) All of this creativity depends in part on access to, and use of,
the already created.
In our present legal regime, some of this content is free; some is con-
trolled. A poet has a copyright on his or her poetry. Others cannot simply
take and reproduce it without the copyright holder’s permission. The same
with plays and novels: A play that is close enough to the plot of a novel is
a derivative work. Copyright law gives the copyright holder control over
these derivative works. Musical chords cannot be controlled; the design of
public buildings cannot be copyrighted. These bits of content in these tra-
ditions are free, even if the control created by copyright is strong.
But this control is still limited—indeed, it is constitutionally limited.
While a poet or author has the right to control copies of his or her work, that
right is limited by the rights of “fair use.” Regardless of the will of the own-
ers of a copyright, others have a defense against copyright infringement if
their use of the copyrighted work is within the bounds of “fair use.” Quoting
a bit of a poem to demonstrate how it scans, or making a copy of a chapter
of a novel for one’s own critical use—these are paradigmatic examples of
use that is “fair” even if the copyright owner forbids it.
A similar limitation protects the historian. For content to be controlled, it
must be “creative.” Facts on their own are not “creative.” As the Supreme
Court has said, “[T]he sine qua non of copyright is originality. To qualify for
copyright protection, a work must be original to the author. . . . [But] facts
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 106

1 0 6 L a w r e n c e L e s s i g

do not owe their origin to an act of authorship. The distinction is one be-
tween creation and discovery.”3 Thus, facts remain in the commons for any-
one to draw upon—even if these facts were discovered only because of the
hard work of some investigator. Hard work does not entitle someone to a
copyright. Only “creativity” does. Thus facts remain a resource that—
constitutionally—cannot be subject to a system of legal control.
So too with all creative works—eventually. Disney, for example, did not
license the right to make The Hunchback of Notre Dame or Pocahontas.
These works, though originally copyrighted, are no longer subject to copy-
right’s control. Copyright is, in the United States, at least, constitutionally
required to be for a “limited time.” After that limited time, the work falls
into the public domain—free of restraint, so that “second comers,” as Judge
Learned Hand described them, “might do a much better job than the orig-
inator” with the original idea.4
Or at least that’s the theory, though Congress has done its best in recent
years to ignore this theory. The distinctive feature of modern American
copyright law is its almost limitless bloating—its expansion both in scope
and in duration. The framers of the original Copyright Act would not begin
to recognize what the act has become.
Scope: The first Copyright Act gave authors of “maps, charts, and books”
an exclusive right to control the publishing and vending of these works, but
only if their works had been “published,” only after the works were regis-
tered with a copyright registry, and only if the authors were Americans. (Our
outrage at China notwithstanding, we should remember that before 1891,
the copyrights of foreigners were not protected in the United States. We
were born a pirate nation.)5
This initial protection did not restrict “derivative” works: one was free to
translate an original work into a foreign language,6 and one was free to make
a play out of a novel without the original author’s permission. And because
of the burdens of registering, most works were not copyrighted. Between
1790 and 1799, 13,000 titles were published in America, but only 556 copy-
right registrations were filed.7 The vast majority of creative work was free for
others to use; and the work that was protected was protected only for limited
purposes.
Time, with a little help from lobbyists, works changes. After two centuries
of copyright statutes, the scope of copyright has exploded, and the reach of
copyright is now universal. There is no registration requirement—every cre-
ative act reduced to a tangible medium is now subject to copyright protec-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 107

T H E F U T U R E O F I D E A S 1 0 7

tion. Your e-mail to your child or your child’s finger painting: both are au-
tomatically protected.
This protection is not just against competing publications. The target is
not simply piracy. Any act of “copying” is presumptively regulated by the
statute; any derivative use is within the reach of this regulation. We have
gone from a regime where a tiny part of creative content was controlled to a
regime where most of the most useful and valuable creative content is con-
trolled for every significant use.
Duration. The first Congress to grant copyright gave authors an initial
term of 14 years, which could be renewed for 14 years if the author was liv-
ing. The current term is the life of the author plus 70 years—which, for an
author like Irving Berlin, would mean a protection of 140 years. More dis-
turbingly, we have come to this expanded term through an increasingly
familiar practice in Congress of extending the term of copyright both
prospectively (to works not yet created) and retrospectively (to works created
and still under copyright).
These extensions are relatively new. In the first hundred years, Congress
retrospectively extended the term of copyright once. In the next fifty years,
it extended the term once again. But in the last forty years, Congress has ex-
tended the term of copyright retrospectively eleven times. Each time, it is
said, with only a bit of exaggeration, that Mickey Mouse is about to fall into
the public domain, the term of copyright for Mickey Mouse is extended.8
You might think that there is something a bit unfair about a regime where
Disney can make millions off stories that have fallen into the public do-
main, but no one else but Disney can make money off Disney’s work—
apparently forever. You’d be right about that, but we’ll consider the fairness
(and more important, the constitutionality) in greater detail later on. It is
enough for now simply to recognize that even if the scope of controlled con-
tent has grown, in principle there is to be a constitutional limitation on this
expansion. Some content is to remain in the commons, even if most useful
content remains subject to control.

c ontr o l , as I have argued, is not necessarily bad. Copyright is a critical
part of the process of creativity; a great deal of creativity would not exist
without the protections of the law. Without the law, the incentives to pro-
duce creative work would be vastly reduced. Large-budget films could not
be produced; many books would not get written.9 Copyright is therefore an
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 108

1 0 8 L a w r e n c e L e s s i g

integral and crucial part of the creative process. And as it has expanded, it
has expanded the opportunities for creativity.
But just because some control is good, it doesn’t follow that more is bet-
ter.10 As Judge Richard A. Posner has written, “[T]he absence of copyright
protection is, paradoxical as this may seem, a benefit to authors as well as a
cost to them.”11 It is a benefit because, as we’ve seen already, creative works
are both an input and an output in the creative process; if you raise the cost
of the input, you get less of the output.
More important, limited protection has always been the rule. Never has
Congress embraced or the Supreme Court permitted a regime that guaran-
teed perfect control by copyright owners over the use of their copyrighted
material. As the Supreme Court has said, “[T]he Copyright Act does not
give a copyright holder control over all uses of his copyrighted work.”12
Instead, Congress has historically struck a balance between assuring that
copyright owners are compensated and assuring that an adequate range of
material remains in the public domain for others to draw upon and use. And
this is especially true when Congress has confronted new technologies.
Consider the example of piano rolls. In the early 1870s, Henri Fourneaux
invented the player piano, which recorded music on a punch tape as a
pianist played the music.13 The result was a high-quality copy (relative to the
poor quality of phonograph recordings at the time) of music, which could
then be copied and played any number of times on other machines. By
1902, there were “about seventy-five thousand player pianos in the United
States, and over one million piano rolls were sold.”14
Authors of sheet music complained, saying that their content had been
stolen. In terms that echo the cries of the recording industry today, copy-
right holders charged that these commercial entities were making money
off their content, in violation of the copyright law.
The Supreme Court disagreed. Though the content the piano player
played was taken from sheet music, it was not, the Court held, a “copy” of
the music that it, well, copied.15 Piano roll manufacturers (and record com-
panies, too) were therefore free to “steal” the content of the sheet music to
make money with their new inventions.
Congress responded quickly to the Court’s decision by changing the law.
But the change was an interesting compromise. The new law did not give
copyright holders perfect control over their copyrighted material. In grant-
ing authors a “mechanical reproduction right,” Congress gave authors the
exclusive right to decide whether and on what terms a recording of their
music could be made. But once a recording had been made, others had the
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 109

T H E F U T U R E O F I D E A S 1 0 9

right (upon paying two cents per copy) to make subsequent recordings of
the same music—whether or not the original author granted permission.
This was a “compulsory licensing right,” which Congress granted copiers of
copyrighted music to assure that the original owners of the copyrighted
works would not acquire too much control over subsequent innovation with
that work.16
The effect of this compromise, though limiting the rights of original au-
thors, was to expand the creative opportunity of others. New performers had
the right to break into the market, by taking music made famous by others
and rerecording it, after the payment of a small compulsory fee. Again, the
amount of this fee was set by the statute, not by the market power of the au-
thor. It therefore was a far less powerful “exclusive right” than the exclusive
right granted to other authors.17
This balance is the rule, not the exception, when Congress has con-
fronted a new technology affecting creative rights. It did the same thing with
the first real “Napster” in our history—cable television. Cable TV was born
by stealing the content of others and reselling that content to consumers.
Suppliers of cable services would set up an antenna, capture the commer-
cial broadcasts made by television stations, and then resell those broadcasts
to their customers.
The copyright holders did not like this “theft.” Twice they asked the
Supreme Court to shut cable TV down. Twice the Court said no.18 So it fell
to Congress to strike a balance between cable TV and copyright holders.
Congress in turn followed the model set by player pianos: cable TV had to
pay for the content it broadcast, but the content holders did not have an ab-
solute right to grant or deny the right to broadcast its content. Instead, cable
TV got a compulsory licensing system to guarantee that cable operators
would be able to get permission to broadcast content at a relatively modest
level. Thus content holders, or broadcasters, couldn’t leverage their power
in the television broadcasting market into power in the cable services mar-
ket. Innovation in the latter field was protected from power in the former.19
These are not the only examples of Congress striking a balance between
compensation and control. For a time there was a compulsory license for
jukeboxes; there is a compulsory license for music and certain pictorial
works in noncommercial television and radio broadcasts; there is a compul-
sory licensing scheme governing satellite television systems, digital audio
home recorders, and digital audio transmissions.20
These “compromises” give the copyright holder a guarantee of compen-
sation without giving the copyright holder perfect control over the use of its
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 110

1 1 0 L a w r e n c e L e s s i g

copyrighted material. In the language of modern law and economics, these
rules protect authors through a “liability rule” rather than a “property
rule.”21 They are perfect instances of the special character of copyright’s
protection, as they represent the aim to give authors not perfect control of
their copyrighted work, but a balanced right that does what the Constitution
requires—“promote progress.”
Thus, while Congress has expanded the scope of rights protected by the
Copyright Clause, as technologies have changed, it has balanced the rights
of access against these increases in protection. These balances, however, are
not, on balance, even: though limits have been drawn, the net effect is in-
creased control. The unavoidable conclusion about changes in the scope of
copyright’s protections is that the extent of “free content”—meaning con-
tent that is not controlled by an exclusive right—has never been as limited
as it is today. More content is controlled by law today than ever in our
past. In addition to limited compulsory rights, an author is free to take from
work published before 1923; is free to take noncreative work (facts) when-
ever published; and is free to use, consistent with fair use, a limited degree
of others’ work. Beyond that, however, the content of our culture is con-
trolled by an ever-expanding scope of copyright.

physical

at the content layer, I’ve argued, the law aims to strike a balance between
access and control. Copyrights grant control, but copyrights are constitu-
tionally and statutorily limited to ensure some uncontrolled access. Some
parts are controlled; some parts remain free.
No such balance exists at the physical layer, and for the most part, that’s
a good thing, too. Writing is produced and published on paper; paper is a
physical good; in our economy, physical goods are fully controlled by the
market. Films require film stock; nondigital film stock is extremely expen-
sive; no right to steal this physical stock exists in our society. Market control
is the rule at the physical layer; access is at the pleasure of the property
owner.
This control is largely benign, at least where markets are competitive.
If the market is not competitive, then power at the physical layer can be-
come harmful. Control at the physical layer can, in at least some contexts,
be leveraged into another layer.22 But for this danger, antitrust law is an ade-
quate remedy. As long as the other layers remain relatively free, the control
here is not inherently troubling.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 111

T H E F U T U R E O F I D E A S 1 1 1

The problem, of course, is that these other layers are not relatively free—
or at least they weren’t free in the dark ages. They were increasingly not free
for content; they were especially not free at the layer of code.

code

the core constraint on artistic creativity in real space is at the code
layer—the constraint on whose work gets produced and distributed where.
The writer becomes an author when his or her work is published. Publi-
cation is a process controlled by editors. Editors at The New York Times de-
cide what goes on their pages. Editors at Basic Books decide which books
they will print. No one has a right to enter Basic Books and steal access to its
printing presses. Nor does anyone have a right to demand that Basic Books
transport his texts. The production and distribution of printed material are a
wholly privatized activity.
The same is true for music. Rock bands are plenty; many write their own
content; most of that content (fortunately, perhaps) never gets heard beyond
a neighborhood garage. Whether the work of a musician gets distributed
broadly depends upon the decisions of publishers. Record companies
choose what gets floated in the market; radio stations (in effect) get paid to
play what record companies choose.23
So too with television. You are free to buy commercial time on television,
and in some markets you are free to buy program time. But unless you’re
Ross Perot, these freedoms don’t matter much. What gets played on TV is
the decision of network owners; what gets broadcast on cable is the choice
of cable companies.24
These constraints at the code layer plainly affect the choice of creators to
create or not. If the editors of a newspaper are conservative, a liberal colum-
nist is less likely to submit a column to that paper. If newspapers generally
are unwilling to be critical of U.S. policy, then authors who would criticize
U.S. policy are less likely to waste their time penning the criticism. Com-
munists don’t waste much time writing Marxist screenplays. Only the deeply
ill informed waste their time translating Adam Smith’s work to the silver
screen. The author is constrained by the expectation of how the code layer
will respond. And the code layer, in those dark ages, at least, was impor-
tantly controlled. Though the range of outlets expanded dramatically,25 the
concentration in ownership among those outlets increased as well. And the
Net is an important constraint on what is made.
Obviously, the code layer interacts with the physical and the content
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 112

1 1 2 L a w r e n c e L e s s i g

layers. NBC gets to decide what it will broadcast. Because of trespass laws, I
can’t break into NBC and interrupt the evening news. If I do, I will be ar-
rested for trespass. There is no First Amendment right that I can assert to
trespass on NBC’s property.
Likewise, NBC’s right at the code layer is largely protected against state
control by the First Amendment. Congress probably does not have the
power to pass a law requiring that NBC give me access to its station. Edito-
rial judgments of television executives are a constitutionally protected right
at the code layer.

Commerce

issues of control matter not just to artists, and the dark ages did more
than constrain budding Frank Sinatras. Indeed, among the most significant
aspects of the Internet revolution has been the liberation it has given to
commerce—not just to commerce in the mode of IBM or GM, but to com-
merce of the different. The commons of the Net exploded opportunities for
commerce that would not otherwise have existed. And this explosion was
not, given the architecture of telecommunications before the Net, predicted.
We can see this point quite quickly in two contexts that have been dra-
matically affected by the Internet—one in the context of coding, the other
in the expansion of the market. Both of these contexts were quite different
before the Internet, again because of the constraints imposed upon them by
the architectures of real space. The opportunities of both have been
changed as the technology of the Internet has changed.

coding

in 1 9 7 2 , Robert Fano, then a researcher at MIT, published a dark and
pressing essay titled “On the Social Role of Computer Communications.”26
Fano’s fear was that access to computing resources would be increasingly
centralized, and that this centralization would do a great damage to de-
mocracy. As the power to understand and manipulate data about the world
was held by a smaller and smaller number of people, the skew to democracy
caused by this concentration would only increase: what was needed, Fano
argued, was a different architecture for computer communications, one not
centralized within a small number of organizations, but instead made avail-
able generally to many.27
Fano had an idea of how to build this different architecture and what this
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 113

T H E F U T U R E O F I D E A S 1 1 3

different architecture would look like. To build it would require state inter-
vention to break up the concentrations in computer communications that
had emerged. The network thus built would look much like the Internet.
Fano was wrong (though understandably so) about the future. But he
wasn’t wrong about the past. For computers at the time were expensive de-
vices. Except for universities, programming for them required that you work
within an institution that could afford to own one of these devices. If you
wanted to work on a large-scale coding project, you needed to be within a
company that was producing large-scale code.
For many people, of course, that wasn’t a terrible thing. IBM and AT&T
were powerful and well-paying companies. Most would consider it a great
privilege to work for either.
But if you were not the sort likely to be able to work in these places—if
you lived in South Dakota, where there weren’t many IBM coding plants,
or in China, where not many coding companies were allowed—then this
reality was an important constraint. To author code in this world required
working within large, typically American, corporations. And for many, this
meant they could not author code at all. Just as with research in nuclear sci-
ence today, the ability to do this research was limited to those who worked
for specific organizations.
Again, this barrier is easy to understand. No conspiracy is needed to ex-
plain it. Computers were valuable resources; not every Joe could or should
have access to play around with them. The economic and processing con-
straints mean that the system couldn’t well leave itself open for others to
take. The restrictions here were unfortunate and unintended consequences
of economic constraints imposed elsewhere.
Here again we can understand these constraints in terms of Benkler’s
model. The physical layer of the “computer-communications” architecture
was controlled; the very nature of its expense forced users to locate to the
machines. Locating the machines in particular places made it easy to con-
trol access. The logic of the machine may have been open, but only those
with permission were allowed in the “machine room.” And finally, while
the source code for these machines may not have been controlled (content
layer, open), the small number of these machines meant that the value of
the open code was limited. Coding, and the creativity realized in coding,
was dictated by this architecture that mandated control.
This feature of the dark ages, then, limited the supply of resources to a
market of production. Only those in a particular place, only those willing to
work within a given structure, could work within coding projects. A wide
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 114

1 1 4 L a w r e n c e L e s s i g

range of talent was thereby excluded from the practice of coding. The ease
with which those resources might be shared with many outside a single or-
ganization was limited by the technologies of computer communications
that Fano described.

markets

so too are markets constrained. Technology most dramatically affects the
extent of the market. The more interconnected markets are, the easier it is
for goods from one area to affect the price of goods in another area. Geog-
raphy is a physical constraint on that interconnection—in real space,
greater distance means greater cost. But information supported by broad dis-
tributional channels can balance the constraint of geography.
Competition laws and constitutional norms keep this transportation sys-
tem competitive. Competition laws make it hard for distributors to restrict
or control distribution. The Dormant Commerce Clause of the U.S. Con-
stitution makes it hard for states to bias distribution to favor themselves.
These legal constraints balance natural tendencies among commercial and
political actors. They produce a relatively competitive interstate market for
goods and services.
Still, real space constrains. Even if the market were perfectly competitive,
the cost of transportation and the high cost of information restrict the mar-
ket’s scope. If you want to sell very weird widgets, and only a hundred thou-
sand people are within range, then you’re not likely to be able to sell enough
widgets to make it worthwhile. But if you had the world as your market—
if the code layer facilitated broad distribution of selective information about
widgets, thus lowering the cost of information—then you might have a mar-
ket large enough to make your weird widget factory work. As Ronald Coase
puts it:

People talk about increases in improvements in technology, but just as im-
portant are improvements in the way in which people make contracts and
deals. If you can lower the costs there, you can have more specialization
and greater production. . . . By improving the way the market works, you
can produce immense benefits, not because it invents new technologies,
but because it enables new technologies to be used.28

The net of these layers of control in real space is relatively simple to map.
Creativity may well be inspired by the protection these systems of control es-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 115

T H E F U T U R E O F I D E A S 1 1 5

tablish. But it is also constrained by the limits that these systems of control
impose. I can write what I will, but what gets published is a function of what
publishers like. I can sing in the shower, but before we sing “Happy Birth-
day” in a large crowd, we had better call a lawyer.29 My home movies can be
shown in my living room, but art students should not expect their films to
be shown in theaters. And freedom of speech notwithstanding, no one has
the right to fifteen minutes of NBC’s airtime. Creativity in the dark ages
lives in a world largely without a commons. Permission of others is the nec-
essary condition of one’s work being seen elsewhere.

now again, unlike in Lenin’s Russia, these systems of control are not the
product of conspiracy. The constraints that require control in these different
markets for resources are real. Economics is the science of choice in the
context of scarcity; it is a positive (if dismal) science that takes the world as
it finds it. We can no more will a world where real-space printing presses
were free than we can will a spacecraft that could fly as fast as the starship
Enterprise.
So by contrasting this economy governed by layers of control with an
economy governed by large swaths of the commons, I don’t mean to criti-
cize every system of control. Whether control is necessary for a particular
good in a particular context depends upon the context—upon the tech-
nologies of that context and the character of the resource. Resources held in
common in one context (among friends or in a small community) may need
to be controlled in another (in a city or between tribes).
In particular, to the extent a resource is physical—to the extent it is
rivalrous—then organizing that resource within a system of control makes
good sense. This is the nature of real-space economics; it explains our deep
intuition that shifting more to the market always makes sense. And follow-
ing this practice for real-space resources has produced the extraordinary
progress that modern economic society has realized.
A part, however, cannot speak for the whole, especially when changes in
technology render the assumptions of the old obsolete. Even if the control
model makes perfect sense in the world of things, the world of things is not
the digital world. We may need fences and perfect control to assure that the
world of things runs efficiently. That’s what the prosperity of the market,
property, and contract teach us.
But perfect control is not necessary in the world of ideas. Nor is it wise.
That’s the lesson our Framers taught us—in both the limits they placed on
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 116

1 1 6 L a w r e n c e L e s s i g

the Exclusive Rights Clause and the expanse of protection for free speech
they established in the First Amendment. The aim of an economy of ideas
is to create incentives to produce and then to move what has been produced
to an intellectual commons as soon as can be. The lack of rivalrousness
undercuts the justification for governmental regulation. The extreme pro-
tections of property are neither needed for ideas nor beneficial.
For here is the key: The digital world is closer to the world of ideas than to
the world of things. We, in cyberspace, that is, have built a world that is close
to the world of ideas that nature (in Jefferson’s words) created: stuff in cyber-
space can “freely spread from one to another over the globe, for the moral
and mutual instruction of man, and improvement of his condition,” be-
cause we have (at least originally) built cyberspace such that content is,
“like fire, expansible over all space, without lessening [its] density at any
point, and like the air in which we breathe, move, and have our physical
being, incapable of confinement, or exclusive appropriation.”
The digital world is closer to ideas than things, but still it is not quite
there. It is not quite true that the stuff in cyberspace is perfectly nonrivalrous
in the sense that ideas are. Capacity is a constraint; bandwidth is not un-
limited.30 But these are tiny flaws that cannot justify jumping from the
largely free to the perfectly controlled. There are problems of coordination
and constraints of scarcity. But the solution to these problems is not neces-
sarily systems of control or better techniques of excludability. That cyber-
space has flourished as it has largely because of the commons it has built
should lead us to ask whether we should tilt more to the free in organizing
this space than to the controlled that organizes real space.
Put differently: These imperfections in the capacity of cyberspace—that
together may make it more rivalrous than ideas are—should not by them-
selves force us to treat the resources that cyberspace produces as we would
treat real-space resources. If by resisting the model of perfect control we
gain something important, then we should do so.

in the context of the media, we can be a bit stronger than this. Over the
past twenty years, we have seen two changes in the media that seem to pull
in different directions. On the one hand, technology has exploded the num-
ber of media outlets—increasing the number of television and radio stations
as well as newspapers and magazines. On the other hand, concentration in
the ownership of these media outlets has also increased. This increase in
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 117

T H E F U T U R E O F I D E A S 1 1 7

concentration especially should lead us to ask whether the control enabled
in real space should carry over to cyberspace.
The statistics about increased concentration in ownership are undeni-
able and extraordinary. In 1947, 80 percent of daily newspapers were inde-
pendently owned; in 1989, only 20 percent were independently owned.
Most of the business of the nation’s eleven thousand magazines was con-
trolled by twenty companies in 1981; in 1988, that number had fallen to
three.31 Books are much the same. The independent publishing market was
strong just thirty years ago; with Bertelsmann’s purchase of Random House
in 1998, the industry is now much more concentrated, dominated by just
seven firms.32 The significance of this concentration in books is no doubt
less than that in film or other important media. There are still many inde-
pendent publishers, and the range and diversity of book publishing are quite
large. But the inertia is in the direction of concentration. And this inertia
may be a source of concern.
Music is even more concentrated.33 The five largest music groups in the
United States account for over 84 percent of the U.S. market.34 The same is
true of radio. The top three broadcasters control at least 60 percent of the
stations in the top one hundred U.S. markets.35 The same is true in film. In
1985, the twelve largest U.S. theater companies controlled 25 percent of the
screens; “by 1998, that figure was 61 percent and climbing rapidly.”36 Six
firms accounted for over 90 percent of theater revenues in 1997; 132 out of
148 of the “widely distributed” films in 1997 were produced by “companies
that had distribution deals with one of the six majors.”37 With this concen-
tration, there has been a dramatic drop in foreign films. In the mid-1970s,
foreign films accounted for 10 percent of box office receipts. By the late
1990s, the number had fallen to 0.5 percent. 38 Cable and television are no
better. In 1999, Robert McChesney could write that “six firms now possess
effective monopolistic control over more than 80 percent of the nation, and
seven firms control nearly 75 percent of cable channels and program-
ming.”39 Those numbers are now much more extreme.40 Professor Ben
Bagdikian summarizes the result as follows: “[D]espite more than 25,000
outlets in the United States, 23 corporations control most of the business in
daily newspapers, magazines, television, books, and motion pictures.”41 The
top firms in this set vastly outbalance the remainder. The top six have more
annual media revenue than the next twenty combined.42
The reasons for this increase in concentration are many. I don’t mean to
argue, as many others have, that we should necessarily consider this in-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 118

1 1 8 L a w r e n c e L e s s i g

creasing concentration inefficient or illegal. There are important efficien-
cies to be gained by the mergers of large media interests; important gains in
coverage have also been realized. And while the conspiracy theories are
many and practically unending in scope, we need not believe media con-
spirators are behind this radical change. The government has loosened its
restrictions on concentration, sometimes for good economic reasons; tech-
nologies of transmission have changed to the great benefit of all; and the
consequence has been an extraordinary concentration in media produc-
tion.43
But whatever the reason, the results are staggering. And they extend be-
yond the mere structure of the market. They affect its character as well. The
resulting mix of media is strikingly homogenous. The companies that make
up the handful of international conglomerates are cookie-cutter variations
of one another. Some are slightly larger in music than in film; others are
slightly more American in ownership and content. But if you had to char-
acterize the differences in philosophy or attitude among these different
media conglomerates, it would be extremely hard (unlike, for example, the
situation with newspapers in Britain): there are no clear philosophical or
ideological differences among them.44
Many have quite rightly worried that this control by a few who are not
very different from each other will have a significant effect on the kind of
news that is reported. Andrew Kreig tells a compelling story of the effect of
chain management on an American newspaper, driving the respected Hart-
ford Courant to more excessive, sensationalistic reporting.45 The paper he
describes is not dissimilar from many others. There are many stories about
corporate owners influencing the news within their organizations—steering
the news away from stories that reflect negatively upon those corporate own-
ers.46 Congressman Newt Gingrich expressly recommended as much in
1997, when he told the Georgia Chamber of Commerce that business lead-
ers and advertisers “ought to take more direct command of the news-
room.”47
Even if we ignore this most blatant form of bias, if the media are owned
by a handful of companies, each basically holding the very same ideals, how
much diversity can we expect in the production of media content? How
critical can we believe these media will be? How committed to testing the
status quo is this form of organization—itself so dependent upon the status
quo—likely to be?48
You don’t need to be a radical to be worried about this trend. Even the
most committed pro-market ideologues could at least hope for a broader
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 119

T H E F U T U R E O F I D E A S 1 1 9

range of competition in ideas and perspective. There is good evidence that
competition improves the quality of newspapers.49 And there is a general
and broad view that the only justification for the power that the media has
is that there is a broad range of views with the same power.50 No more.
Never in our history has the concentration of media outlets been greater.
Even a believer in the invisible hand might hope that this hand might muck
things up a bit.

something has mucked things up a bit. Something has entered the field
in a way that could make these concentrations change—not the govern-
ment or a regulation imposed by the government, but the architecture of
the Internet we have been describing so far.
For the essence of this power in the handful of media companies that now
dominate media internationally is control over distribution and the power it
can promise artists.51 Movies run in certain places only; getting films into
those places is quite hard. CDs are distributed through predictable channels
of distribution—including radio stations, whose choice of what to play or
not to play determines which content is popular or not. Breaking into this
distribution channel is likewise extremely hard.
The same is true with cable. While many thought that increasing the
number of cable channels would mean more valuable competition, in fact,
the fragmentation of channels simply induced more commercialization.
Fragmentation makes it easier to “slice and dice people demographically”
and “maximize . . . advertising revenues.”52 Cable has thus not been a
source of new innovation (unsurprisingly, as we saw, because the physical,
logical, and content layers are all controlled). Instead, as “one cable execu-
tive put it in 1998, ‘Most entrepreneurs have already gotten the word that
the cable field is closed.’ ”53
But the essence of the Internet that I’ve described so far is an architecture
for distribution that admits of no controllers, architecture that neither needs
nor permits the centralization of control that real-space structures demand.
And while this lack of control won’t on its own mean Hollywood will fail, it
will mean that the success of any particular kind of content is more con-
vincingly a function of the desire for that content. Or at least, as we’ll see,
this is what the traditional media fear.54
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 120

I n n o v a t i o n f r o m

t h e I n t e r n e t

I n both artistic and commercial contexts in real space, there are barriers
that keep innovators out. These barriers, for the most part, have been eco-
nomic and real: the real cost of resources is a real constraint for most who
would create. These barriers are obviously not absolute; ours is an extraor-
dinarily creative culture; plainly some overcome the limits I’ve described.
Indeed, if markets were perfectly competitive, one might imagine the opti-
mal number that overcomes the barriers I have described. But markets are
not perfect, and costs can be regretted. Hence these barriers are enough to
keep innovators away whom we would not otherwise want to exclude. The
hassle, the uncertainty, the absolute cost: no doubt these together chill many.
These barriers in real space are a function of its nature or, we could say,
its architecture. Not “architecture” in its ordinary sense—buildings and
streets—but architecture in a much broader sense: architecture in the sense
of the set of physical constraints that one finds, even if these are constraints
that man has made. The constraints that are reflected through economics
are constraints of architecture in this sense. You can’t perfectly and cost-
lessly copy a nutritious meal; that takes real resources. You can’t costlessly
and instantly move your car from one coast to another: that takes time and
energy. The constraints of real space are built into the nature of real space,
and though technology presses against this nature, it is only so effective.
Real constraints remain.
Cyberspace has a different architecture. Its nature is therefore different as
well. Digital content can be copied perfectly and practically freely. You can
move a great deal of content almost freely and instantly. And you can repli-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 121

T H E F U T U R E O F I D E A S 1 2 1

cate whatever good there is in one place in many places—almost instanta-
neously. The barriers of cyberspace in its natural state are radically different
from the barriers in real space.
“In its natural state.” I spent many pages in Code arguing against just this
way of speaking. Cyberspace has no nature. How it is—what barriers there
are—is a function of its design, its code. Thus, in this abstract sense, it makes
no sense to speak about the nature of this system that is wholly designed by
man. Its nature is as man designs it.
But cyberspace at its birth did have a certain character. I’ve described
some of it here and more of it elsewhere.1 The feature of its character at its
birth that is most significant for our purposes here is an architecture that dis-
abled the power of any in the middle to control how those at the ends inter-
acted: this is the principle of end-to-end. This design choice of end-to-end
assures that those with a new idea get to sell that new idea, the views of the
network owner notwithstanding.
This principle can operate at very different levels. I described it initially
in the context of a network design. I have argued that the same principle ap-
plies to open code. Spectrum organized in a commons would implement
the principle in the physical layer. The same idea can operate within any so-
cial system. Within law, this is the principle of subsidiarity—decisions are
made at the lowest level appropriate for the decision. Within politics, it is a
principle embraced by libertarians, who urge not no control, but control by
the individual.
We can argue about how far this principle should extend in politics.
Tomes have been written about how far it should extend in law. But my aim
is to push its embrace in the context of creativity. In this domain, at least,
our presumption should be libertarian. And we should build that presump-
tion into the architecture of the space.

as the dot.coms crash and the pundits ask whether there was anything
really new in the new economy of the Internet, it is useful to frame just what
this new space has given us so far. That is my aim in the balance of this
chapter. I want to show how we already have something new, or at least orig-
inally did. My hope is to link instances of innovation to changes in the
layers of control that the Internet effected. This is not a survey; my examples
are illustrative, not representative. But by the end we should have a clearer
sense of the link between these different commons and the innovation these
commons produced.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 122

1 2 2 L a w r e n c e L e s s i g

NEW PRODUCTS FROM THE NET

at the code layer, the Internet is a set of protocols. These protocols make
new digital products possible. These are products that could not, or would
not, have been built before the Net. Among these we could include the dy-
namically generated maps with driving directions;2 massive translation en-
gines, covering scores of languages, translating texts and Web sites on the
fly;3 and on-line dictionaries covering hundreds of languages that otherwise
would not be available except in the largest libraries.4
But let’s focus on a few of these products and their relationship to the ar-
chitecture of the Net.

HTML Books

physical books are extremely durable information sources. They are
stable and preserve relatively well. They read well in many contexts; they
will be a central part of culture for the next century, at least.
But there are things paper books can’t do, and constraints on paper books
that limit how far the knowledge they carry is carried. These limits thus
suggest the place of a different kind of book—the “HTML book,” or a book
produced for the World Wide Web.
HTML books are the passion of Eric Eldred. Eldred was a computer pro-
grammer in the navy. In the mid-1980s, he became aware of the Internet.
The Internet then, of course, was not the World Wide Web—the Web
would not appear for another five years. Nonetheless, the Net facilitated an
exchange of information long before the Web made that exchange hyper-
textual.
When the Web came on-line, Eldred wanted to experiment to see what
the Web might do for books. Eldred’s daughter had an assignment to read
Hawthorne’s The Scarlet Letter. He tried to locate the text on-line. What he
found was essentially unusable. So Eldred decided to make a version that
was usable. He cleaned up the text, added a few links, and created his first
HTML book—a book designed to be read on the World Wide Web.
An HTML book can do things that a paper book cannot. The author of
an HTML book can add links to aid the reader or to guide the reader to
other related texts. It can be easily searched and copied into other texts. And
because it lives on the Web, it is available to anyone anywhere—including
to people who can’t afford to purchase that particular work and to machines
that index the work to be included within search engines.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 123

T H E F U T U R E O F I D E A S 1 2 3

An HTML book is a derivative work under copyright law. If the original
text is protected by copyright, then to publish a derivative work, you would
need the permission of the original copyright holder. The Scarlet Letter,
however, is a work in the public domain. So Eldred was free to take that
work and do with it as he wanted.
This started Eldred on a hobby that soon became a cause. With the pub-
lication of Hawthorne, Eldred began Eldritch Press—a free site devoted
to publishing HTML versions of public domain works (http://eldred.ne.
mediaone.net/). With a relatively cheap computer and an inexpensive scan-
ner, Eldred took books that had fallen into the public domain and made
them available for others on the Net. Soon his site had pulled together an
extraordinary collection of work, including a large collection of the works of
Oliver Wendell Holmes (Sr., not the Supreme Court justice).
Eldred, of course, is not the only on-line publisher of public domain
works. Michael Hart’s Project Gutenberg has been publishing public do-
main texts on the Internet since 1970. But the point is not the uniqueness of
Eldred’s efforts. Indeed, the point is exactly the opposite: The physical and
code layers of the Net enabled this kind of innovation—for Eldred and for
anyone else. The physical layer was cheap; the code layer was open. His
only constraint would come at the content layer—but more on that later.5

MP3

internet texts are not the only innovation enabled by the Net. Much
more dramatic is the innovation in audio and video technologies. MP3
technologies are at the core of the audio changes. They too can be consid-
ered a new product that the Internet has made available.
As I have described, MP3 is the name of a compression technology. It is
a tool for compacting the size of a digital music recording. It works in part
by removing parts of the file that are inaudible to humans. Dogs would no-
tice the difference an MP3 file makes; most of the rest of us are blissfully ig-
norant.
Blissfully—because this deafness of ours means that music can be made
available on the Internet in an efficient and simple way with relatively little
loss in fidelity. A five-minute song can be compressed to a file just
6 megabytes in size. And as connection speeds increase, that 6 MB file can
be shipped to someone else in less than a minute.
This means that the Net becomes a possible distributor for music, and
therefore it inspires a new kind of production: music written and performed
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 124

1 2 4 L a w r e n c e L e s s i g

for sampling on the Internet. This is not just a substitute for CDs or audio-
tapes. The cheap distribution—at different levels of quality—makes it pos-
sible to sample music in a different way. This in turn expands the market for
music, as more can be tried without the commitment to purchase.

Film

a similar change has happened with film. As I suggested at the start of
this book, the costs of production of film have dropped dramatically as digi-
tal equipment has become more powerful and less expensive. We are soon
to enter a time when filmmakers will be able to produce high-quality film
for digital devices at 1 percent of the cost of the same production with tradi-
tional tools. Apple Computer is fueling this demand, with cheap, high-
quality digital film technologies bundled into its popular iMac computer.
This same technology, costing a few thousand dollars today, would have
cost $150,000 just five years ago.
This drop in the cost of production is due to changes in the physical layer
that enables film production. It is also supported at the code layer by a wide
range of tools for manipulating and editing digital images. These together
create an important blurring of the line between amateur and professional.
As a recent Apple ad put it, “And now you’re the purveyor of, you’re the gen-
erator of, you’re the author of great stuff.”
How great, of course, is a matter of taste. In 2001, Apple proudly adver-
tised the first iMac movie purchased by HBO—a short dealing with teenage
pregnancy, produced by a fourteen-year-old kid.6 This is not quite Holly-
wood in the den, but it points to a future that just ten years ago could not
have been imagined—a broad range of film content produced by a wider
range of creators and, in turn, potentially available to others.7

Lyric Server s and Culture Databases

popular culture is diverse and expansive. Finding information about
this culture, however, is often quite difficult. Fans may be many, but the
systematic cataloging of data about such creativity has, so far, been quite
lacking.
Think in particular about music lyrics. Music is an important part of our
life. We grow up listening to songs on the radio; we buy records and listen to
those songs over and over. Our ability to recall music is extraordinary; a few
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 125

T H E F U T U R E O F I D E A S 1 2 5

bars from a song we heard thirty years before will bring back memories of a
certain party or an evening at a concert.
Our memory is of both songs and lyrics. But it is often extremely hard to
locate either. You might remember a particular song, but recall only a few
words. Or you might remember the name of the song, but be unable to find
its author.
The Internet provided an obvious solution to this problem, and by the
mid-1990s, there was an explosion of lyric sites across the Web. These sites
had grown from earlier sites located on the Net. But they quickly became
extremely popular locations where fans might find the words that were
echoing around in their heads.
These sites did not make money. They were produced by fans and hobby-
ists. But though there was no money to be earned, thousands participated in
the building of these sites. And these thousands produced a far better, more
complete, and richer database of culture than commercial sites had pro-
duced. For a time, one could find an extraordinary range of songs archived
throughout the Web.
Slowly these services have migrated to commercial sites. This migration
means the commercial sites can support the costs of developing and main-
taining this information. And in some cases, with some databases, the
Internet provided a simple way to collect and link data about music in par-
ticular.8
Here the CDDB—or “CD database”—is the most famous example. As
MP3 equipment became common, people needed a simple way to get
information about CD titles and tracks onto the MP3 device. Of course,
one could type in that information, but why should everyone have to type in
that information? Many MP3 services thus enabled a cooperative process.
When a user installed a CD, the system queried the central database to see
whether that CD had been cataloged thus far. If it had, then the track and
title information was automatically transferred to the user. If it had not, then
the users were given a chance to contribute to the system by adding the
records necessary to complete the database for that recording.
This meant that quickly, all but the most obscure music was entered into
this large cooperative database. And this database itself then became a prod-
uct that otherwise would not have been available generally.
Each of these new products grows out of the different economics of digi-
tal production and the ability of innovators to add value at the edge of the
network. Some surprise; others are obvious extensions; and still others are
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 126

1 2 6 L a w r e n c e L e s s i g

ideas that are timeless but possible only in this time. But regardless of their
character, they were enabled by the environment of innovation of the origi-
nal Net.

NEW MARKETS

new products beget new markets. And new modes of distribution (in-
cluding the removal of barriers to distribution) induce the creation of new
markets for existing products as well.
Consider just one example: the production of poetry. The market for po-
etry is extremely small; the burden in getting poems published is excep-
tionally great. These real-space constraints translate into an extraordinarily
difficult market for poets.
But using both the cheap distribution of the Internet and tools for better
structuring the delivery of content, the market for poetry on the Internet has
taken off. There has been an explosion of sites dedicated to producing and
distributing poetry. Poetry Daily, for example, launched in 1997, receives
over 150,000 visitors a month. At peak usage, 12,000 users come to the site
per day. And over 16,000 subscribe to a regular content newsletter.9
These sites are not simply tools for delivering poems more cheaply. Their
technology also enables better control over how that content is consumed.
Some sites have technologies for guiding the reading of poetry—thereby
making it accessible to or understandable by a much wider audience.10
Others enable the audio reading of poetry, similarly enabling a market for
the blind that otherwise would have been restricted.11 These tools in turn
expand the reach of this form of creativity to people who otherwise would
not consume this poetry.

NEW MEANS OF DISTRIBUTION

the most dramatic potential for affecting creativity is, as Coase described
earlier, the lowering of the transaction costs of distribution, and hence the
expansion of the extent of the market. Here there are a number of well-
known examples that we will consider again when we examine changes in
copyright law and their effects on the Net.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 127

T H E F U T U R E O F I D E A S 1 2 7

My.MP3

the compression technique of MP3 is “free” in the sense I have de-
scribed it: anyone is allowed to develop technologies that use it.12 And
through the 1990s, thousands of such technologies developed. It was ex-
tremely easy to find a program on the Net that would “rip” the contents of
your CDs—meaning copy the music in an MP3 format so that you could
store the music on your machine. And many MP3 players were offered for
free or for sale.
But though MP3 files were small relative to their original data file, they
were still quite large to be sent across the Internet. While a fast connection
could chomp through an MP3 song in a few seconds, on a standard tele-
phone connection it could take twenty to thirty minutes.
This restriction in bandwidth gave birth to an important industry of
streaming technologies. The idea was quite simple: Rather than download-
ing the full copy of music and then playing it, streaming technologies allow
the user to stream the desired content and play it at the same time. No
copies of the file must be made first, which means the user need not waste
time waiting for the music to be delivered.
RealAudio was the innovator here, though its idea was soon mimicked by
Microsoft and Apple. RealAudio sold tuners that enabled people to tune in
to audio, and then video, content. That content was compressed and
streamed across the Net.
MP3.com took this idea further. Started in 1997, MP3.com has no real
relation to the technology MP3. It didn’t invent the technology, and it had
no exclusive license. Nor did its founder, Michael Robertson, have any real
relation to the music industry. Robertson was simply an entrepreneur who
saw the Internet as a great new opportunity. MP3.com was started to find
new ways to use the technology to produce and distribute music.
MP3.com pushed production by encouraging artists to produce and dis-
tribute music across its site. This was not unique to MP3.com. It was the
business model of companies like EMusic.com as well.
But more interesting were MP3.com’s new ideas about how to push dis-
tribution. The existing labels had a clear idea about how the business
worked. Their business, Robertson explains, “is making a bet. They make
big bets on a small number of acts, hoping that one of them is the lottery
ticket and pays off.”13 And to help make sure their bets paid off, as artist
Courtney Love explains, “record companies controlled the proportion and
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 128

1 2 8 L a w r e n c e L e s s i g

the marketing; only they had the ability to get lots of radio play, and get
records into the big chain stores.”14
When representatives from the existing labels spoke to Robertson, they
tried to convince him to follow the same model. “They were all about, okay,
we gotta find the next U2 or Backstreet Boys, or whatever. . . . Try to find
that one, and promote the heck out of him, and hope we break even.” Own-
ing the artist is key to this model; the traditional labels thus demand exclu-
sivity.
But Robertson’s model was different.

I just said, that’s not our model. Why don’t we let somebody else make the
music and produce the music and do whatever they do in the creative
process? And we’ll just pick up after that creation is already done, and
worry about the delivery.15

So MP3.com rejected exclusivity (Robertson: “Exclusivity is a very bad
thing for content owners”) and worked instead on technologies to make it
easier for customers to get access to music. And using subscriptions and ad-
vertising, the company expected to make this way of getting content pay.
A core feature of this technology was an “automatic” way for popular con-
tent to find its way to the top. As Robertson describes it, the key is collecting
good data. For “data changes the balance of power.” Consumers “listen to
the good music, and we’ll make sure the good music floats to the top.”16
The floating follows the listening. The more users listen and download
music, the more the music “floats”; and the more the system learns the pat-
terns of who likes what, the more the system can make sure that the music
you like is likely to float to the top of your screen.
MP3.com thus conceives of itself as a “service bureau.”

[Artists] can come and go as they please. . . . You don’t even have to agree
to exclusively give it to us. It defies all logic in the music industry. I can’t
tell you how many people came into my office and said, “You don’t get it.
You don’t get the business. I don’t know why you’re not forcing these guys
to give you a piece of their intellectual property. Because you’re gonna
make the next Madonna, and you’re not gonna own the next Madonna.”17

There was one innovation, however, that earned MP3.com more than
the sucker scorn about Madonna. This was the My.MP3 service, launched
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 129

T H E F U T U R E O F I D E A S 1 2 9

in January 2000. Using this service, consumers could get access to their
music in two different ways. First, through cooperating sites, they could pur-
chase music. Those cooperating sites would then send a CD to the cus-
tomer but immediately make the music purchased by the customer
available in the customer’s MP3 account. Once available in a user’s ac-
count, the customer would be able to stream that music to his or her com-
puter, wherever the customer was. MP3 would then keep a collection of
music stored in that account, and give the customer access to it wherever he
or she had access to the Web.
The second aspect of My.MP3 was a bit bolder. MP3.com released a pro-
gram called Beam-it. If you (the customer) had Beam-it installed on your ma-
chine, you could insert a CD into the computer, and the MP3.com service
would then identify what the CD was. If MP3.com had that CD in its li-
brary, it would make that music available to you. Thus, you could take all
the CDs you had at home, “beam” them to the MP3 server, and subse-
quently get access to your music anywhere else you happened to connect to
the Web.
Plainly, consumers could do much of this without MP3.com’s help. My
former colleague at Harvard Jonathan Zittrain was an early on-line music fa-
natic. He bought a basic computer from Dell Corporation with a large hard
drive, and proceeded to copy all his CDs onto the computer. He then con-
nected the computer to his cable modem and designed the system so that
wherever he was (and mainly in his office) he could listen to his music. Zit-
train thus built a music server like MP3.com which distributed his music on
demand.
The difference with MP3.com was that you didn’t need to be Zittrain to
get your music. Nor did you have to waste the time that Zittrain wasted to
copy all of your music. The Beam-it service would recognize your disk in
less than ten seconds; you would have access to your music after a few sec-
onds more. There was no need for a large disk drive to store all your music.
All the music was stored at MP3.com’s servers.
This service by MP3.com made it easier for consumers to get access to
the music they had purchased. It was not a service for giving people free ac-
cess to music. Of course, people could borrow other people’s CDs and
hence “steal” the content of those CDs if they wanted. But that was possible
before MP3.com came along. MP3.com’s aim was simply to make it easier
to use what you’d already bought.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 130

1 3 0 L a w r e n c e L e s s i g

Napster

no dou b t the most famous story of musical “innovation” has been the ex-
plosion called Napster—a technology simplifying file sharing for MP3 files.
The idea was the brainchild of Shawn Fanning and Sean Parker—in the
eyes of many, children themselves.
Fanning and Parker’s idea was just this: Individuals had music stored on
their computers—think again about the hoarder Zittrain. Others would
want copies of that music. Fanning devised a way to engineer a “meeting”
between the copies and those wanting them. His system, Napster, would
collect a database of who had what. And when someone searched for a par-
ticular song, the database would produce a list of who had that song and was
on the line at that moment. The user could then select the copy he or she
wanted to copy, and the computer would establish a connection between
the user and the computer with the copy. The system would function as a
kind of music matchmaking service—responsible for finding the links, but
not responsible for what happened after that.
Napster is an “ah-ha” technology: you don’t quite get its significance until
you use it. The experience of opening a Napster search window, rummag-
ing through your memories for songs you’d like to hear, and then, within a
few seconds, finding and hearing those songs is extraordinary. As with the
lyric database, you can easily find what is almost impossible to locate; as
with the MP3 server, you can then hear what you want almost immediately.
Music exchanged on Napster is free—in the sense of costing nothing. And
at any particular moment, literally thousands of songs are available.
The innovation excited an immediate legal reaction. The Recording In-
dustry Association of America (RIAA) immediately filed suit against Napster
for facilitating copyright violation. That may have been a mistake. At the
time the RIAA filed suit, the number of Napster users was under two hun-
dred thousand; after the suit hit the press, the number of users grew to fifty-
seven million.
In chapter 11, we will consider in some depth the legal questions that
Napster raised. Focus for the moment just on the innovation. For what Fan-
ning had done was to find a way to use the dark matter of the Internet—the
personal computers connecting the Net. Rather than depending upon con-
tent located on a server somewhere—in this strict hierarchical client/server
model of computing—Fanning turned to the many individual computers
that are linked to the Net. They could be the place where content resides.
Using the protocols of the code layer, he was able to find an underutilized
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 131

T H E F U T U R E O F I D E A S 1 3 1

asset at the physical layer. Hence the importance of these centralized
servers would be reduced.
Many believe the central motivation behind Napster was to find a way to
avoid “the copyright police.” For some content (but not streaming content),
centralized servers are more efficient places for storage. But they are also
more efficient places to harbor illegal content. And if it is illegal content
that is being stored, the efficiency of the storage gets outweighed by the risk
of getting caught.
But if the content is located on many machines set up individually, then
the content is hard to find and it becomes difficult to prosecute those har-
boring it. So Napster imagined that individuals would put up content that
would be available for others, but the individuals holding the content would
not be so regular as to be targets of prosecution.
These groups, however, would serve a separate function as well. They
would induce the exchange of information about preferences among mem-
bers of these groups. That information would induce an expansion of de-
mand by consumers of this music. And that demand in turn could be
satisfied by music from Napster or from the ordinary channels of distribu-
tion.
To the extent you view Napster as nothing more than a device for facili-
tating the theft of content, there is little usefulness in this new mode of
distribution. But the extraordinary feature of Napster was not so much the
ability to steal content as it is the range of content that Napster makes avail-
able. The important fact is not that a user can get Madonna’s latest songs for
free; it is that one can find a recording of New Orleans jazz drummer Jason
Marsalis’s band playing “There’s a Thing Called Rhythm.”
This ability competes with the labels, but it doesn’t really substitute for
the demand they serve. A significant portion of the content served by Nap-
ster is music that is no longer sold by the labels.18 This mode of
distribution—whatever copyright problems it has—gives the world access to
a range of music that has not existed in the history of music production.
Once you taste this world of almost limitless access to content, it is hard
to imagine going back. What Napster did more effectively than any other
technology was to demonstrate what a fully enabled “celestial jukebox”
might be.19 Not just for the music that distributors want to push at any par-
ticular moment; not just for the music that a large market would demand;
but also for practically any recording with any fans using the service any-
where, the music was available.
This represents the end of the progression that began when broadcast
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 132

1 3 2 L a w r e n c e L e s s i g

channels started multiplying. When television, for example, is just a few sta-
tions, then the producers of television have a great deal of power to guide
the audience as the program director chooses. As the channels multiply,
each channel becomes a competitor of all others. At any one time, the num-
ber of competitors for attention has increased, and the effective choice is
much greater.
Napster represents the extreme of this trend. Channels here no longer
channel consumers. Consumers have the broadest range of choices possi-
ble. Thus, just as it is for an avid reader in a very large library, the content of
music becomes available for individuals to choose rather than available as
disc jockeys choose.

NEW DEMAND

in the stories so far, the Internet affects the content and the applications
that meet existing demand. But this is only one part of how the Net is dif-
ferent from real space and ultimately, I believe, the less interesting part. Far
more important is how the Net changes how people learn what they want
and how these wants might be changed.
Consider here the bright part of Amazon.com. There are many who are
concerned by the emergence of this powerful bookseller. Shifting book sales
to the superstore can’t help but reduce sales in the smaller shops.
But to stop here in the analysis is to make the story a zero-sum game when
it is far too soon to know the net effect. The emergence of Amazon.com and
the like has created a boom in demand that could not be built in real space.
The reason is the complexity of our preferences and how hard it is for oth-
ers to speak to them. Media observers looking at real-space media doubt this
complexity; most media to them look extremely homogeneous. But a
speaker may utter a general message either because he or she wants to say
something bland or because the costs of saying something more specific are
too high. And these costs can be too high either because the message can’t
be targeted or because the costs of knowing what each person would want
to hear are prohibitive.
Focus on the last point first—knowing what someone wants to hear. You
all have friends; some of you have more friends than others; some of your
friends are closer than others. But I doubt any of your friends knows your
tastes in music and books as well as Amazon knows mine. After a three-year
relationship, dutifully remembered by Amazon’s data-mining engine, Ama-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 133

T H E F U T U R E O F I D E A S 1 3 3

zon can recommend to me things that I ought to buy. It advertises to me,
but its advertisements—unlike 99 percent of the ads I see in real space—
actually speak to me. They actually say something that I want to hear. And
because they speak to me, I listen.
It is technology that makes this possible. There is no editor at Amazon
who decides who should read what. There is just a machine. Preference-
matching engines are constantly gathering data about what I buy and what
others buy; Amazon adds to that data preferences that I express. It has a
handy way for me to signal how much I liked something I bought or
whether it should exclude something I bought from its data review. And
when it lists recommendations, it lets me tell it that I own what it recom-
mends or that I would never buy what it recommends. Amazon knows,
based on real data.
This technology for preference matching was not born at Amazon. A
company called Firefly first deployed the technology.20 But the idea spread
quickly. And every successful on-line merchant now competes to under-
stand what its customer wants. As Michael Robertson of MP3.com said,
“Data changes everything.” With the kind of data that can be collected, the
Web can deliver a kind of service that would otherwise be impossible.
This technology is in its infancy, but it is old enough for us to catch a
glimpse of where it is going. As it matures, the technology (unless stopped)
will increasingly be able to predict what you, as an individual, will want;
it will increasingly waste less of your time on what you don’t want. And this
will increase your demand for the things the Net sells.
From the start, there have been skeptics about these technologies. Some
fear the loss of privacy resulting from these systems of perpetual monitor-
ing.21 Others fear that these systems for giving you just what you want will
result in increasingly isolated individuals.22 Noise, the argument goes, helps
build communities. My ability to perfectly select what I hear and what I
read may help me, but it may harm society. Increasingly isolated individu-
als are not the stuff that communities are made of.
Both concerns may prove to be true, but I want to put both aside for the
moment and focus on the positive consequences of this emerging tech-
nology. The most direct effect of this technology is to make data about peo-
ple’s preferences much more usable—usable, that is, in informing others
about individual preferences.
These data are a social resource that the architectures of social life have
made hard to collect so far. We’ve had ways of collecting how many people
bought DVD X, or book Y, but no good way of collecting how many who
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 134

1 3 4 L a w r e n c e L e s s i g

bought book X also bought book Y. And just as the invention of a new ma-
chine can in turn reduce the cost of some production process, the ability to
capture and use this new resource will reduce the costs of advertising dra-
matically. Rather than technologies that produce 0.5 percent return,23 these
technologies will produce a much greater return at a lower cost.24
Thus the architecture of the Net enables a resource otherwise unavail-
able. And through this resource, a barrier to entry is reduced. For if one of
the hardest parts of breaking into music is the cost of promotion, then as
long as these data mines remain competitive, lowering the cost of promo-
tion will make it easier to break into music. The same is true with books or
any other content. By increasing the demand for a diverse selection of con-
tent, and by enabling the cheaper identification of that demand, the Net
widens the range of potential contributors.

N E W PA R T I C I PA T I O N : P 2 P

finally, consider an innovation that enables a new kind of participa-
tion that many have called the next great revolution for the Internet, and
that in light of our discussion of e2e will be quite familiar: the emergence of
peer-to-peer (P2P) networks.25
A peer-to-peer network is one where the content is being served not by a
single central server, but by equal, or “peer,” machines linked across the
network. Formally, “p2p is a class of applications that take advantage of
resources—storage, cycles, content, human presence—available at the
edges of the Internet.”26 In the sense I’ve described, this was the architecture
of the original computers on the Internet—there wasn’t a set of central
servers that the machines connected to; instead, there was a set of e2e pro-
tocols that enabled data to be shared among the machines.
But as commerce came to the Net, the character of this architecture
changed. As the World Wide Web became more popular, Web servers be-
came dominant. And as servers grew, the equal peer-to-peer structure of the
Internet was replaced by the hierarchical structures of client and server.
This was not intended. When Berners-Lee invented the World Wide
Web, he didn’t really have in mind centralized Web servers broadcasting
tons of content to the many; from the very start, he tried to push developers
of browsers to develop them as two-way devices—allowing both the viewing
and the writing of HTML code. Berners-Lee wanted a peer-to-peer Web,
and his technology enabled that. But in the first generation of its deploy-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 135

T H E F U T U R E O F I D E A S 1 3 5

ment, that wasn’t how it was deployed.27 To large ISPs like AOL, the ends
are important only in the way customers in a mall are important—they con-
sume the commerce but don’t participate in its design.
More recently, however, a change in the Net’s architecture has occurred.
As machines have become more powerful, and as they have become more
reliably and permanently connected to the Net, the possibility of using
peers to process and forward data on the Net has increased. Peer-to-peer
services are returning to the Internet as machines mature and are persis-
tently on the Net. The character of what can happen is changing, and the
potential—if left free to develop—is extraordinary.
Napster is the most famous peer-to-peer technology, even though it is not
exactly peer-to-peer. (There is a central server that keeps a database of who
has what; the music itself is kept on other people’s machines.) But Napster
is the horse and buggy in this transportation system. It is only the beginning.
Consider the SETI project. SETI—the Search for Extraterrestrial
Intelligence—scans the radio waves for evidence of intelligent life some-
where else in the universe. It does this by recording the noise of the radio
spectrum that we receive on planet Earth. This noise is then analyzed by
computers, looking for telltale signs of something unexplained.28
Who cares about wandering X-Files types?, you might ask. Is this really it?
But the point is the potential that SETI evinces. The computation involved
in the SETI project is immense, and they soon discovered that the cost of
renting computers to process these recorded radio waves was increasingly
prohibitive. So researchers at Berkeley had an idea: Facilitate the distribu-
tion of chunks of this recorded data to machines across the Net, then enable
these machines across the Net to do the computation required on this data.
A package of data would be delivered to the participating computer along
with a program to be run; that program would run on the data and send it
back to the mother ship.
When the SETI@home project first began, within ten days it had
350,000 participants in 203 countries. In four months, it broke a million
users. The service grew so fast that it had to stop processing data for a while.
The speed at which data was being collected had surpassed the processing
speed.
In mid-2000, the system could boast the equivalent of 280,000 years of
processing time devoted to the SETI mission.29
Just as Napster had latched on to unused disk space, SETI@home had
latched on to unused computer cycles living at the edge of the Net. Idle ma-
chines could be turned to large-scale cooperative projects.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 136

1 3 6 L a w r e n c e L e s s i g

Others had seen the same point. Before SETI, in response to a challenge
by RSA Labs, distributed.net deployed a technology that would enable
computers linked to the Net around the world to help distributed.net crack
a DES (digital encryption standard)-encrypted message. The first successful
crack of a DES-encrypted message was in 1997, by Rocke Verser. His group
cracked the message in ninety-six days. By 1999, distributed.net had im-
proved this “distributed technology” (meaning software that ran remotely
on a large number of independent machines) sufficiently that one hundred
thousand computers around the world—using idle computer time—were
able to crack an encrypted message in just twenty-four hours. The govern-
ment had estimated this cracking would take many years. distributed.net
proved the government wrong.
These examples suggest a future that many companies are scrambling to
adopt. Large computing projects can be carved into manageable bites and
then shipped to cooperating computers everywhere on the Net. The com-
puters simply process the data as instructed by the organizing machine, and
the processed data is then shipped back to the organizing machine. From
the user’s perspective, the nature of the data being processed could be com-
pletely obscure. The point is that the idle time of the machine could be
harnessed to the end of getting something done.30 As Howard Rheingold de-
scribes it:

At its most basic level, distributed processing is a way of harvesting a re-
source that until now has been squandered on a massive scale: unused
CPU cycles. Even if you type two characters per second on your keyboard,
you’re using only a fraction of your machine’s power. [Distributed com-
puting bands together millions of computers] on the Net to create, with
their downtime, ad hoc supercomputers.31

The potential of peer-to-peer technologies reaches far beyond simple file
transfer or the sharing of processing cycles. Indeed, as researchers are com-
ing to understand, the most important peer-to-peer technologies could be
more efficient caching technologies. A “cache” is simply a copy of content
kept close to the user, so that less time is needed to get access to that con-
tent. P2P caching solutions imagine using computers at the edge to more
efficiently store data. A user might be paid by a network, for example, to re-
serve 1 gigabyte of his or her hard disk for the network to do with as it wants.
The network could then store content on that disk and keep it closer to the
users in the customer’s region. By sharing space at the edge of the network,
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 137

T H E F U T U R E O F I D E A S 1 3 7

these caching systems can deliver content more efficiently—more cheaply
and more quickly. The speed and reliability of streaming technology using
peer-to-peer technologies are significantly increased.
So too is the intelligence. Consider Gnutella. Gnutella is best known as
an alternative to Napster. Rather than relying upon a centralized server to
keep the list of which users have which files, Gnutella uses a peer-to-peer
querying algorithm, which makes the central server unnecessary. This
means that it is much harder to control the content that gets exchanged in
a Gnutella system. And this means in turn that Gnutella so far has been
used to facilitate the exchange of a wide range of content too dangerous to
post on the Web.
But the real value to Gnutella is not this ability to exchange questionable
content. The real potential is for Gnutella to expand the power of searching
technologies beyond their presently limited scope. Search engines today
miss an increasingly large proportion of the Internet. The engines were ini-
tially designed to index static Web pages. As more and more of the Net be-
comes dynamic (meaning the content displayed is generated each time a
Web page request is made), the search engines miss this content.
The Gnutella technology suggests a way around this limitation. By en-
abling a more sophisticated language with which to relay requests to end de-
vices, Gnutella would make it possible for a search to be launched on the
Internet, in which individual machines within the Gnutella network pro-
cess that search however they think best. As Andy Oram describes it:

Gnutella offers the path forward. It governs how sites exchange informa-
tion, but says nothing about what each site does with the information. A
site can plug the user’s search string into a database query or perform any
other processing it finds useful. Search engines adapted to use Gnutella
would thus become the union of all searches provided by all sites. A
merger of the most advanced technologies available (standard formats like
XML for data exchange, database-driven content provision, and distrib-
uted computing) could take the Internet to new levels.32

This is innovation at the edge of the Internet that implicates its core func-
tionality. It would radically advance the function of the Net, moving the Net
from its costly, centralized server architecture to a more distributed and
flexible architecture.
These peer-to-peer innovations are enabled by the commons at the code
layer. And indeed, the strongest complaint against them is that some of
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 138

1 3 8 L a w r e n c e L e s s i g

these technologies tend to suck up all available bandwidth. Napster—at
least while threatened with extinction—was an example of this use. Univer-
sity systems were overburdened as student accounts became public servers.
There is some fear that a generalized Gnutella system would similarly con-
sume bandwidth.
But for this technical problem there are technical solutions. And better-
implemented peer-to-peer technologies should better balance demand and
supply. What is certain, however, is that they provide a kind of innovation
that would not have been possible without the commons of the Net.

in each of these stories the lesson is the same. The platform of the Inter-
net removes real-space barriers; removing these barriers enables individuals
with ideas to deploy those ideas. The architecture is different; the innova-
tion it encourages is therefore different.
This encouragement comes from a number of sources. First: Because of
the commons at the code layer, there is no cop on the block. The Net does
not enable control; in this sense, it therefore encourages those ideas that
would have been blocked by a system of control. Where many people have
to sign on before a project gets going, the opportunity for irrational vetoes
becomes quite great (corporations protecting their vision of the market;
management restricting how much they want their departments to change).
Second: Because access to the physical layer is so inexpensive, the mar-
ket linked to this commons is vast. The market for solar-powered Beanie Ba-
bies might be quite small in relative terms. But if the market is the whole
world, then the Net would encourage what otherwise could not be sus-
tained.
Third: Because of the character of the code layer, there is an opportunity
to exploit a resource that is prohibitively expensive in real space, save for
the very large organization—data. Top-down advertising is replaced by
bottom-up marketing, which in turn makes it easier for creators without
great backing to enter the channel of distribution.33
The innovations that I have described flow from the environment the Net
is. The environment is a mix of control and freedom. It is sensitive to
changes in that mix. If the constraints on the content layer are increased, in-
novation that depends upon free content will be restricted. If the access
guaranteed by a commons at the code layer becomes conditioned or re-
stricted, then innovation that depends upon this access will be threatened.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 139

T H E F U T U R E O F I D E A S 1 3 9

The environment balances the free against the controlled. Thus, preserving
this environment means preserving this balance.
But nothing guarantees that this mix will remain as it is. There is no “na-
ture” of the Internet that will assure a continued commons at the code layer,
no strong protections limiting Congress to ensure that adequate resources
remain free at the content layer. And most troubling in this environment,
there is an increasing pressure to use contracts to muck about with the free-
doms enabled at the physical layer—to retrofit those technologies, that is, to
ensure that they don’t threaten existing interests (as we’ll see more of in
chapter 10).
This mix constitutes an environment for innovation; this environment
enables the new against the old. But the old have an interest in under-
mining this environment, to the extent it threatens the old. And the means
to pollute this environment are well within reach. This requires environ-
mentalism in the Internet era.34
A crude economics is skeptical of this dynamic of the old versus the
new—not skeptical that the old have an interest in resisting the new, but
skeptical that they have the means.35 If the new represent a more efficient
technology, then, over time, that efficiency will drive out the old. There may
be struggles in the meantime, and these struggles will no doubt waste re-
sources. But the question is not just whether the old will interfere, but
whether intervention to protect the new won’t itself be more costly.
At any particular moment, this question of trade-offs will be important.
But the issue as I’ve presented it is not the question of trade-off. The issue is
not which technology we can expect to win in the long run. It is, instead,
what architecture for innovation best speeds us along the path to the long
run. Which architectures encourage experimentation? Which permit the
old to protect themselves against the new? Which permit the new the most
freedom to question the old?
This raises again precisely the issues that Christensen describes in his In-
novator’s Dilemma. As I’ve said, Christensen’s argument doesn’t depend
upon stupidity. He is not identifying a failure in the market; he’s identifying
a feature of successful companies’ perspective. The problem is not that the
decision maker is irrational; the problem is that innovation is controlled by
the wrong decision maker.
The solution is to architect the system to vest the power to innovate in a
more decentralized manner. Or, put a different way, the solution is to archi-
tect innovation to be free rather than controlled. The lesson that the failure
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 140

1 4 0 L a w r e n c e L e s s i g

of the Soviet East teaches is that innovation controlled by the state fails. The
lesson that Christensen and others teach is that innovation controlled by the
most successful in the market will be systematically blind to new forms of
creativity. This is the same lesson that the Internet teaches.

let’s take stock: The architecture of the original Internet minimized the
opportunity for control, and that environment of minimal control encour-
ages innovation. In this sense the argument is linked to an argument about
the source of liberty on the original Internet. At its birth, the Internet gave
individuals great freedoms of speech and privacy. This was because it was
hard, under its original design, for behavior on the Net to be monitored or
controlled. And the consequence of its being hard was that control was
rarely exercised. Freedom was purchased by the high price of control, just
as innovation is assured by the high prices of control.
But the story about liberty on the original Net had a sequel: what the ar-
chitecture could give, the architecture could take away. The inability to
control was not fixed in nature. It was a function of the architecture. And as
that architecture changed, the ability to control would change as well.
In Code, I argued that the original architecture of cyberspace was chang-
ing, as governments and commerce increased the ability to control behav-
ior in cyberspace. Technologies were being deployed to better monitor and
control behavior, with the consequence, for better or worse, of limiting the
liberty of the space. As the architecture changed, the freedom of the space
would change, and change it did.
In the balance of this book, I argue that a similar change is occurring with
respect to innovation. Here, too, the architecture of the space is changing,
interfering with the features that made innovation so rich. And the conse-
quence again will be a decrease in this value that we thought defined the
original Net.
But here, the change in architecture is both a change in the architecture
in a technical sense and a change in the legal architecture within which
cyberspace exists. Much more significant than in the story I told about lib-
erty, the emphasis here is on the interaction between changes in law and
changes in code that together will undermine innovation.
Or at least innovation of a certain kind. The story I want to tell is not
about the death of innovation generally; it is about the relocation of inno-
vation from the diverse, decentralized Internet back to the institutions that
policed innovation before. The story is about the bureaucratization and
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 141

T H E F U T U R E O F I D E A S 1 4 1

capture of the innovation process—relocating it back to where it was—as a
response to the structures originally enabled by the Internet itself.
Put differently, this is a story about changes in code, both East Coast (by
lawmakers in Washington) and West Coast (by software writers in Silicon
Valley) code, which will restore some of the power of the old against some
of the threat of the new.
My claim is not that therefore no change will occur. Obviously, the world
will be different when it is Internet enabled from how it was before the
Internet. But it will not be as different as the platform promised, and the
changes back to the model of the dark ages are not supported, or justified,
by the economics that justified the dark ages. Instead, the power that is cre-
ated here is importantly artificial—the product of legal rights created in the
air and defended with the vigor of courts and code.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 142
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 143

P A R T

I I I

///

DOT.COntrol
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 144
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 145

O l d v s . N e w

P icture the Soviet Union during its final days. Think about the most
powerful within that system. These were not inherently evil people,
though no doubt, especially during the Stalinist era, there were plenty of
psychopaths hanging about. They were quite ordinary in many ways; if you
met them at a party, many would strike you as quite liberal and sensible.
They were not terribly rich, though some were. And they didn’t live in a
flourishing society, though they were promised just this as they grew up in
the USSR. But by the late 1980s, everyone knew the system had failed. Yet
very few were willing to take steps to free that society from state control.
Why? Why wouldn’t these “leaders” try to move their society to a better
place? Why wouldn’t they voluntarily push for a different system of control?
It doesn’t take a deep understanding of human psychology to answer this
question. Things may have been bad, but how would these leaders know
that for them and their families, things would be better under any other
system? What incentive was there to release the reins of control, when the
resulting system could promise so little that was certain? Like the manage-
ment of a successful company, they could see the marginal improvements
that were possible if they stayed on course. But they could not be confident
of improvement if they jumped the other way.
(Here the story of the malevolent giant begins to make more sense. One
could well believe the leaders of the Soviet Union expected their society as
a whole would be better off under freedom but also believed they would not
be able to extract enough of that social gain to make them individually as
well off.)
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 146

1 4 6 L a w r e n c e L e s s i g

Now picture the leaders of dominant industries, faced with a disruptive
technology. What is their rational response? Is it any different?
The perfect marketeers presume these actors would behave differently
from the Soviets. They presume the leaders of dinosaur firms would spin
those firms on a dime, to become something radically different.
But why would one believe that? Faced with a disruptive technology that
threatens their way of life—their mode of doing business, their vision of the
market—why would these leaders voluntarily step down from their place
and enter a different market with uncertain returns? Why instead isn’t the
story that Christensen tells—like all deep truth—obvious, once we see it?
And even more obvious, why wouldn’t we expect these leaders of existing
dominant industries to use whatever power they have to protect themselves?
Rather than yielding to the new technology, wouldn’t they take steps to pro-
tect the old against the new?
What steps would these be? In the story I have told, there are any number
of levers that the old Soviets might use. Most obviously, they could use the
force of law to stifle innovation that challenges their power. Or they could
use market power to chill the willingness of innovators to challenge their
position. Or they could use norms to stigmatize the deviants. Or they could
use architecture to hinder the opportunity for innovators to innovate. Any
one of these techniques could help strengthen the power of the existing So-
viets; any one could be deployed to weaken the opportunity of a challenger.
The balance of this book is a story about how our “old Soviets” are doing
precisely this with the Internet. “Soviets” is an unfair term, I know, but the
image is precise even if unfair. Changes threaten the power of those now in
power; they will work in turn to protect themselves from the changes. In the
balance of this book, I want to detail their work to change the Internet, and
the legal culture surrounding it, to better protect themselves. Some of these
changes are legal; some are technical; and some use the power of the mar-
ket. But all are driven by the desire to assure that this revolution doesn’t
muck things up—for them.
There’s nothing immoral in this desire. This is not a battle between good
and evil. Stockholders demand that management maximize its income; we
shouldn’t expect management to do anything different.
But even if this is “only business” to them, that does not mean it should
be “just business” for us. We need not stand by idly as the Internet is
changed. They have their interests; we have ours. And for those who believe
that the environment of creativity that the Internet produced was worth
something, there is reason to resist the changes that I will describe.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 147

C o n t r o l l i n g t h e W i r e s

( a n d H e n c e t h e C o d e L a y e r )

In chapter 3, I described an architectural principle that I said helped
build an innovation commons: end-to-end. I also described the struggle to
assure that in effect that principle would govern on the telephone lines.
Keeping those channels open to enable this commons of innovation was an
important, if forgotten, part of the history that gave us the Internet.
The lesson from that story was of the power that came from an inability to
control: the innovation and creativity that were inspired by a platform that
was free.
If there was a time in the past decade when we had learned this lesson,
the story of this chapter is that we have now forgotten it. The changes that I
will describe in the pages that follow are all examples of the network being
rearchitected for control. I have called the inability to discriminate a feature
of the original Net’s design. But to many—and especially those building out
what the network called the Internet will become—this “feature” is a bug.
The power to discriminate is increasingly the norm; building a Net to en-
able it is the aim.

the internet was born on networks linking universities, but it took its
first step when it came to the phones. It was when ordinary individuals
could dial up an Internet connection that the Internet came alive.
Long before people started dialing into the Internet, however, many were
already members of on-line services and on-line communities. Compu-
Serve and Prodigy were early market players. America Online came a bit
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 148

1 4 8 L a w r e n c e L e s s i g

later. These services were born serving content of their own. They were not
Internet portals. There was indeed no way to move from their proprietary
system to the nonproprietary Internet.
By the mid-1990s, all this changed as the attraction of the Internet grew,
and as the competitive threat of ISPs increased. As more and more saw the
Internet as an attractive alternative to the edited content of the existing ser-
vice providers, they pressed their service providers to provide access to the
Internet.
As I’ve suggested, the part of this story that is too often missed is the role
that the telephone company played in the birth of the Net or, more accu-
rately, the role the telephone company did not play. For what is striking
about the birth of this different mode of communication is how little the
telephone companies did in response. As their wires were being used for this
new and different purpose, they did not balk. They instead stood by as the
Internet was served across their wires.
This was no slight change. When telephones are used for talking, the av-
erage usage at a particular house is quite small. Calls are ordinarily short, so
the number of circuits needed in a particular region is few.
But when phones began to be used to link to the Internet, this usage
changed dramatically. Calls no longer lasted a few minutes on average. Peo-
ple were dialing in and hanging on, and the burden placed on the tele-
phone system was great. The average voice call typically lasts only three to
five minutes; the average Internet “call” lasts seventeen to twenty minutes.1
Ordinarily, one imagines that telephone companies would be quick to re-
spond to this change in usage. They would either be quick to increase rates
for calls over a certain length or they might restrict usage to certain kinds of
telephone numbers (such as those to the ISPs). And we might imagine that
telephone companies, if they were creative, would decide to become their
own Internet service providers, offering better rates internally than they did
to other Internet service providers. In short, there are any number of games
telephone companies might play to respond to this demand for Internet ser-
vices.
Phone companies, however, did not play these games, because they were
not allowed to. And they were not allowed to because regulators stopped
them.2
As we saw in chapter 3, the telephone company had become a disfavored
monopoly. Its power over its wires had first been limited in 1968, in the
Carterfone decision,3 and then after growing resistance by Congress and the
FCC, most dramatically by the Justice Department in the early 1980s. In
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 149

T H E F U T U R E O F I D E A S 1 4 9

1984, a decree breaking up AT&T was entered, and over the next ten years,
Congress and the FCC struggled to find a model under which the Bells cre-
ated by the breakup would be regulated.
The model finally fixed upon—and ratified by a statute of Congress in
1996—imposed an obligation on the Baby Bells to be neutral about how
their lines would be used. The Baby Bells were required to unbundle the
services they offered and make it possible for others to compete directly with
them. If you wanted to start an ISP, you could connect your service into the
telephone company’s office. Their wires in a sense became your wires. The
important point was preserving and defending neutrality.4
This imposed neutrality had an unintended effect on the Internet and its
growth, because while the regulators imagined creating competition in the
telephone service, they did not have in their head the idea that this might
create a kind of competition with telephone service. They did not imagine
the birth of the Internet as a product of their accidental regulation. But that
is precisely what their regulation produced. This imposed neutrality about
how the wires would be used left the field open for others to use the wires in
ways no one ever expected. The Internet was one such way.

THE END-TO-END IN TELEPHONES

as i described in chapter 3, the end-to-end argument says intelligence in a
network should be located at the edge, or ends, of the network, and that the
network itself should remain simple. Only those functions that must be
placed in the network are placed in the network. Other functions—other
intelligence—are left to the applications that run on the network.
The TCP/IP Internet was designed as an end-to-end network. The proto-
cols of TCP/IP simply enable data to be sent across the network. They regu-
late how data is to be divided and how the resulting packets are shipped.
They don’t at all care about what is built into the data or how that built-in
part works.
Not all networks are end-to-end in this sense. A contrasting network de-
sign is, for example, an asynchronous transfer mode (ATM) design. Under
the ATM design, the network first establishes a virtual circuit between two
endpoints in the network; it then ships data along that circuit. The virtual
circuit means the network can control quality of service. But the virtual cir-
cuit also means the network is more “intelligent” than another network.5
The circuit could be programmed to be compliant with the end-to-end
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 150

1 5 0 L a w r e n c e L e s s i g

character of the original network design. But it need not be; it can be much
more (intelligent) and hence much less.
These differences make it sound as if there is a fairly technical way to
describe whether a network complies with “end-to-end” principles. They
suggest that the question of whether a network is end-to-end is simply a
question to be answered by technologists.
But let’s step back from the technical aspects and look more broadly at
the types of control there might be over a network. For if value comes from
the absence of control architected into a network, then that value may be
compromised by other techniques of control.6
The point should be obvious, but it bears emphasis. In principle, a
network could be architected such that each application must “register”
itself with the network before that application will run on the network.
A program would then send a request to the network—“May I run X?”—
and the network would give it a digital token as permission to run. Such a
network would not comply with the end-to-end argument. It would be a
control-centered network that requires permission before computer re-
sources are used. The permission this network would require is negotiated
technically. The machine does the negotiating, and if you don’t get a token,
your code doesn’t run.
Notice, however, that the very same control could be implemented
through other means. The network, for example, could have a rule—
imposed through contract—that before your computer ran any program,
you would have to register that program with the network administrator.
This rule would not be enforced through code—you could cheat and some-
times get away with it. But you might expect the network administrator to
have code to detect whether you are cheating. And if it finds that you are
cheating, it might force you off the network.
Or we might imagine a community network, where there is an under-
standing about the kinds of applications that would be run on the network
and the kinds of applications that would not be run. Roommates might have
a network, and to keep it running fast, they might have an understanding
not to use the network to download MP3s. Or better yet, the network news
protocol that enables USENET to function might include a norm that the
system would not be used to distribute commercial advertisements.
Finally, we could imagine a pricing system for controlling how a network
is used. The code could charge users based on the bandwidth used or on the
amount of time connected. This was the technique, for example, of AOL
for many years. It is the technique of many on-line service providers today.
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 151

T H E F U T U R E O F I D E A S 1 5 1

These different techniques—whether architecture, or rules, or norms, or
the market—all effect control over what gets used on a network. Control
through architecture is just one kind. And since these techniques could
overlap, a network could technically be end-to-end, at least from the per-
spective of the network architect, but because of other rules imposed
through these different techniques, it would deviate from the values pro-
tected by end-to-end. If rules, norms, or the market vested control in the
center, then the values of a decentralized, end-to-end architecture could be
lost.
In this sense, the rules governing a network, whether through laws, con-
tract, or norms, can function as a kind of intelligence in the network. This
intelligence can be advantageous or not, just as architectural intelligence
can be advantageous or not. But whether or not beneficial, my point so far
to see is the change they effect over a network where resources are free.

FA T P I P E

internet access across telephone lines is slow, even though modem
technology has improved dramatically. When I first connected with a
modem, I was happy to get a speed of 300 baud. My laptop modem now
sends and receives data at up to 56,000 bits per second. But still, that speed is
far too slow for the kinds of work people do on the network today. One can’t
surf the Internet quickly at even 56K; nor can one share large files quickly.
This limitation has pushed the market to supply faster and faster ways of
getting access to the Net. And the most important new technology for get-
ting fast access to the Net—at least in the immediate term—is “broadband”
through cable lines.7
Cable technology was developed in the 1960s as a way of giving remote
communities access to television.8 CATV stood for “community access tele-
vision,” and the very first installations simply placed an antenna on a moun-
tain and ran a cable line down to a community in a valley. When it was first
built, cable television was essentially an end-to-end system—there was little
intelligence in the network; all the power was provided by the broadcaster
and the TV; and both could be conceived to be at the ends. It was also es-
sentially one-way, analog content. Television broadcasts were piped to TVs.
There was no way for TVs to talk back.
Congress liked cable TV. The idea of spreading television to many was at-
tractive to many politicians. So in the early 1970s, Congress and the FCC
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 152

1 5 2 L a w r e n c e L e s s i g

began providing incentives for cable networks to be built. And among these
incentives was a particularly lucrative asset—monopoly control.
The argument of the cable industry in favor of monopoly was simple: We
need, they argued, incentives to risk the investment to build out cable TV.
That build-out would be worth it to us only if we could be certain to recover
our investment. This certainty would be adequately provided if we had
complete control over the programming on our network. If we get to pick
and choose the shows we run, and we get protected monopoly status in the
local markets we run cable for, then we will have sufficient incentive to
build out cable to secure our needs.
Not a bad deal, if you can get it. And even though “every major policy
study on how cable should be regulated recommended that cable operators
be required to provide at least some degree of non-discriminatory access to
unaffiliated program suppliers,”9 Congress and the FCC ignored these rec-
ommendations. Cable was given control both over the physical infrastruc-
ture that built their network and over the code layer that made their network
run.
From our perspective, however, there should be something odd about
this decision. Telephones and television were both technologies that de-
pended upon wires. Yet just as the nation was resolving to limit the control
that the network owner had over one set of wires—telephone—it was in-
creasing the control the network owner would have over a different set of
wires—cable. From our perspective, these different policies for the same
thing—wires—deserve an explanation, at least.
But at the time, telephones were as different from television as cars are
different from buggies. It was not obvious to legislators (or if it was, they
didn’t let on) why the rules governing one should also govern the other. And
even if it was obvious to some, the commercial pressure for exceptionalism
was too great to resist. Just at the time America was coming to second-guess
its first great network monopoly (telephones), it embraced and supported
the construction of a second with the potential to be just as powerful.
So cable entered its golden years, which were brightened in the late
1970s only by the innovations of Ted Turner. Turner looked at cable and
saw a waste of wires. Cable, he felt, could become a competing broadcast-
ing network, not simply the supplicant to television broadcasters. So Turner
bought access to a satellite and started broadcasting content across the satel-
lite to cable stations everywhere. Cable thus became a content provider as
well as a conduit for other people’s content.10
By the early 1990s, cable was the dominant mode of accessing television
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 153

T H E F U T U R E O F I D E A S 1 5 3

in America.11 It had gone from the farms to the centers of the largest cities.
The number of stations increased dramatically, as the technology enabled
hundreds of channels. And the range of channels exploded with the de-
crease in the number of viewers needed to make any particular channel suc-
ceed. When channels multiplied, the opportunity cost for each new channel
fell; when opportunity costs fell, then uses of the networks increased.
Cable was about to hit a number of bumps in the road, however. Some
were of its own creation—perceived “price gouging” led Congress twice to
regulate the prices of cable services. But some it did not control.12 Satel-
lite TV was the first of these; the Internet was the second.
Satellite TV offered competition to cable in the same way that cable had
offered competition to TV. Services like DirecTV provided access to many
more channels of television than cable, as well as the possibility to sell TV
on a pay-per-view basis. Yet because it used no wires, the costs of providing
this service were relatively low—at least when compared to cable. Thus,
satellite provided a great challenge to the monopoly that cable was.
To respond to this competitive threat, cable needed to upgrade its systems
to make it easier to supply two-way communication. Two-way communica-
tion was needed so consumers could make pay-per-view selections for tele-
vision; fatter pipe would make it possible for cable to provide a wider range
of content.
But while upgrading to compete with satellite, cable soon realized that it
could also upgrade to provide two-way Internet service. And if it upgraded
to provide Internet service, then cable could also be used to provide tele-
phony. Thus the upgrade could secure cable in its primary market, while so-
lidifying cable in these two new and growing markets.

AT & T C A B L E

to upgrade, however, would require a great deal of investment and,
more significantly, technological development. First, there was no standard
for enabling Internet across cable. Second, there was a great deal of poor-
quality cable that needed to be upgraded. Some was quite old. And even the
cable that was not old would require new technologies to make two-way
cable work. So the cable companies formed an independent company—
Cable Labs—to develop an open standard for serving cable. This standard—
called the DOCSIS standard—would then be usable by modem providers
that wanted to build cable modems to serve the growing Internet commu-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 154

1 5 4 L a w r e n c e L e s s i g

nity.13 And those with low-quality cable lines began replacing their lines
with newer technology.
At first, and quite slowly, a number of local cable companies began to ex-
periment with Internet access. An Internet access service provider, @Home,
and Road Runner helped the cable systems come on-line. But soon the
push for this change in technology came from an entity quite familiar with
national communications networks: AT&T.
AT&T was looking for a way to get into the Internet market. In 1995, the
Internet had just taken off; AT&T’s president, C. Michael Armstrong, de-
cided AT&T had to be a part of the future. So AT&T devised a plan
whereby it would purchase an interest in as many cable ventures as possible
and slowly combine these cable systems under a single network enabled for
broadband content. These networks, in turn, would be supported by se-
lected ISPs—either @Home or Road Runner. Thus the design AT&T envi-
sioned was of an Internet service network that would be supported by a
limited number of Internet service providers—namely, those that it would
control.
AT&T’s reasons for restricting its network to just two ISPs were many, and
over time the reasons changed. The essence of its argument was that exclu-
sive dealing with a small number of ISPs was “necessary.” At one point, they
said it was “technically necessary”—claiming that it would be technically
infeasible for AT&T to connect other Internet service providers to the
AT&T network. But later, when other cable systems demonstrated how it
might be done, AT&T claimed it was “economically necessary”—to give
it adequate incentive to develop broadband cable.14
AT&T had eaten a bunch of cable monopolies and was now beginning to
prove that you are what you eat: like the cable monopolies in the 1970s (and
like AT&T in the 1920s), AT&T claimed a protected network was needed if
broadband was to develop.

cable was not, and is not, the only broadband game in town. Wireless, as
I suggested in chapter 5, in principle could become an important competi-
tor. And in many communities, cable has a competitor serving broadband
across the telephone lines—DSL.
DSL (digital subscriber line) was developed many years ago.15 It is a way
of transmitting data over a telephone line that is also being used for voice.
The data is modulated above the frequency where voice service flows, so it
doesn’t interfere with the telephone conversation. And in tests inside DSL
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 155

T H E F U T U R E O F I D E A S 1 5 5

laboratories, there is some hope that it could transmit data at an extraordi-
narily high rate—52 megabits per second, by some estimates.16
But DSL faces many hurdles, much like the hurdles that handicap cable.
For a DSL connection to work, the copper wires in the local loop must be
reasonably clean. This requires extensive work by telephone companies to
find usable wires at the local loop. And it requires installing new routers no
more than two miles from DSL customers. The cost of this upgrade to the
copper wire world is huge, though some estimates that I have seen demon-
strate that the per customer cost is the same for cable and DSL.
DSL does not have the option, however, of running a closed network.
DSL is deployed by telephone companies. Telephone companies (by which
I mean local Bells) are regulated to be open.17 That means that telephone
companies must give ISPs the right to run their own DSL networks across
the telephone companies’ wires. And that means that the telephone com-
panies’ networks cannot exercise any real power over the kind of Internet
service made available across their wires.
It might strike you as odd that the law would require one kind of broad-
band service—DSL—to remain open to other competitors, while allowing
another broadband service—cable—to build the Internet of the future the
way cable and telephones were built in the past. Why would the govern-
ment permit control over the Internet in one case but require open compe-
tition in the other?
The answer is that there is no good reason for this inconsistency. It is
solely a product of regulatory accident. The regulations governing tele-
phones and all “telecommunications services” are found under Title II of
the Communications Act of 1934. The regulations governing cable and all
“cable services” are found under Title VI of the Communications Act.
Title II requires open access to telecommunications services; Title VI does
not. The telephone company is stuck with the position that DSL is a kind of
telecommunications service. The cable companies have vigorously argued
that broadband cable is not. And so far, though the battles are many, the law
is in favor of cable. Cable companies have been allowed to limit the range
of ISPs that use their wires, while the telephone company has been required
to permit any number of ISPs to have access to its wires.

but forget what the law is for a moment. Which should it be? Should
the lines be kept open, or should cable companies, and phone companies,
be allowed to close the lines? Should the government do nothing to protect
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 156

1 5 6 L a w r e n c e L e s s i g

openness in either case? Or should it consistently demand openness where
closed systems reign?
Well, let’s first be clear about what’s at stake. Recall what end-to-end en-
sured: that the network would remain simple, and that it would be un-
able to discriminate against content or applications it didn’t like, so that
innovations—including those the network didn’t like—would be possible
on this network. That value is threatened if end-to-end on the Internet is
compromised—either technically, by building control into the network (in
ways that will become clear later on), or effectively, by layering onto the net-
work rules or requirements that replicate this control. Whatever other
closed and proprietary networks there might be, polluting the Internet with
these systems of control is a certain way to undermine the innovation it in-
spires.
This is precisely what is happening on the cable networks right now.
While the networks are being architected to be technically consistent with
the principle of end-to-end, by requiring that everyone who gets access to
cable do so through a small number of controlled ISPs, the cable compa-
nies will reserve to themselves the power to control what access they get—
in particular, the power to decide whether some content will be favored
over other content, whether some sites surf faster, and whether certain kinds
of applications are permitted.18
And on the assumption that this control will be allowed, technology firms
such as Cisco are developing technologies to enable this control. Rather
than a neutral, nondiscriminatory Internet, they are deploying technologies
to enable the “walled garden” Internet. The network is built to prefer con-
tent and applications within the garden; access to content and applications
outside the garden is “disfavored.” “Policy-based routing” replaces the
neutral “best efforts” rule. The content favored by the policy becomes the
content that flows most easily.19
Already cable has exercised this power to decide which kinds of applica-
tions should be permitted and which kinds not. As Jerome Saltzer, one of
the coauthors of the “end-to-end” argument, describes, cable networks have
already begun to be gatekeepers on the Net. As he writes:

Here are five examples of gatekeeping that have been reported by Internet
customers of cable companies . . . :

1. Video limits. Some access providers limit the number of minutes
that a customer may use a “streaming video” connection. . . . The
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 157

T H E F U T U R E O F I D E A S 1 5 7

technical excuse for this restriction is that the provider doesn’t have
enough capacity for all customers to use streaming video at the same
time. But cable companies have a conflict of interest—they are re-
stricting a service that will someday directly compete with cable TV.

2. Server restrictions. While advertising the benefits of being “always
on” the Internet, some providers impose an “acceptable use” con-
tract that forbids customers from operating an Internet service, such
as a Web site. The technical excuse is that Web sites tend to attract
lots of traffic, and the provider doesn’t have enough capacity. But
again the access provider has a conflict of interest, because it also of-
fers a Web site hosting service. . . . (Some providers have adopted a
more subtle approach: they refuse to assign a stable Internet address
to home computers, thereby making it hard for the customer to offer
an Internet service that others can reliably find. And some access
providers have placed an artificial bottleneck on outbound data rates
to discourage people from running Internet services.)

3. Fixed backbone choice. Access providers choose where they attach
to a long distance carrier for the Internet, known as a “backbone
provider.” The route to the backbone provider and the choice of the
backbone provider are important decisions, bundled with the access
service. . . .

4. Filtering. Data is carried on the Internet in batches called packets,
and every Internet packet contains an identifier that gives a rough in-
dication of what this packet is for: e-mail, a Web page, a name
lookup, a remote login, or file sharing. Several access providers have
begun to examine every packet that they carry, and discard those
with certain purposes, particularly those used for file sharing. The
technical excuse for this filtering is that many users don’t realize that
their computer allows sharing of files, and filtering prevents other
customers from misusing that feature. But some access providers
have imposed filtering on every customer, including those who want
to share files. . . . And again, there can be a conflict of interest—the
access provider has an incentive to find a technical or political ex-
cuse to filter out services that compete with the entertainment or
Internet services it also offers.

5. No home network. An increasing number of homes have two or
more computers interconnected by a home network, and as time
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 158

1 5 8 L a w r e n c e L e s s i g

goes on we are likely to find that this home network connects tele-
vision sets, household appliances, and many other things. Some ac-
cess providers have suggested that they aren’t technically prepared to
attach home networks, but the technology for doing it was developed
in the 1970’s. In refusing to attach home networks, providers are ac-
tually protecting their ability to assign the network address of the cus-
tomer. By refusing to carry traffic to Internet addresses they didn’t
assign, the access provider can prevent the customer from con-
tracting for simultaneous service with any other Internet access
provider.20

The most telling of these limits is video. Cable companies make a lot of
money streaming video to television sets. The Internet, in the view of some,
could become a competitor to cable, by streaming video to computers.
Under @Home rules, users were not permitted to stream more than ten
minutes of video to their computers.21 And though AT&T offers congestion
as a reason for this limitation, at times it is a bit more forthcoming. As AT&T
executive Daniel Somers is reported to have said, when asked whether
AT&T would permit the streaming of video to computers, “[W]e didn’t
spend $56 billion on a cable network to have the blood sucked out of our
veins.”22
Cable’s intent to exercise control is clear; it has already exercised control.
And if the business model that Cisco sells is as attractive as Cisco sells
it to be, then we should expect that cable will continue to exercise control
in the future. It will architect and enforce a network where the kinds of uses
and content that run on the network are as the network chooses—which is
to say, it will build a network just the opposite of the network the Internet
originally was.
The evidence of this intent to discriminate was strongest at AT&T’s
@Home media. As François Bar reports, “[T]he @Home 1998 annual re-
port is very clear” on the strategy of discrimination.23 It proposed to steer its
customers, unknowingly, toward merchants that partnered with @Home. It
would do this through code and marketing—through placement of ads, as
well as through “how do I” wizards that would direct customers to selected
sites. Their reports “explain how they will provide superior quality perfor-
mance to partnering merchants.”24 In this respect, Bar argues, “@Home is
acting very much like Microsoft, using its control of the operating system’s
architecture to favor some applications over others.”25 This closed-access
control would allow cable owners to pursue only the exploration and devel-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 159

T H E F U T U R E O F I D E A S 1 5 9

opment of new technologies that directly benefit them. “This is not to say
that no innovation will take place,” Bar argues, “simply that only the tech-
nology trajectories that line up with their interest will be pursued.”26

one choice about these trajectories is particularly important to high-
light. Recall from chapter 8 my description of the emerging technologies of
peer-to-peer. These technologies presume “peers”—that is, machines that
are roughly equal. And though connection speeds on narrowband connec-
tions were slow, they were equal upstream and downstream.
Not so with the emerging technologies of broadband. Most of these tech-
nologies are faster downstream than they are upstream. Most broadcast
more quickly than they receive.
Given the way the Internet is used right now, this imbalance makes good
sense. E-mail and Web clicks going up take far less bandwidth than stream-
ing video going down. Hence this structure makes sense of the uses of
today—it is optimized, that is, for the uses of today.
But as with any optimization, what’s good for today is not necessarily good
for tomorrow. More important, how we optimize the network today will af-
fect what good is possible tomorrow. Thus, as we optimize the network for
this broadcasting mode, it becomes harder for the peer-to-peer structures to
evolve. A world where users are servers doesn’t scale well when the connec-
tion to the Internet is biased in favor of servers at the center.
Ordinarily, we don’t have to worry much about this sort of thing in
advance. The original PC market didn’t care much about design; Apple
Computer then changed that preference. The early PCs didn’t have much
capacity for sound. Later innovation created an incentive for the early PCs
to change. The reason in both cases is the power of the market: as long as
the market is free and competitive, these new uses will evolve as consumers
want them.
In the context of broadband, however, there may well be a reason to be
more skeptical about the market. If the concentrations in ownership con-
tinue as they have in the past few years (recently encouraged by an FCC
that wants to take a more hands-off approach), then at a certain point there
may be a strategic reason for these networks to resist the peer-to-peer way. By
architecting networks to enable peer-to-peer, broadband providers will be
reducing the power they have to direct users as they wish.
Here’s an analogy that might suggest the point: Imagine you’re a cable
company serving twenty channels in your market. A new technology comes
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 160

1 6 0 L a w r e n c e L e s s i g

along that would open up two hundred more channels on your cable sys-
tem. This technology is, let’s assume, relatively cheap, and with it you could
be certain to increase your communications capability ten times over.
But the catch is that the content on these two hundred channels will be
provided by your customers. Would you, as the cable company, adopt this
technology?
The answer is that it depends. If the cable company could charge its cus-
tomers differently because they used these different channels—if they could
make up for the loss they suffered because fewer were watching the twenty
channels by charging something for the use of the two hundred channels—
then the cable company in principle should have no problem with the new
technology. Prices would be adjusted to assure that the revenues the cable
company received were as high as they were before.
But there are three problems with this happy assumption as applied to the
Net. First, customers don’t notice that they are living within a closed system.
If a travel site comes up slowly because it is not a favored site, the user is
likely to consider this congestion, not something owing to the network. Sec-
ond, the business model of some networks is based on “owning” the cus-
tomer, not on charges for access. The last thing these business models can
accept is an architecture that opens more channels. But third, and more im-
portant, even if there’s a price at which the cable company would be willing
to allow this new innovation, that price may be too high to inspire investment
in this new form of innovation. The innovation that gets devoted to a free,
neutral platform is different from the innovation that gets devoted to a plat-
form where the platform owner can, down the road, simply change its mind.
We’ve seen this lesson before, and we’re at a point where we can state a
general claim: Where a disruptive technology emerges, there may be good
reason not to extend the power of existing interests into power over that
technology. That doesn’t mean the new technology should be allowed to
defeat the old or, at least, defeat the old for free. For example, no doubt the
customers who use the “two hundred new channels” technology on
the cable system should have to pay something for this new capacity. But the
price they pay should not necessarily be within the control of the dinosaurs.
Instead, while compensation is justified, control is not required. Or, better,
separating control from compensation may well be a way to induce more in-
novation.

* * *
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 161

T H E F U T U R E O F I D E A S 1 6 1

my claim so far can be summarized like this: When the Internet was first
born, both norms (among core network facilities) and law (restricting the
telephone company) effected an end-to-end environment. This created the
initial neutral platform. But as the Internet moves to broadband platforms,
neither norms nor the law require network providers to preserve the same
innovation environment. The trend instead is toward control—toward lay-
ering onto the original code layer of the Internet new technologies that fa-
cilitate greater discrimination, and hence control, over the content and
applications that can run on the Internet. The Net is thereby moving from
the principle of end-to-end that defined its birth to something very differ-
ent.27

many resist the view that this control is anything to worry about. Cable
is just one of a number of broadband technologies. DSL, as I have noted, is
not free to be closed. In many contexts, DSL competes with cable. Hence if
consumers value openness, then they can choose DSL over cable. And if
they choose DSL because DSL is open, then cable will be pressed to be
open as well.28 Thus competitive forces will force the network to open up,
even if cable desires to be closed.
I agree with this claim, as far as it goes. To the extent consumers prefer
open to closed, they will put pressure on the closed providers to be open.
But to move from that claim to the conclusion that therefore there is noth-
ing to worry about is, in my view, premature. There are plenty of reasons to
worry that the closed character of cable won’t correct itself.
First, there is the issue of numbers, and numbers of two sorts—the num-
ber of people on cable broadband, and the number and character of other
broadband providers. Cable now has a great lead over DSL in subscribers to
the cable system. There were 5 million cable broadband customers and
1.8 million DSL customers in 2000.29 Some predictions suggest that DSL
may close in on cable by 2002,30 especially in nonresidential areas where
cable does not now exist. But there are just as many who predict that cable
will continue to lead.
But second, there is the number of different broadband providers and the
character of their business models. For cable is not the only closed system.
Wireless—the great hope from chapter 5—is being deployed now in a way
that is primarily closed. The architecture for wireless broadband uses the
same specifications that cable does—DOCSIS.31 DOCSIS, as you’ll re-
less_0375505784_4p_01_r1.qxd 9/21/01 13:49 Page 162

1 6 2 L a w r e n c e L e s s i g

member, doesn’t yet provide for simple open access. Thus the future is not
cable vs. DSL, but DSL vs. a scad of providers, all of which are closed.
Third, while DSL will be a strong competitor to simple Internet access,
cable access provides the opportunity to mix television content with Inter-
net content. And while DSL presses its limits to serve Internet as fast as
cable does, cable has a great deal of bandwidth that it can use to supply
Internet-related content. Right now cable provides 10 percent of its band-
width to be used for Internet service. It could easily multiply the number of
channels supplying Internet service and become a much more attractive op-
tion. Thus, while openness might be on DSL’s side, the value of openness
to the consumer may be outweighed by the ability to bundle cable more ef-
fectively with other video content.
But fourth, and most important, consumers’ preferences might not be
enough to motivate the market. This is the point we have seen both in chap-
ter 3’s discussion of the value of e2e and in chapter 4’s discussion of neutral
platforms: A closed network creates an externality on innovation generally.
It increases the cost of innovation by increasing the range of actors that
must license any new innovation. That cost is not borne directly by the con-
sumer. In the long run, of course, if it is a cost, it is borne by the consumer.
But in the short run, the consumer doesn’t notice the innovation that the
closed model chills. Thus the consumer does not completely internalize
the costs imposed by a closed system. And hence the pressure the consumer
puts on closed systems to open themselves up is not equal to the costs that
such closed systems impose on innovation generally.
These are good reasons, I believe, for being skeptical about whether the
invisible hand will solve the problem of closed networks. The observation
that never in the history of telecommunications has a network voluntarily
been opened after being closed is another reason to be skeptical. Finally,
the interest of those who own these networks to keep control within the net-
work is huge, and a huge reason to be skeptical about their control.
To see this part of the story, however, we need to shift to a different battle
about open architectures—AOL.

america online was born far from the Internet. Its birth was as an on-
line service that gave members access to other members. While other ser-
vices were focused on how to sell product, AOL understood from the very
beginning that networks are built by communities.
AOL’s community was built by making computers easy and access simple.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 163

T H E F U T U R E O F I D E A S 1 6 3

The company “carpet-bombed” America with AOL disks; it made sign-up
simple and access cheap. Quickly AOL built a following that was extraordi-
nary for on-line services.32
The AOL network was not really end-to-end. Lots of intelligence was
built into the software with which one connected to AOL. AOL made that
intelligence work to assure ease of access as well as control where control
was needed. The service held the user’s hand, but it required some intelli-
gence to know where the hands were. It was a preprogrammed world, which
users took as they found it. No one built additions to AOL or added func-
tionality to AOL without AOL’s permission.
That wasn’t the case with all on-line communities. MUDs (multi-user do-
mains), for example, were on-line communities where people were free to
develop new parts of the on-line, virtual, text-based world.33 If you wanted to
add a room to an existing MUD, you simply wrote the code to add the room
and submitted it. The space that got built was as the members built it.
In AOL, the only building was that approved by the town planner—AOL.
And AOL succeeded in building an extraordinarily popular place.
When the Internet came along, many thought AOL would die. Why pay
to get access to preselected content when you could get access much more
cheaply to the Internet as a whole? But AOL responded to this challenge by
doing what it does best: by building its service to make it easy for users to
find their way onto the Internet. The Internet was one place AOL users
could go, but then there was also the content on AOL. Both would be avail-
able to AOL customers; only the Internet was available to others.
AOL then became another Internet service provider, but with something
extra that came from the content it served. It was an ISP plus, because it also
had its own content. But many simply used the service to get easy access to
the Internet. And AOL then was subject to the fierce competition that every
ISP faced. With some five thousand ISPs across America, there was only so
much power any one ISP had—even if that ISP had a very large number of
customers.

AOL was built on narrowband telephone lines. When broadband came
along, AOL faced a critical threat. If broadband service was reserved to just
two ISPs, and if it was far superior to the service one could get across the
telephone lines, then AOL faced a great challenge from this emerging
Internet opportunity. If AOL was barred from broadband, then AOL would
be history.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 164

1 6 4 L a w r e n c e L e s s i g

AOL thus joined many who were pushing the FCC as well as local gov-
ernments to require that broadband cable lines be kept open for competi-
tion. This was the “open access” movement; AOL was a key player. In 1999,
AOL argued to the city of San Francisco during its open access implemen-
tation hearings:

AOL applauds the City for taking this critical step in the implementation
of the Board of Supervisors’ open access resolution, which wisely supports
consumers’ freedom to choose their Internet service provider and to access
any content they desire—unimpeded by the cable operator.34

AOL had made the same arguments in favor of governmental interven-
tion to the FCC.35 In this campaign, AOL’s allies were many. Indeed, before
AT&T started buying cable lines, AT&T too was an ally. In Canada, AT&T
argued to the Canadian government that access to cable in Canada should
be regulated to be open.

AT&T Canada LDS submits that the application of the Commission’s for-
bearance test to the two separate markets for broadband access and infor-
mation services supports a finding that there is insufficient competition in
the market for broadband access services and the market for information
services to warrant forbearance at this time from the regulation of services
when they are provided by broadcast carriers. As noted above, these carri-
ers have the ability to exercise market power by controlling access to
bottleneck facilities required by other service providers. It would appear,
therefore, that if these services were deregulated at this time, it would
likely impair the development of competition in this market as well as in
upstream markets for which such services are essential inputs.36

Vertically integrated cable and telephone facility owners, AT&T had ar-
gued, possessed market power and had to be prevented from engaging in
anticompetitive practices.37
But when AT&T bought its own cable lines, its story changed. No longer
did it believe that cable should be regulated. Instead, AT&T began to argue
that the market should regulate cable, and the government should stand
aside.
This would become a familiar pattern.
In January 2000, AOL and Time Warner announced to a startled world
that they had agreed to merge. Time Warner owned many cable companies;
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 165

T H E F U T U R E O F I D E A S 1 6 5

these cable companies would serve AOL content at high speed. AOL had
many Internet customers. These customers would be able to get access to
Time Warner content. The merger was an ideal opportunity, both compa-
nies argued, for synergy in this market. The old and the new would form to-
gether one of the most important media companies in the world.
At the same time, AOL announced its policy on open access had
changed. It, like AT&T, no longer believed that the government should reg-
ulate access. It, like AT&T, believed that the market should regulate itself.38

i’m not sure why people are surprised by flips in corporate policy, any
more than we are surprised by flips of politicians. Corporations have a duty
to their shareholders. Their job is to make money. If the opportunities pre-
sent themselves, they will, and should, change their views. They are not in-
stitutions of public policy. And they don’t deserve the attack that would
befall an institution of public policy that so radically, and transparently,
switched sides.
But the other side of this obvious point is that we should not treat
what corporations say is good public policy as what is good public pol-
icy. We should treat them as statements by individuals who are required
by law to be self-serving. This is not just “bias”—this is legally mandated
bias.
Thus, I discount both AOL’s support and AOL’s opposition to govern-
ment regulations to support open access as evidence about whether open
access is good policy. The question is not what AOL believes is good for
AOL. The question is what is good for the Internet.
And here again we return to the question, What trend should we expect?
The opponents of any governmental role here argue that the market will
take care of itself. I think that’s true—the market will take care of itself.
AOL/TW will build itself to maximize its market power. The question is
what shape that building will take.

the danger is what economists would call the problems of vertical
integration—where one provider controls the full range of services across
the layers I described—content, logical, and physical.39 Outside the Inter-
net, the danger of vertical integration is less.40 But within a network, the
danger grows. Such integration, a report by the National Research Council
has concluded, “could, if successful, cause a change in the Internet market,
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 166

1 6 6 L a w r e n c e L e s s i g

with innovation and creativity becoming more the province of vertically in-
tegrated corporations.”41 It would, Web founder Tim Berners-Lee worries,
be dangerous for innovation generally. “Keeping the medium and the con-
tent separate,” Berners-Lee writes, “is a good rule in most media. When I
turn on the television, I don’t expect it to deliberately jump to a particular
channel, or to give a better picture when I choose a channel that has the
‘right’ commercials. I expect my television to be an impartial box. I also ex-
pect the same neutrality of software.”42
The danger with the AOL–Time Warner merger is the danger that this
vertical integration will induce AOL/TW to engage in discrimination—
both discrimination in conduits (favoring their own lines over others) and
discrimination in content (favoring their own content over others).
This danger is real. As economists Daniel Rubinfeld and Hal Singer have
concluded, given the existing concentration in cable broadband, AOL/TW
will have a significant incentive to engage in both forms of discrimination.43
And by mid-2001, AOL Time Warner had begun to prohibit advertisements
on their sites for competing Internet access providers.44 Discrimination was
threatened; discrimination is being realized.

as the Clinton administration came to an end, one of the last acts of the
(by statute, at least, neutral) Federal Trade Commission (FTC) was a sign of
some hope. After a long and extensive investigation into the risks of the pro-
posed AOL–Time Warner merger, the FTC, led by its chairman, Robert
Pitofsky, conditioned the merger of AOL and Time Warner upon the es-
sential elements of open access. Access to the cable broadband pipes must
be kept open, the FTC insisted. Nonaffiliated content must flow without
hindrance from AOL or Time Warner. And this unhindered access must in-
clude access to Internet-active TV.45
This decision by the FTC was an important breakthrough in the attitude
of the government. Until this point, the government’s view had been that
the market here had to take care of itself. The problem, as the increasing
mergers and restrictive access conditions demonstrated, was that the market
was taking care of itself. The market was building a protection into the ar-
chitecture that could change the commons for innovation dramatically.
But this decision is just a first round. (And as I describe in chapter 11, this
first round may well be overturned by the courts.) As cable gets built out, as
an administration emerges that is more open to allowing the market rather
than rules to regulate, as other modes of broadband are built, the constant
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 167

T H E F U T U R E O F I D E A S 1 6 7

pressure will be to allow this founding principle of neutrality to itself be neu-
tralized.
And then the question becomes this: If the original Internet architected
an innovation space that was free, if it built that space by creating an envi-
ronment where innovations would not be checked, if it was defined by a
code layer that, in Benkler’s terms, was open, then as the Internet moves
onto fat pipes, will the same principle govern the code layer of the Net? Will
broadband respect the principle of end-to-end as narrowband has? And if it
doesn’t, will the government do anything to resist the change?

what’s at stake here are two models for organizing a communications
network, and the choice for us is which model will prevail. On the one
hand, there is the model of the perfectly controlled cable provider—owning
and controlling the physical, logical, and content layers of its network. On
the other hand, there is the model of the Internet—which exerts no control
over a physical layer beyond the decision to include equipment or not, and
which enables the free exchange of content over a code layer that remains
open.
As the Internet moves from the telephone wires to cable, which model
should govern? When you buy a book from Amazon.com, you don’t expect
AOL to demand a cut. When you run a search at Yahoo!, you don’t ex-
pect your MSN network to slow down anti-Microsoft sites. You don’t expect
that because the norm of neutrality on the Internet is so strong. Providers
provide access to a network that is neutral. That’s the essence of what the
Internet means.
But the same neutrality does not guide our thinking about cable. If the
cable companies prefer some content over others, that’s the natural image
of a cable provider. If your provider declines to show certain stations, that’s
the sort of freedom we imagine it should have. Discrimination and choice
are at the core of what a cable monopoly does; neutrality here seems silly.
So which model should govern when the Internet moves to cable? Free-
dom or control?

not every increase in control violates the principle of end-to-end. Obvi-
ously the ends are free, as far as this principle is concerned, to do what they
want with their machines, and while some would resist calling the cable
networks an “end,” they could well argue that they are just a private network
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 168

1 6 8 L a w r e n c e L e s s i g

connected to the Internet. To link to the network is not to commit your hard
disk to the use of anyone. The physical layer remains controlled, even if the
code layer is free.
Here we see the source of the compromise that this chapter is all about.
For in an important sense, the cable network is simply asserting the same
rights with “its” equipment that I assert over my machine when connected
to the Internet. My machine is mine; I’m not required to make it open to
the world. To the extent I leave it open, good for the world. But nothing
compels me to support it.
Leaving the ends free to choose, then, creates an opportunity for them to
choose control where the norm of the Internet has been freedom. And con-
trol will be exercised when control is in the interest of the ends. When it
benefits the ends to restrict access, when it benefits the ends to discriminate,
then the ends will restrict and discriminate regardless of the effect on others.
Here, then, we have the beginnings of a classic “tragedy of the com-
mons.”46 For if keeping the network as a commons provides a benefit to all,
yet closing individual links in the network provides a benefit to individuals,
then by the logic that Garrett Hardin describes in chapter 2 above, we
should expect the network “naturally” to slide from dot.commons to
dot.control. We should expect these private incentives for control to dis-
place the public benefit of neutrality.47
The closing of the network by the cable companies at the code layer is
one example of this slide. If DSL providers were given the choice, they too
would do the same. Wireless providers are implementing essentially the
same sort of control. AOL Time Warner is insisting that code using its net-
work be code that it controls.
In all these cases, the pressure to exert control is strong; each step makes
sense for each company. The effect on innovation is nowhere reckoned.
The value of the innovation commons that dot.commons produces is whit-
tled away as the dot.coms rebuild the assumptions of the original Net.
Consider another example of this tragedy in play:
The World Wide Web is crawling with spiders. These spiders capture
content and carry it back to a home site. The most common kind of spider
is one that indexes the contents of a site. The spider will come to a Web
page, index the words on that Web page, and then follow the links on the
Web page to other sites. And by following this process as far as the links go,
these spiders index the Web.
This index, then, is what you use when you run a search on the Web.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 169

T H E F U T U R E O F I D E A S 1 6 9

There are many Web search engines, each with a slightly different tech-
nique. But they all rely upon the ability to spider the Web and gather the
data the Web makes available.
These “spiders” are also called “bots.” A bot is simply a computer pro-
gram that runs remotely on another machine. Searching is just one exam-
ple of the kinds of things computer “bots” do to one another on the Web.
Some of those other things are awful: “denial of service attack” is an event
where either one or a number of coordinating computers sends repeated re-
quests to a Web page, ultimately overwhelming the server for that page. But
in the main, these “things computers do to each other” have been produc-
tive and extraordinarily creative.
One example of this creativity comes in the context of auction sites. Auc-
tion sites make products available to real-time, wide-scale auctions. eBay is
the most famous, but not the only one. eBay opened in 1995 as a place
where individuals could offer their stuff in an auction to others. The
idea caught on, and competing sites started offering the same service.
Amazon.com has its own auction site, as does Yahoo!.
But then customers interested in auctions faced another “metaproblem.” If
they had things they were watching on many different sites, they had the has-
sle of traipsing through all those sites to find what they wanted to watch. So
where there was a problem, the market quickly provided a response. Bidder’s
Edge, among others, began to offer a site that did the surfing for you. On one
page you could see the status of all your auctions. And Bidder’s Edge prom-
ised to update this information regularly.
In each case, the innovation is the same. The Web is an open architec-
ture; it begs for people to discover new ways to combine the resources it
makes available. In each of these cases, someone did discover a new way of
combining resources. And this discovery then produced a new kind of mar-
ket. Search engines were a defining feature of the original World Wide
Web. And the opportunity to quickly compare prices was one of the early
promises for competition on the Web.
But in each case, too, there is this undeniable fact: When a search engine
spiders the Web, it uses resources of others to build its index. When Best
Book Buys enters Amazon.com, it collects the price Amazon offers by using
Amazon’s servers. In a sense, then, we could say that each of these bots tres-
passes on the servers of other sites.
To many, this idea of trespassing bots will seem bizarre. But it did not
seem too bizarre to the lawyers at eBay. For eBay didn’t want bots that cre-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 170

1 7 0 L a w r e n c e L e s s i g

ated competitors to eBay. And it had imposed a NO-BOT policy on access
to its Web site. That is, it indicated in the code of its site that it did not want
unlicensed bots to enter its site.
Bidder’s Edge ignored that sign. It continued to gather data even though
the eBay lawyers told it not to. And eventually it found itself in court, in a
lawsuit brought by eBay charging Bidder’s Edge with “trespass.”
In one sense, of course, the lawsuit was completely right. In the virtual
sense in which one “goes” to a Web site, Bidder’s Edge’s bot was “entering”
a computer without the permission of its owner. And “entering” without
permission is the classic definition of “trespass.”
But in another sense, the claim seemed bizarre. The Web was built on a
norm of open access; this was a community that kept its doors unlocked. No
one forced eBay to open itself to the World Wide Web. But if it did, it should
live by the norm. And if the norm was openness, then it was eBay that com-
mitted the offense.
Both sides brought in lawyers to argue their respective points of view. On
the side of eBay was an outspoken, and famous, law professor from the Uni-
versity of Chicago—Richard Epstein. Epstein pushed the law-focused an-
swer: Trespass law made perfect sense in the Internet context. Indeed, it
made more sense here than in real space. It was simple to establish signs
that stated the conditions under which entry was permitted; those signs
could be easily read by bots. If a site wanted to restrict access to all save those
who pay, then that was perfectly permissible, Epstein argued. The site, after
all, “owned” the equipment. Control over the property one owns is perfectly
ordinary.48
On the other side was a lawyer who was a bit more careful with the legal
tradition. Law professor Dan Burk argued that the law had been strict only
when it came to “land.” Other property was protected against unauthorized
use. But that protection was not absolute. To support a lawsuit based on tres-
pass to property other than land, the plaintiff would have to demonstrate
some sort of harm. But in the Bidder’s Edge case, no harm had been
pleaded. Thus, under traditional trespass doctrine, eBay should lose.
Both sides had a point, and while my bias is with Burk, I don’t mean to
deny the plausibility of a different regime. What I do deny, however, is that
the answer to this question is obvious. What is most damaging about the
submission made by Epstein is its obliviousness to any issue on the other
side.
For no doubt we could move to a world where every use of data on the
Web had to be licensed. We could generalize from the control everyone has
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 171

T H E F U T U R E O F I D E A S 1 7 1

over “his” machine to the power to deny the neutrality of the network gen-
erally.
But there are costs to that world. Closing access based on this argument
grounded in the physical layer of the Net increases the costs of innovation
for the Net generally. If to deploy a technology the innovator must first li-
cense its use, this legal requirement then functions as a kind of tax on inno-
vation on the Net.
This is especially true with bot technologies. These devices—the next
generation after the spiders that gather data from the Net—would enable
agent-driven, fluid marketplaces on the Net. These bots could search out
prices, negotiate contracts, and schedule delivery in a way that is far more
efficient than any of the existing markets.49
The response to this is that we don’t want rules that force people to devote
their resources to something they don’t want to support. Bidder’s Edge didn’t
pay the servers that it used when it linked to eBay’s data. Why should eBay
be forced to subsidize a competitor?
But this story could be told both ways around. eBay benefits greatly from
a network that is open and where access is free. It is this general feature of
the Net that makes the Net so valuable to users and a source of great inno-
vation. And to the extent that individual sites begin to impose their own
rules of exclusion, the value of the network as a network declines. If ma-
chines must negotiate before entering any individual site, then the costs of
using the network climb.
As I said at the start, this closing of sites selectively changes the character
of the Net, but not necessarily its compliance with end-to-end. As it in-
creases, however, it does change the commons of the Internet into some-
thing different. To the extent that this ability—to select the uses that access
to the Net permits—grows, then this permission changes the character of
the commons the Internet creates.

this discrimination is growing in other contexts as well. Sometimes it
happens for innocent reasons, sometimes less innocently. An innocent case
is the emergence of a technology called “network address technologies”
(NATs). NATs are devices for multiplying IP addresses. Every machine on
the Internet needs a unique IP address—that’s how the Net knows where to
send the packets. But NATs make it so many machines can share the same
IP address.
NATs were created initially because of an expected shortage of IP ad-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 172

1 7 2 L a w r e n c e L e s s i g

dresses. The technology has subsequently grown simply because of the dif-
ficulty in coordinating devices in many contexts. Apple, for example, uses
NATs to connect machines to its AirPort wireless server. You can plug an
AirPort into your cable or DSL modem and then an unspecified number of
machines can share the very same IP address.
The problem with NATs is that the techniques used to share IP addresses
are not standard. The NAT inserts points of control into the network. Data
passing onto a NAT-controlled network must pass through the NAT before
the NAT permits it to pass to the end user. If the NAT is unaware of how to
process the data from that particular application (either because the NAT
was unaware of that application or because it was coded to ignore data of
that type), then that application won’t function on that NAT-empowered
network. Developers of technologies that need to be certain they are talking
to a particular machine must therefore survey the world of NATs to make
certain their systems will work on all the major brands. This in turn in-
creases the costs of development and, on the margin, may reduce innova-
tion.
No one thinks NAT boxes are part of a conspiracy. This compromise of
end-to-end is innocent in the sense that we don’t imagine it is implemented
for strategic purposes. Nonetheless, it reduces the flexibility of the Internet
as a whole.
But there is a solution to the problem that NATs were initially designed
to solve—and again, it is to increase capacity. The name space for the Inter-
net (IPv4) is in the process of being upgraded (to IPv6). That will have a
practically endless number of addresses,50 thereby eliminating the need
for NATs. With endless address space, technologies for “conserving” ad-
dresses become unnecessary at best. Thus, rather than imposing this high-
coordination cost on technologists developing technology for the Net,
increasing the name space would remove the initial reason for the compro-
mise.
Other compromises with end-to-end are less benign. Consider firewalls,
for example. A firewall is a technology for controlling interaction between a
local network and the Internet. Like the NAT, it is a technology that adds a
point of control within the network that could block everything that has not
explicitly been admitted by the local network manager. Unforeseen appli-
cations thus again pay a heavy price.
Firewall technology, for example, no doubt serves a legitimate purpose in
many cases. Sometimes, however, its purpose is expressly to impose a policy
on the Net. Many universities, for example, forbid the use of Napster tech-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 173

T H E F U T U R E O F I D E A S 1 7 3

nologies. They enforce this ban by telling their firewalls to block Napster
content. This in turn produces something of an arms race, as developers
shift their systems to channels that will never be filtered by a firewall. But
that shift will only make it harder to use those channels in different applica-
tions efficiently.51
Here too there is a solution that could solve the problem that firewalls an-
swer, but without compromising end-to-end. A technology called IPSec
could enable better control over access consistent with end-to-end.52
In each of these cases, then, there are two issues at stake. As technologies
for facilitating discrimination increase, one question is where these tech-
nologies get located in the Net—on the Net or at the edge. A second ques-
tion is the effect such discrimination will have, even if it is located at the
ends. The end-to-end principle counsels that we locate such discrimination
at the ends rather than in the network; but even when it is located at the
ends, a widespread pattern of certain types of discrimination could weaken
the commons the network now provides.
There are reasons not to worry so much about this kind of discrimination.
Where concentration is slight and many different services are available, the
risk that any particular concentration will harm innovation is slight as well.
Some ends may be Christian Right; as long as they don’t interfere with ac-
cess to the Christian Left, innovation for the Christian Left will not be
harmed. The key is to preserve user autonomy; the danger is a technology
that might undermine autonomy.
The danger is discrimination engaged in by concentrated actors. Here
again we return to the story of concentrating cable. For if we were in a world
where there was significant competition in broadband services, with many
different suppliers each essentially open—and hence, each not discriminat-
ing in the kind of access that is provided—then the danger from closed ac-
cess in one channel would be greatly reduced. The value of the commons
in the highway is not lost simply because some roads become private. But
when there isn’t a great deal of competition in access, when a small number
of companies can set the rules for the whole system, then the dangers in dis-
crimination return. When a few can make decisions about what kinds of
innovation will be permitted, the innovation promised by an end-to-end ar-
chitecture is lost.
The danger of the changes that I have described in this chapter is that just
this concentration is occurring. And the dangers in this concentration in-
clude the fear that an opportunity for innovation will be lost. We will have
used architecture and rules to shift control over how the network can be
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 174

1 7 4 L a w r e n c e L e s s i g

used, from the many ends that constituted the Internet originally, to a few
that own the wires. Control will have been returned to this medium born
free.

there is another side to the stories I have told. Not every increase in con-
trol is driven by a desire to lessen competition; not every increase will have
the effect of undermining innovation.
Indeed, some increase in control may well be necessary if investment to
build a network is to proceed. Just as cable companies argued initially that
control over their cable lines was essential if there were to be a sufficient re-
turn from laying cable, so too cable companies today may rightly argue that
control is needed if the return is to be enough.
The cable companies may be right. And striking a monopoly deal with a
provider is a strategy that governments have employed since the start of gov-
ernments. My argument cannot begin to resolve the question of whether or
not the cable companies are right in their defense. If this infrastructure is to
be built without public support, then protected monopoly may well be nec-
essary.
My argument is meant simply to highlight a cost that may well run with
a benefit. The Internet is not a community antenna. It is not simply a system
for delivering a given kind of content more efficiently. The critical feature
of the Internet that sets it apart from every other network before it is that it
could be a platform upon which a whole world of activity might be built.
The Internet is not a fancy cable television system; the Internet is the high-
way system, or the system of public roads, carrying bits rather than trucks,
but carrying them in ways no one can predict.
When the United States built its highway system, we might have imag-
ined that rather than fund the highways through public resources, the gov-
ernment might have turned to Detroit and said, Build it as you wish, and we
will protect your right to build it to benefit you. We might then imagine
roads over which only American cars can run efficiently, or exits and en-
trances that tilt against anything built outside Detroit. Or we could imagine
Detroit then auctioning rights to use its network to the highest bidder, or ex-
cluding Coke trucks because of an exclusive contract with Pepsi.
This power in Detroit might well have been necessary if Detroit were to
have had sufficient incentive to build the highways. But it does not follow
that Detroit should be given this power. For however much the state may
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 175

T H E F U T U R E O F I D E A S 1 7 5

gain by not having to fund roads on its own, society would lose in the ag-
gregate if the open commons of transportation were lost.
That loss is even more pronounced in the context of the Internet. Roads
have many uses, but “many” is still not infinite. Any kind of commerce gets
to use the roads: trucks as well as VW bugs; campers as well as pickups. But
the physical nature of roads limits the possible “many” uses. Lots are possi-
ble, but “the possible” is constrained.
The constraints on the Internet—properly architected—are far fewer.
The range of uses is far less constrained. The Internet could be a platform
for innovation across the full range of social and political life. Its possible
uses are, even this far into its growth, unknowable.
We may gain something by giving network owners power over the net-
work. I don’t question that. But we will lose something as well. To the extent
we chill innovation that threatens disruption, disruption will be slower
in coming. That slowness is a cost that society must account for. We may
gain something from the “free” infrastructure monopoly builds. But we lose
something with the “controlled” infrastructure that monopoly inevitably
wants.
Even more significant, we have no good way to make sure that the gains
outweigh the losses. To the extent that the code layer builds an innovation
commons, changes at the code layer threaten to exhaust that commons.
Changes imposed by broadband providers weaken the value of the neutral
platform; changes effected through NATs or firewalls similarly weaken the
innovation potential of the Net. All these changes are effected locally, but
they also have a global effect. Each may make sense locally, but there’s no
obvious way to be certain that their effect globally will also make sense.
In this way, changes at the code layer create their own tragedy of the in-
novation commons. As we might paraphrase Hardin:

Therein is the tragedy. Each [firm] is locked into a system that compels [it]
to increase [its control] without limit—in a world that is limited. Ruin is
the destination toward which all [firms] rush, each pursuing [its] own best
interest in a society that believes in the freedom of the commons. Freedom in
a commons brings ruin to all.53

“Ruin” is a strong word, I’ll concede. But the dynamic is the same
nonetheless: the incentive is for companies to layer control onto the Net;
that has been the history of the past five years. But the effect of that incen-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 176

1 7 6 L a w r e n c e L e s s i g

tive is felt by the Net as a whole. Yet its effect on the innovation commons
is almost completely ignored.
Against this trend, some rightly argue that government has a role to play
to assure that ISPs continue to offer “open IP service.” As a recent National
Research Council report put it:

[C]oncerns about the vertical integration of the data transport and content
businesses and about content control, as seen in recent debates about ac-
cess to cable broadband Internet systems, could be eased if ISPs com-
mitted to providing their customers with open IP service. From this
standpoint, the continued delivery of open IP service would be an en-
lightened move in the long-term interest of the industry.54

Just the sort of wisdom that finds its way into NRC reports and is ignored
almost everywhere else.

the change that is happening in the context of wires has a particular
form. We are in the midst of a radical change in technology; that change
threatens existing interests; those interests have an interest in minimizing
the threat that this change presents; they can minimize that threat by
reestablishing choke points on the system that emerges. They can, in the
words of Gerald Faulhaber, use the architecture to regain strategic control.55
This is precisely the change that is happening. As Charles Platt put it in a
recent article in Wired, “Everyone knows that the broadband era will breed
a new generation of online services, but this is only half the story. Like any
innovation, broadband will inflict major changes on its environment. It will
destroy, once and for all, the egalitarian vision of the Internet.”56
Dinosaurs should die. This lesson we have learned over and over again.
And innovators should resist efforts by dinosaurs to keep control. Not be-
cause dinosaurs are evil; not because they can’t change; but because the
greatest innovation will come from those outside these old institutions.
Whatever the scientists at Bell Labs understood, AT&T didn’t get it. Some
may offer a theory to explain why AT&T wouldn’t get it. But this is a point
most understand without needing to invoke a fancy theory.
Because the Internet is inherently mixed—because it is a commons built
upon a layer that is controlled—this tension between the free and the con-
trolled is perpetual. The need for balance is likewise perpetual. But the
value of balance is not always seen. This value we need to keep in focus.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 177

C o n t r o l l i n g t h e W i r e d

( a n d H e n c e t h e C o n t e n t L a y e r )

I n the last chapter, I argued that there is a tension between control at the
physical layer and freedom at the code layer, and that this tension affects
the incentives for innovation. The original freedom built a commons; more
control can undermine that commons; the tragedy is our forgetting the
value of the free in our race to perfect control.
The same tension exists at the content layer. Some content the law treats
as “owned”—copyright and patents are “intellectual property,” owned by in-
dividuals and corporations. Other content can’t be owned—either content
that has fallen into the public domain or content that is outside the scope of
Congress’s power under the copyright and patent clause of the Constitu-
tion. Here, too, balance is important. Yet here, too, the owned chases out
the unowned. The pressure to protect the controlled is increasingly under-
mining the scope for the free.
My aim in this chapter is to describe this dynamic and to suggest how
changes that we are seeing right now will affect this dynamic. By the time
this book is published, I fear the struggle I am describing will be finished.
The courts will have resolved these questions, and the politicians will have
no courage to interfere with this resolve. Already the endgame is clear; al-
ready property has queered the balance. Hence, already the value of this
freedom will have been lost.

this chapter is meant to mirror chapter 4, “Commons Among the
Wired.” Yet it is not directly about the people I spoke of in chapter 4. The
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 178

1 7 8 L a w r e n c e L e s s i g

“wired” who are affected by the changes I am describing here are not ex-
actly the same “wired” who built the open source and free software move-
ments that I spoke about there.
But in a critical sense, they are the same. Both innovate by building on
the content that has gone before. Both therefore reveal how much creativity
depends upon the creativity that has gone before. Both show, that is, inno-
vation as adding something to the work of others.
In some cases, the restrictions I describe in this chapter apply directly to
the innovators of chapter 4. Patent law, for example, poses one of the most
significant threats to the open code movement that there is. But in general,
the changes I describe in this chapter are aimed at controlling a new gen-
eration of “wired” folks—those who see the platform of the Internet as an
opportunity for a different way of producing and distributing content and
those who see the content on the Net as a resource for making better and
different content. The changes in this chapter are changes that reestablish
control over this class of potentially wired souls.

when the Net emerged into the popular press, there was an anxiety
among many about what the Net would make possible. People could do
things there that we had discouraged or made illegal here.
Pornography was the most dramatic example of this anxiety. The freedom
of the Net meant, the world quickly learned, the freedom of anyone—
regardless of age—to read the obscene. The news was filled with instances
of kids getting access to material deemed “harmful to minors.” The demand
of many was that Congress do something to respond.
In 1996, Congress did respond, by passing the Communications De-
cency Act (CDA).1 Its aim was to protect children from “indecent content”
in cyberspace. The act was stupidly drafted, practically impaling itself upon
the First Amendment, but its aim was nothing new. Laws have long been
used to protect children from material deemed “harmful to minors.” Con-
gress was attempting to extend that protection here.
Congress failed. It failed because the CDA was overbroad, regulating
speech that could not be regulated constitutionally. And it failed because it
had not properly considered the burden this regulation would impose upon
activity in cyberspace. The statute required adult IDs before adult content
could be made available. But to require sites to keep and run ID machines
was to burden Internet speech too severely. Congress would have to guar-
antee that the burden it was imposing on the Internet generally was no
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 179

T H E F U T U R E O F I D E A S 1 7 9

greater than necessary to advance its legitimate state interest—protecting
children.
In 1998, Congress tried again. This time it focused on clearly regulable
speech—speech that was “harmful to minors.” And it was much more for-
giving about the technology that would permissibly block kids from “harm-
ful to minors” speech. Still, federal courts struck down the law on the
ground that the burden it would impose on the Internet generally was just
too great.2
These cases evince a distinctive attitude. Though the state’s interest in
protecting children is compelling, courts have insisted that this compelling
state interest be pursued with care. In effect, a demonstration that the regu-
lation won’t harm the Net too broadly is required before this state interest
can be promoted. Facts, and patient review, are the rule in this area of the
law of cyberspace.
Keep this picture in mind as we work through the examples that follow.
For the meaning of Reno v. ACLU is not that porn is okay for kids or that the
state’s interest in enabling parents to protect their kids from porn is out-
dated. The Court in Reno was quite explicit: Protecting children from
speech harmful to minors is a “compelling” state interest. But this com-
pelling interest must be advanced in ways that are consistent with the other
free speech values. The state was free to advance its compelling state inter-
est; but it was required, in so doing, not to kill the rest of the Net.

about the same time that parents were panicking about porn on the Net,
copyright holders were panicking about copyright on the Net. Just as par-
ents worried that there was no way to keep control over their kids, copyright
holders worried that there was no way to keep control over copyrighted con-
tent. The same features of the Internet that made it hard to keep kids from
porn also made it hard to keep copyrights under control.
Both forms of panicking were premature. While it is true that the Net as
it was originally built made it hard to control content (by either keeping it
from kids or keeping it from being copied by kids), the Net as it was origi-
nally built is not the Net as it must be. Code made the Net as it was; that
code could change. And the real issue for policy makers should be whether
we can expect code to be developed that would solve this problem of con-
trol.
In Code I argued that in the context of copyright, we should certainly ex-
pect such code to be developed.3 And if it were developed as its architects
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 180

1 8 0 L a w r e n c e L e s s i g

described, then the real danger, I argued, is not that copyrighted material
would be uncontrolled: the real danger is that copyrighted material would
be too perfectly controlled. That the technologies that were possible and
that were being deployed would give content owners more control over
copyrighted material than the law of copyright ever intended.
This is precisely what we have seen in the past two years, but with a twist
that I never expected. Content providers have been eager to deploy code to
protect content; that much I and others expected. But now, not only Con-
gress but also the courts have been doubly eager to back up their protections
with law.
This part I didn’t predict. And indeed, in light of Reno v. ACLU, one
would be justified in not predicting it. If parents must go slowly before de-
manding that the law protect their kids, why would we expect Hollywood to
get expedited service?
The answer to that question is best left until after we have surveyed the
field. So consider the work of the courts, legislatures, and code writers in
their crusade to expand the protections for a kind of “property” called IP.

INCREASING CONTROL

in dorm rooms around the country, there are taped copies of old LPs.
Taped to the windows, there are posters of rock stars. Books borrowed from
friends are on the shelves in some of these rooms. Photocopies of class ma-
terial, or chapters from assigned texts, are strewn across the floor. In some of
these rooms, fans live; they have lyrics to favorite songs scribbled on
notepads; they may have pictures of favorite cartoon characters pinned to
the wall. Their computer may have icons based on characters from The
Simpsons.
The content in these dorm rooms is being used without direct compen-
sation to the original creator. No doubt, no permission was granted for the
taping of the LPs. Posters displayed to the public are not displayed with the
permission of the poster producers. Books may have been purchased, but
there was no contract forbidding passing them to other friends. Photocopy-
ing goes on without anyone knowing what gets copied. The lyrics from
songs copied down from a recording are not copied with the permission of
the original author. Cartoon characters, the exclusive right of their authors,
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 181

T H E F U T U R E O F I D E A S 1 8 1

are not copied and posted, on walls or on computer desktops, with the per-
mission of anyone.
All these uses occur without the express permission of the copyright
holder. They are unlicensed and uncompensated ways in which copy-
righted works get used.
Not all of these uses are impermissible uses. Many are protected by ex-
ceptions built into the Copyright Act. When you buy a book, you are free to
loan it to someone else. You are free to copy a small section of the book and
give it to a friend. Under the Audio Home Recording Act, you are free to
copy music from one medium to another. Taped recordings of records are
therefore quite legal.
But some of these uses of copyrighted works may well be illegal. To post
the poster may be a public display of the poster not authorized by the pur-
chase.4 To use icons on your computer of Simpsons cartoons is said by Fox
to violate its rights. And if too much of an assigned text has simply been
copied by the student, then that copying may well exceed the scope of “fair
use.”
The reality of dorm rooms, however—and, for that matter, most private
space in real space—is that these violations, if they are violations, don’t mat-
ter much. Whether or not the law technically gives a student the right to
have a Simpsons cartoon on his desktop, there is no practical way for Fox
Broadcasting Company to enforce its rights against overeager fans. The fric-
tion of real space sets the law of real space. And that friction means that for
most of these “violations,” there is no meaningful violation at all.
Now imagine all this activity moved to cyberspace. Rather than a dorm
room, imagine that a student builds a home page. Rather than taped LPs,
imagine he produces MP3 translations of the original records. The Simp-
sons cartoon is no longer just on his desktop; imagine it is also on his Web
server. And likewise with the poster: the rock star, we can imagine, is now
scanned into an image file and introduces this student’s Web page.
How have things changed?
Well, in one sense, one might say the change is quite dramatic. Now,
rather than simply posting this content to a few friends who might pass
through the dorm room, this student is making this content available to mil-
lions around the world. After all, pages on the World Wide Web are avail-
able anywhere in the world. Millions use the World Wide Web. Millions
can now, for free, download the content that this student posted.
But there’s a gap in this logic. There are millions who use the World
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 182

1 8 2 L a w r e n c e L e s s i g

Wide Web. But there are billions of Web pages. The chances that anyone
will stumble across this student’s page are quite slight. Search engines bal-
ance this point, though that depends upon what’s on a particular page. Most
Web pages are not even seen by the author’s mother. The World Wide Web
has amazing potential for publishing; but a potential is not a million-hit site.
Thus, in reality, this page is effectively the same as the student’s dorm
room. Probably more people view the poster on the dorm room window
than will wade through the student’s Web page. In terms of exposure, then,
moving to cyberspace doesn’t change much.
But in terms of the capacity for monitoring the use of this copyrighted
material, the change in the move from real space to cyberspace is quite sig-
nificant. The dorm room in cyberspace is subject to a kind of monitoring
that the dorm room in real space is not. Bots, or computer programs, can
scan the Web and find content that the bot author wants to flag. The bot au-
thor can then collect links to that content and follow through however it
seems most sensible.
Consider the story of fans of The Simpsons who find themselves sum-
moned to court when their Simpsons fan pages are discovered by a bot hired
by the television network Fox. The fans are not allowed, Fox said, to collect
friends and strangers around these images of Bart Simpson and his dad.
These images are “owned” by Fox, and Fox has the right to exercise perfect
control.5 Though “[t]he sites are the Internet equivalent of taping posters of
favorite actors to a bedroom wall,”6 they are not permitted by copyright law.
Fan sites are not the only examples here. Dunkin’ Donuts used the threat
of a copyright lawsuit to force a site devoted to criticism of the nationwide
chain to sell the site to the company. The company claimed it could “more
effectively capture the comments and inquiries” if it owned the site.7
Maybe, but it is also certainly true that it could more effectively edit the
content the site made public.
A more telling example is the history of OLGA—an on-line guitar
archive started by James Bender at the University of Nevada, Las Vegas. As
the Web site describes it:

OLGA is a library of files that show you how to play songs on guitar. The
files come from other Internet guitar enthusiasts like yourself, who took
the time to write down chords or tablature and send them to the archive or
to the newsgroups rec.music.makers.guitar.tablature and alt.guitar.tab.
Since they come from amateur contributors, the files vary greatly in
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 183

T H E F U T U R E O F I D E A S 1 8 3

quality, but they should all give you somewhere to start in trying to play
your favorite tunes.8

In 1996, the University of Nevada, Las Vegas, was contacted by EMI Pub-
lishing, which alleged that the site violated EMI’s copyright. The university
shut the site down. The then-current archivist, cathal woods, moved the
archive to another host. Then in 1998, OLGA was contacted again, this
time by the Harry Fox Agency, which, like EMI, complained of copyright
violations without specifying precisely what was being infringed. OLGA
closed the archive in that year and then began a long (and as yet unresolved)
campaign to establish the right of hobbyists to exchange chord sequences.
The pattern here is extremely common. Copyright holders vaguely allege
copyright violations; a hosting site, fearing liability and seeking safe harbor,
immediately shuts down the site. The examples could be multiplied thou-
sands of times over, and only then would you begin to have a sense of the
regime of control that is slowly emerging over content posted by ordinary in-
dividuals in cyberspace. Yahoo!, MSN, and AOL have whole departments
devoted to the task of taking down “copyrighted” content from any Web site,
however popular, simply because the copyright holder demands it.9 Ma-
chines find this content; ISPs are ordered to remove it; fearing liability, and
encouraged by a federal law that gives them immunity if they remove the
content quickly,10 they move quickly to take down the content.
This is the second side of the effect that cyberspace will have on copy-
right. Copyright interests obsess about the ability for content to be “stolen”;
but we must also keep in view the potential for use to be more perfectly con-
trolled. And the pattern so far has tracked that potential. Increasingly, as ac-
tivity that would be permitted in real space (either because the law protects
it or because the costs of tracking it are too high) moves to cyberspace, con-
trol over that activity has increased.
This is not a picture of copyrights imperfectly protected; this is a picture
of copyright control out of control. As millions move their life to cyberspace,
the power of copyright owners to monitor and police the use of “their” con-
tent only increases. This increase, in turn, benefits the copyright holders,
but with what benefit to society and with what cost to ordinary users? Is it
progress if every use must be licensed? If control is maximized?
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 184

1 8 4 L a w r e n c e L e s s i g

CPHack

there’s lots of junk on the World Wide Web. And there’s lots that’s
worse than junk. Some of the stuff, for some people, is offensive or worse.
The worse includes material deemed obscene or, and this is a very different
category, “harmful to minors”—aka pornography.
As I’ve described, there’s a long and tedious history of Congress’s efforts to
regulate porn in cyberspace.11 I’m not interested in that story here. I’m in-
terested here in the efforts of companies to regulate porn in cyberspace by
producing code that filters content.
The code I mean is referred to affectionately as “censorware.” Censor-
ware is a class of technology intended to block access to Internet content by
forbidding a Web browser to link to the blocked sites. Censorware compa-
nies make it their job to skim the Web looking for content that is objection-
able, and they then add the link to that content to their list. Their list of
banned books is then sold to parents who want to protect their kids.
There is obviously nothing wrong with parents exercising judgment over
what their kids get to see. And obviously, if the choice is no Internet or a fil-
tered Internet, it is better that kids have access to the Internet.
But this does not mean that censorware is untroubling. For often the sites
blocked by censorware systems are themselves completely unobjectionable.
Worse, sites often are blocked merely because they oppose the technology
of censorware. In December 2000, free speech activists at the civil rights
group Peacefire reported that a number of censorware systems had begun
to block Web sites affiliated with Amnesty International.12 This is just the
latest in an endless series of similar cases. They all point to a technology that
is fundamentally at odds with the openness and free access of the original
Net.
In 1999, Eddy Jansson of Sweden and Matthew Skala of Canada decided
they wanted to test out one instance of censorware—a product called Cyber
Patrol. They therefore wrote a program, CPHack, with which a user could
disable Cyber Patrol and then see which sites Cyber Patrol banned. The
code thus made it easier, for example, for a number of sites to complain
about the censorious practices of Cyber Patrol.
The owner of Cyber Patrol was not happy about CPHack. So like most
owners unhappy with what others do, it raced into federal court. In March,
Mattel brought suit against the authors and Peacefire, demanding it stop dis-
tributing its code for liberating the CP list.
Its claim was copyright violation. These coders, Mattel argued, had
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 185

T H E F U T U R E O F I D E A S 1 8 5

violated Mattel’s copyright by reverse engineering the code for Cyber Patrol—
contrary to the license under which Cyber Patrol was sold. Because their
use of Cyber Patrol was unlicensed, it was illegal.
There is something very odd about the claim that Mattel was making.
Copyright’s core is to protect authors from the theft of others. It is to protect
Mattel, in other words, from someone who would steal Cyber Patrol and use
it without paying for the program. Copyright is not ordinarily aimed at pro-
tecting authors from criticism. It doesn’t “promote progress” to forbid criti-
cism of what has happened before. But this is exactly how the law was being
used in this case. By claiming that a contract that was attached to the copy-
righted code banned a user from criticizing the code, the law was being
used to restrict criticism.
Within two weeks, Mattel had received a worldwide injunction against
the distribution of CPHack.13 The injunction was not just against the au-
thors of the program; it also extended to those who linked to the program’s
site or who merely posted the program. These secondary posters believed
they had a fairly strong right to post the code for CPHack. The code stated
it was “GPL’ed,” which meant that anyone was free to take it and post it as
he wished. But all these “conspirators” (as the law had to call them to justify
this extraordinary federal action) were now bound by this emergency in-
junction of a U.S. court. And Mattel then moved quickly to perfect and
make permanent this force of law.
Yet here, cracks in the case began to show. First there was the problem of
jurisdiction. The authors of CPHack were not citizens of the United States,
and their work was not done in the United States. Copyright law, in the
main, is national. Just because these two people somewhere in the world did
something that would constitute a violation of copyright law in the United
States does not show they violated United States copyright law.14
But even if there had been jurisdiction, there was a much more funda-
mental flaw. What exactly was the wrong that these defendants were said to
have committed? Mattel said they had “reverse engineered” Cyber Patrol.
Reverse engineering is ordinarily a permissible “fair use” under copyright
law—copyright law has no incentive to make it impossibly difficult for
others to compete with software programs.15 But, Mattel said, the license
that Cyber Patrol was sold under did not give the purchasers any right to re-
verse engineer. Indeed, it expressly waived the right of the purchaser to
reverse engineer the product.
The contract Mattel was speaking of was the sort of shrink-wrapped li-
cense that comes with most software today. When you install Cyber Patrol,
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 186

1 8 6 L a w r e n c e L e s s i g

you are said to have agreed with everything on that license. Now whether
such a license in general is enforceable is a hard question. The strongest
case in the United States supporting its enforcement is a decision by Judge
Frank Easterbrook in the Seventh Circuit Court of Appeals. But Easter-
brook is clear that the restrictions beyond copyright law depend upon there
being a contract. As he said, “Someone who found a copy of [a copyrighted
work] on the street would not be affected by the shrink-wrap license—
though the federal copyright laws of their own force would limit the finder’s
ability to copy or transmit the application program.”16 Thus, to demonstrate
that the authors violated the law, you would have to demonstrate they had
purchased the product in a way that would have made them liable under
the contract.
All that was going to be very hard to prove. But just at the moment the
case was to come to trial, Mattel had a surprise. It had purchased the rights
to CPHack from the original authors, and now it was simply enforcing the
rights it was purchasing. No one, Mattel said, was free to distribute this
code, because this code was now Mattel’s.
There was a squabble at this point about whether in fact the code was
Mattel’s. The code had been distributed in a form that indicated it was gov-
erned by the GPL. The GPL made it impossible to sell the product in a way
that would revoke that license—at least to those down the chain of distribu-
tion. The original sellers—who received nothing except the promise that
this gaggle of American lawyers would go home—were quick then to deny
that they had released the program under the GPL. But that denial rang hol-
low. The Mattel lawyers had apparently informed them that if Mattel had
been tricked, they would be guilty of fraud. And while that would have been
an idle threat (at least if the authors had simply agreed to transfer whatever
rights they had), it was apparently threat enough to get the authors to deny
that CPHack was in fact under the GPL.
Armed with this purchase, Mattel was able to convert the temporary in-
junction into something permanent. And the judge forbade others who had
apparently been restricted by the injunction from intervening to challenge
the injunction. As the case settled, and was affirmed by a court of appeals in
Boston, Mattel had the rights to CPHack; no one else could distribute it,
even if the purpose was simply to criticize the company, Mattel.
The first two centuries of copyright’s history were two centuries of cen-
sorship.17 Copyright was the censor’s tool: the only things that could be
printed were those things printed by authorized presses; the only authorized
presses were those cooperating with the Crown.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 187

T H E F U T U R E O F I D E A S 1 8 7

Here history has repeated itself, though the protected is not the Crown,
but commerce. The law has become a tool for effectively disabling the
ability of others to criticize a corporation. Coders can release code that cen-
sors the Net, and efforts to release the list of censors are censored by the law.

DeCSS

the lawyers for Mattel relied directly upon copyright law. But there was
another tack they might have taken—one that will prove much more im-
portant as time goes on.
In 1998, Congress passed the Digital Millennium Copyright Act
(DMCA).18 That act strengthened copyright in a number of ways, but one
way was particularly troubling. This was its “anticircumvention” provision.
The anticircumvention provision regulates code that cracks code that is
intended to protect copyrighted material. There are two parts to the
provision—one that restricts the cracking of code that protects copyrighted
material, and one that forbids the creation of code that cracks code that pro-
tects copyrighted material. In both cases, the aim of the law is to lend legal
support to the tools that copyright holders deploy to protect their copy-
righted material.19
In the ordinary case—with ordinary property—there can be little in this
to complain about. It is a crime to steal my car. But obviously, that isn’t
enough to stop car theft. So many people install a burglar alarm in their car
to further inhibit car theft. But obviously again, that too isn’t enough. So if
a legislature, wanting to reduce the risk of theft even more, passes a law that
makes it a crime to disable burglar alarms, or to sell tools whose sole purpose
is to disable burglar alarms, there can’t be any complaint about these rules,
either. If it is wrong to steal a car, and permissible for people to protect their
property, it is wrong to crack technology designed to protect the property.
But this story about real property doesn’t map directly onto intellectual
property. For as I have described, intellectual property is a balanced form
of property protection. I don’t have the right to fair use of your car; I do have
the right to fair use of your book. Your right to your car is perpetual; your
right to a copyright is for a limited term. The law protecting my copyright
protects it in a more limited way than the law protecting my car.
This limitation is not just laziness on the part of Congress. The limits on
the law’s power to protect copyright are inherent in the clause granting
Congress power to regulate copyright, and in the First Amendment’s re-
strictions on Congress’s power. Copyright law, for example, cannot protect
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 188

1 8 8 L a w r e n c e L e s s i g

ideas; it can protect only expression. The law’s protection can extend only
for limited times. And fair use of copyrighted works is understood to be con-
stitutionally required.
These limitations distinguish copyright as property from ordinary prop-
erty. And that distinction suggests the trouble with direct analogy from laws
protecting burglar alarms to laws protecting code protecting copyrighted
work. If copyright law must protect fair use—meaning the law cannot pro-
tect copyrighted material without leaving space for fair use—then laws
protecting code protecting copyrighted material should also leave room for
fair use. You can’t do indirectly (protect fair-use-denying-code protecting
copyright) what you can’t do directly (protect copyright without protecting
fair use).
I am not arguing that it is illegal or somehow unconstitutional for indi-
viduals to deploy code that protects copyrighted material more than the law
does. There are troubles with this, and I don’t think the law can ignore
them. But there is ordinarily no constitutional problem unless the law has
actually done something.
But in the case I’ve described, the law has done something. The anti-
circumvention provision is law that protects code that protects copyrighted
material. And my claim is simply that that law must be subject to the same
limitations that a law protecting copyrighted material directly is.
How does all this relate to Mattel?
Well, Mattel released a product that was copyrighted. Arguably, at least,
its compilation of sites is copyrighted. It protected this copyrighted material
using code. This code is what CPHack hacked. Thus, arguably, CPHack
violated the anticircumvention provision of the DMCA.
Mattel didn’t bring this case, though I wish it had. Had it claimed the
anticircumvention provision protected it, then the courts would have had a
clear shot at the question of whether or not there are constitutional limita-
tions on the power of Congress to protect code protecting copyright. The
Cyber Patrol case would have been a perfect case to raise that claim. If
cracking code to demonstrate that the code is censoring speech isn’t fair
use, then I’m not sure what would be.
Instead, this question of fair use was raised in a very different case.
In 1994, Hollywood started releasing movies on DVD disks. These
movies were extremely high fidelity and relatively compact. The disks fit in
an ordinary CD-ROM-size drive. And very quickly, manufacturers started
producing drives that would read DVD disks.
To protect the movies on these disks, the industry developed an encryp-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 189

T H E F U T U R E O F I D E A S 1 8 9

tion system. This system was named CSS—Content Scramble System. CSS
would make it difficult for a user to play back DVD content unless the user
was using a machine that could properly decode the CSS routines.
The machines were DVD players that had been licensed to decrypt
CSS-encrypted content. These licenses were issued by the consortium that
developed and deployed CSS. And they were granted initially to compa-
nies that produced Windows- and Macintosh-compatible machines. Those
running Windows, or those using a Mac, could play DVD movies on their
machines.
Let’s be clear first about what CSS did. CSS was not like those early soft-
ware protection systems. It didn’t interfere with the ability to copy DVD
disks. If you wanted to pirate a DVD disk, all you needed to do was copy the
contents from one disk to another. There was no need to decrypt the system
in order to copy it.
So CSS didn’t disable copying. All it did was limit the range of machines
that DVD disks could be played on. And that in turn was the limitation that
gave rise to the need for a crack.
For—surprise, surprise!—Macintosh and Windows are not the only oper-
ating systems out there. In addition, there are Linux PCs, among others.
These machines could not play DVD movies. And owners of these ma-
chines were not happy about this limitation. So a number of them decided
to develop a program that would crack CSS, so that DVDs could be played
on other machines. And when open source coders developed such a pro-
gram, they called it DeCSS.
DeCSS disabled the encryption system on a DVD disk. It turned out that
CSS itself was a terribly poor encryption technology. And once the system
had been cracked, it became possible to play DVD content on other com-
puters. With DeCSS, DVD disks could be played on any machine.
Now again, DeCSS didn’t make it any easier to copy DVDs than before.
There’s no reason you can’t simply copy a CSS-protected movie and ship it
to your friends. All that CSS did was ensure that you played the movie on a
properly licensed machine. Thus, DeCSS didn’t increase the likelihood of
piracy.20 All that DeCSS did was (1) reveal how bad an existing encryption
system was; and (2) enable disks presumptively legally purchased to be
played on Linux (and other) computers.
But upon the release of DeCSS, the industry went nuts. Within six weeks,
four lawsuits had been filed in four separate jurisdictions, seeking under
many legal theories the quashing of this code.21 Within three weeks of the
filing of the suits, two injunctions had been entered against people who
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 190

1 9 0 L a w r e n c e L e s s i g

posted DeCSS code and even against journalists who linked to DeCSS.22
Once again, as with CPHack, the legal system had been fired up to silence
this dangerous code.
The core case here was tried in New York. The defendants were many.
Some had linked to the sites carrying DeCSS. Others had written articles
about the sites and had linked to the links. And others were active distribu-
tors of DeCSS. None of these defendants was in the business of selling pi-
rated movies. And at no time in the case did the plaintiffs demonstrate that
any movies had been pirated because of DeCSS.
Instead, the sole claim in the case was that these defendants were in the
business of distributing code that cracked an encryption system, and hence,
these defendants were in violation of the anticircumvention provisions of
the DMCA.
The district court judge in the New York case issued an immediate in-
junction stopping the distribution of DeCSS. After a long trial, he issued an
opinion making permanent that injunction. The opinion making the in-
junction permanent rejected the argument that “fair use” entitled the de-
fendants to produce or distribute this code. Fair use, the court concluded,
was something copyright law must allow. This was a law regulating code,
not a copyright. The court concluded that Congress had the power to allow
private actors to pile on protection on top of the copyright law. No First
Amendment interests were violated.
This case was appealed to the circuit court. At the time of this writing,
that appeal has not been resolved. But the importance of the case is not how
it ends; the importance is the signal that Hollywood sends: any system that
threatens its control will be threatened with an army of Hollywood law-
yers.

iCraveTV

iCraveTV was a site that streamed television content over the Internet.23
The site was located in Canada, where Canadian broadcasting law made
such streaming legal. Under Canadian law, anyone has the right to re-
broadcast television content, as long as he doesn’t change the content in
any way. iCraveTV wanted to take advantage of that right to give computer
users access to TV.
The problem was that though TV was free in Canada, it was not free in
the United States. To rebroadcast content in the United States requires the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 191

T H E F U T U R E O F I D E A S 1 9 1

permission of the original broadcaster. So behavior legal in Canada would
be illegal in the United States.
But then where was iCraveTV? In one obvious sense, it was in Canada.
But when it made itself available on the Internet, it was also, simultane-
ously, everywhere. That has been the character of the Internet since its
birth—to be on the site at any place is to be on the site in every place.
iCraveTV took some steps to limit itself to one place. It tried to block
non-Canadians from the site. But when it began this process, the technolo-
gies for blocking were not strong. iCraveTV asked for a telephone number,
but of course it had no easy way to verify that the telephone number you
gave it was your telephone number.
Soon after iCraveTV went on-line, copyright holders in the United States
brought suit to shut it down. The theory? By setting up an Internet service
to broadcast TV, iCraveTV was broadcasting TV into the United States. It
was therefore violating U.S. copyright law (by “publicly performing” what
iCraveTV streamed to American viewers). Until it could “guarantee,” as the
Hollywood lawyers put it, that no United States citizen would get access to
this free Canadian TV, the Canadian site had to be shut down.
There was a significant dispute about how hard iCraveTV was working to
keep non-Canadians out of its site. The Hollywood lawyers hired Harvard
Law School Berkman Center’s boy genius Ben Edelman to demonstrate
just how easy it was to hack the iCraveTV site. But whether easy or not, the
significant issue about the case is this: How much should someone in one
country have to be burdened by the laws of another country?
For example: Imagine the Chinese government telling the American site
China Online24 that it must shut down until it is able to block out all Chi-
nese citizens, since the content on China Online is illegal in China. Or
imagine a German court telling Amazon.com that it must stop its selling of
Mein Kampf until it can guarantee that no German citizen will be able to
get access to that book—since that book is illegal in Germany. Or imagine
a French court telling Yahoo! that it has to block French citizens from pur-
chasing Nazi paraphernalia, since that is illegal in France. (Oops, no need
to imagine. A French court did just this.25)
In all these cases, we are likely to think that the action of these foreign
governments is somehow illicit. That the free exchange of the Net tilts us in
favor of open and regular access. That steps to shut down foreign sites be-
cause of local laws are the very essence of what the Internet was designed to
avoid.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 192

1 9 2 L a w r e n c e L e s s i g

But when it comes to copyright law, we become like the Chinese, or Ger-
mans, or French. With respect to law, we too want to insist upon local
control—especially because local law here is so strong. So with respect to
copyright law, we push local control. And the result is the birth of tech-
nologies that will facilitate better local control.
iCraveTV, for example, promised the court that it would develop tech-
nology to make it possible to block out everyone except Canadians. Jack
Goldsmith and Alan Sykes have described the growing collection of tech-
nologies that will achieve the same end.26 These suggest that the future will
be very much like the past: life on the future Internet will be regulated lo-
cally, just as life before the Internet was regulated locally.
How we will get to that future world was one point of Code and Other
Laws of Cyberspace. But for now, the significance of iCraveTV is again the
attitude it evinces. Though there was no proof that any revenue would be
lost by virtue of people streaming content through their TV, and though
Canadian law was assumed to protect this behavior in Canada, the control
industry raced to court to shut down the alternative. The courts complied.27

MP3

in chapter 8, I told the story of My.MP3—an innovative new service
whose users could “beam” the content of their CD collection to a Web site
and then get access to their music at that Web site. This service was provided
by the company MP3.com. To provide access to this music, MP3.com had
to purchase a very large collection of CDs. It then copied those CDs into its
computer database. When a user of My.MP3 placed a CD into the Beam-it
program, the system identified whether that CD was in MP3.com’s library.
If it was, then that user account got access to the content of that CD when-
ever he or she accessed the account.
Ten days after launching the service, MP3.com received a letter from
RIAA attorneys.28 Its service was a “blatant” violation of copyright laws, said
the letter, and MP3.com should take the service down immediately.
MP3.com refused, and the lawyers did what lawyers do when someone re-
fuses: they filed suit in U.S. district court, asking for over $100 million in
damages.29
The RIAA lawyers had a point, if you looked at the statute quite literally.
MP3.com may have purchased a bunch of CDs, but it had clearly “copied”
these CDs when it created its single, massive database. There was, on its
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 193

T H E F U T U R E O F I D E A S 1 9 3

face, then, an unauthorized copy of each of these CDs, and the question be-
came whether or not this copy was nonetheless fair use.
Applying the ordinary standard for fair use, the RIAA argued that it was
clearly not. This was for a commercial purpose. Thus, fair use was not a de-
fense, and the blatant and willful copying was then a prosecutable offense.
When lawyers have such a clean, slam-dunk case, they get very, very sure
of themselves. And the papers in the My.MP3 case are filled with outrage
and certainty.
But when you stand back from the outrage and ask, “What’s really go-
ing on here?,” this case looks a lot different. First, as should be clear,
My.MP3 was not facilitating the theft of any music. You had to insert a real
CD into your computer before you could get access to the copy on
MP3.com’s server. Of course, you could borrow someone else’s CD and
hence trick the system into thinking you were the rightful owner of the CD.
But you could borrow someone else’s CD and copy it anyway. The existing
system permits theft; My.MP3 didn’t add to that.
Second, it should be fairly clear that this service would increase the value
of any given CD. Using this technology, a consumer could listen to his or
her CD in many different places. Once the system recognized your rights to
the music on the CD, the system gave you those rights whenever you were
at a browser. That means that the same piece of plastic is now more valu-
able. That increase in value should only increase the number of CDs that
are purchased. And that increase would benefit the sellers of CDs.
Third, it is also fairly clear that exactly the sort of thing that MP3.com was
doing could easily have been done by the consumers themselves. Any num-
ber of companies have created free disk space on the Internet. Anyone
could “rip” his or her CDs and then post them to this site. This ripped con-
tent could then be downloaded from any computer. And this download
could be “streamed” to be just like the service MP3.com was providing.
The difference is simply that users don’t have to upload their CDs. On a
slow connection, that could take hours; on a fast connection, it still can be
quite tedious. And a second difference is that the duplication that would be
necessary for everyone to have his or her CDs on-line would be much less.
Ironically, by shutting down MP3.com, the RIAA was inducing the produc-
tion of many more copies of the very same music.
Thus the battle here was between two ways of viewing the law—one very
strict and formal and the other much more sensitive to the consequences of
one outcome over the other. And the claim of MP3.com was simply that the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 194

1 9 4 L a w r e n c e L e s s i g

court should consider the facts in the case before it shut down this inno-
vative structure for distributing content. MP3.com was arguing for a right
to “space-shift” content, so that a user’s content could be accessible any-
where.
But the court had no patience for MP3.com’s innovation. In a stunning
decision, the court not only found MP3.com guilty of copyright violation, it
also found the violation “willful.” And rather than giving nominal or mini-
mal damages for this violation, the court imposed $110 million in damages.
For experimenting with a different way to give consumers access to their
data, MP3.com was severely punished.

Napster

i described the technology that is Napster in chapter 8. The essence was
this: Napster enables individuals to identify and transfer music from other
individuals. It enables peers, that is, to get music from peers. It does this not
through a completely peer-to-peer architecture—there is a centralized
database of who has what, and who, at any particular moment, is on-line.
But the effect is peer-to-peer. Once the service identifies that X has the song
that Y wants, it transfers control to the clients of X and Y, and these clients
oversee the transfer. The Napster server has just made the link.30
But that was enough in the eyes of the recording industry. And with pre-
dictably lightning speed, it filed suit here as well. Napster was just a system
for stealing copyrighted material. It should, the RIAA demanded, be shut
down.
Against the background of MP3.com, Napster does look a bit dicey. After
all, the service at issue in MP3.com was a service to give individuals access
to content that they presumptively had purchased. On Napster, the pre-
sumption is the opposite. There seems little reason for me to download
music I already own.
But even that is not quite correct. I’ve been a Napster user, though I am
not an imaginative user, and I am generally quite lazy. I know exactly what
I want to hear, and I know that because I own the music already. But it is
easier simply to download and play the music I own on Napster than it is for
me to go through the CDs I own (most of which are at home, anyway) and
insert the one I want in a player. Thus, while I won’t say that none of the
music I have listened to on Napster is music I don’t own, probably only
5 percent is.
That the user owned the music, however, didn’t stop the court in the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 195

T H E F U T U R E O F I D E A S 1 9 5

MP3.com case. And the assurance that users were only downloading music
they already owned was not likely to satisfy the RIAA. Most people, the
RIAA argued, used Napster’s technology to “steal” copyrighted work. It was
a technology designed to enable stealing; it should be banned like burglar’s
tools.
Copyright law is not new to a technology said to be designed solely to fa-
cilitate theft. Think of the VCR. The VCR records content from television
sets. It is designed to record content from television sets. The designers
could well have chosen to disable the record button when the input was
from a TV. They could, that is, have permitted recording when the input
was from a camera and not a TV. But instead, they designed it so that tele-
vision content could be copied for free.
No one in the television industry gave individuals the right to copy tele-
vision content. The television industry instead insisted that copying
television content was a crime. The industry launched a massive legal ac-
tion against producers of VCRs, claiming that it was a technology de-
signed to enable stealing and that it should be banned like burglar’s tools. As
Motion Picture Association of America president Jack Valenti testified, the
VCR was the “Boston Strangler” of the American film industry.31
This legal campaign ended up in the courtroom of Judge Warren Fer-
guson.32 After “three years of litigation, five weeks of trial and careful con-
sideration of extensive briefing by both sides,”33 the trial court judge found
that the use of VCRs should be considered “fair use” under the copyright
act. The court of appeals quickly reversed, but the important work had been
done in the trial court. The judge had listened to the facts. Sony was per-
mitted weeks of testimony to demonstrate that, in fact, the VCR would not
harm the industry. Sony was permitted, in other words, to show how this
technology should be influenced by the law.
These findings were critical in the appellate review of the case. And when
the case finally reached the Supreme Court, it gave the Supreme Court suf-
ficient ground to understand matters in a balanced and reasonable way.
Though the VCR was designed to steal, the Court concluded that it could
not be banned as an infringing technology unless there was no “potential”
for a “substantial noninfringing use.”
Potential. For a substantial noninfringing use. Notice what this standard
does not say. It does not require that a majority of the uses of the technology
be noninfringing. It requires only that a “substantial” portion be nonin-
fringing. And it does not require that this noninfringement be proven today.
It requires only that there be a potential for this noninfringing use. As long
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 196

1 9 6 L a w r e n c e L e s s i g

as one can demonstrate how the technology could be used in a way that was
legitimate, the technology would not be banned by a court.
The Supreme Court’s test is rightly permissive. The tradition of American
law is not to ban technologies, but to punish infringing use. And that test
should have had an obvious answer in the context of the Napster case. Here
there are no doubt lots of infringing uses. But there are also lots that under
any fair estimation constitute fair or noninfringing use. Music that has been
released to the Net to be freely distributed is freely distributed through Nap-
ster. That use is clearly noninfringing and is substantial. Music that has
fallen into the public domain is available on Napster. That use is clearly
noninfringing, and is substantial. And lots of recordings that are not
music—lectures, for example—can be made available on Napster. The
Electronic Frontier Foundation has a series of lectures that are traded on
Napster; they are offered as content that is free.
But when this claim was made to Judge Marilyn Hall Patel in California,
she, unlike Judge Ferguson in the Sony case, had no patience for the argu-
ment. Without a trial, and with barely contained contempt, she ordered the
site shut down.
Within thirty-six hours, Napster attorney David Boies had received a stay
of that order from the Ninth Circuit Court of Appeals. And after hearing ar-
guments in the case, that court affirmed much in the injunction of Judge
Patel.34 The court did, however, make one important modification: Napster
was not responsible for contributory infringement unless the copyright
holder made Napster aware of the violation. Napster therefore wasn’t closed
down by the court; it wasn’t required to become the copyright police. But it
was required to remove music posted contrary to the copyright holder’s
wish. So, like the circuits of the computer Hal in the movie 2001, the music
in the memory of the Napster system will be slowly turned off, as copyright
holders will demand the right to control the sharing of their content.

Eldred

recall the story of Eric Eldred’s HTML book library from chapter 8. As
I described there, Eldred has a passion for producing HTML books from
public domain works. As the Framers of our Constitution plainly envi-
sioned, after a limited time, copyrights expire, and the work previously pro-
tected then falls into the public’s hands without restraint. Eldred takes those
public domain works and turns them into freely accessible on-line texts.
But in recent years, Congress has changed the rules. In 1998, Congress
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 197

T H E F U T U R E O F I D E A S 1 9 7

extended the term of existing copyrights by twenty years. As I’ve said, this
was simply the latest extension in a pattern that began forty years ago. While
Congress changed the term of copyright just once in the first hundred years
of copyright, and once again in the next fifty years, it has extended the term
of subsisting copyrights eleven times in the past forty years.
This latest extension meant that works that were to fall into the public do-
main in 1999 would now not be “free” until the year 2019. Thus, works that
Eldred had prepared to be released were now bottled up for another gen-
eration.
This latest change outraged many, and especially Eric Eldred. Eldred
threatened civil disobedience—promising to publish a series of Robert Frost
poems that would have fallen into the public domain. After some of us
convinced him that that was a very dangerous strategy, Eldred chose instead
to challenge the statute in court. In January 1999, in a federal court in
Washington, D.C., Eldred filed his complaint.
Eldred’s claims were simple. If the Constitution permits Congress to
grant authors an exclusive right “for limited times,” then the Framers of that
power clearly intended that that exclusive right must come to an end. Per-
mitting Congress the power to perpetually extend copyrights would defeat
the purpose of the express limitation.
This was Eldred’s claim based on the language of the copyright clause of
the Constitution. He also raised an argument based on the First Amend-
ment. The First Amendment says that Congress “shall make no law . . .
abridging the freedom of speech, or of the press.” Copyright is a law that cer-
tainly limits Eric Eldred’s HTML press. So how are these two provisions of
the Constitution—one granting Congress the power to issue copyrights,
and the other limiting Congress’s power to “abridge” the freedom of the
press—to be reconciled?
The Supreme Court has explained how the two coexist. Copyright, the
Court has written, is an “engine of free expression.”35 Because of the incen-
tives that copyright law provides, work gets created that otherwise would not
have been produced. This means that copyright law both increases speech
and restricts it. And a fairly balanced copyright law can, in principle, at least,
increase more than it restricts. That means that copyright law does not nec-
essarily “abridge” speech; and hence the copyright clause does not neces-
sarily conflict with the guarantees of the First Amendment.
But as Eldred argued, this rationale cannot justify extending the terms for
existing copyrights. Existing copyrights protect work that is already created;
extending the terms for this work restricts speech without any promise of fu-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 198

1 9 8 L a w r e n c e L e s s i g

ture creativity. The one thing we know about incentives, Eldred argued, is
that incentives are prospective. Whatever we promise Hawthorne, he isn’t
going to produce any more work.
Both claims appealed to the Framers’ sense of balance in establishing the
copyright power. As Justice Joseph Story described it, the power gave au-
thors exclusive control for a “short interval”; after that interval, the work was
to fall into the public’s hands “without restraint.”36 At the time Story wrote
that, a “short interval” was an initial term of fourteen years. Today, that
“short interval” can easily reach ten times that term.
The courts, however, had little patience for the Framers’ sense of bal-
ance. Both the District Court and the Court of Appeals for the D.C. Circuit
held that the copyright clause did not constrain Congress to a single “lim-
ited time.” It was free to grant extensions, as long as the extensions them-
selves were limited. (As Professor Peter Jaszi described it, Congress is
therefore free to grant a perpetual term “on the installment plan.”)37 And
more dramatically, in rejecting the First Amendment claim, the court of
appeals held that “copyrights are categorically immune from First Amend-
ment scrutiny.”38
The meaning of these two holdings together is that the ability to proper-
tize culture in America is essentially unlimited by the Constitution—even
though the plain text of the Constitution speaks volumes against such ex-
pansive control. And the consequence of this power to propertize was per-
haps best exemplified by a lawsuit to stop the publication of what many
considered a sequel to Margaret Mitchell’s Gone with the Wind.
Gone with the Wind was published in 1936. Under the law as it existed
then, Mitchell’s copyright would have expired at the end of 1992. But be-
cause of the extensions that Eldred was fighting, that copyright now extends
until 2031. Until then (or later, if Congress extends the term again), the
Mitchell estate has exclusive rights over the story, as well as over other sto-
ries that are sufficiently close to the original to be called “derivative.”
In 2001, Alice Randall tried to publish a work called The Wind Done
Gone. While she called it a parody of Gone with the Wind, that was
her lawyers speaking more than Randall. The work is clearly based on
Mitchell’s work; in telling the story of Gone with the Wind from the per-
spective of the African slaves, it clearly relies upon Mitchell’s work in an in-
timate and extensive manner. The Mitchell estate called the work a sequel
and brought a federal lawsuit to stop its publication. This story, the Mitchell
estate essentially argued, was theirs to control well into the twenty-first cen-
tury.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 199

T H E F U T U R E O F I D E A S 1 9 9

To most people, this is plainly absurd. Gone with the Wind is an extraor-
dinarily important part of American culture; at some point, the story should
be free for others to take and criticize in whatever way they want. It should be
free, that is, not only for the academic, who would certainly be allowed to
quote the book in a critical essay; it should be free as well for authors like
Alice Randall as well as film directors or playwrights to adapt or attack as
they wish. That’s the meaning of a free society, and whatever compromise
on that freedom copyright law creates, at some point that compro-
mise should end.
The Gone with the Wind case, as well as Eldred’s case, is still working its
way through the courts. But both tell a similar story: The freedom to build
upon and create new works is increasingly, and almost perpetually, re-
stricted under existing law. To a degree unimaginable by the Framers of our
Constitution, that control has been concentrated in the hands of the hold-
ers of copyrights—increasingly, large media companies.

C O N S E QU E N C E S O F C O N T R O L

the internet in its nature shocks real-space law. That’s often great; it is
sometimes awful. The question policy makers must face is how to respond
to this shock.
Courts are policy makers, and they too must ask how best to respond.
Should they respond by intervening immediately to remedy the “wrong”
said to exist? Or should they wait to allow the system to mature and to see
just what harm there is?
In the context of porn, as I have already argued, the courts’ response is to
wait and see. And indeed, this is the response of the government in many
different contexts. Porn, privacy, taxation: in each case, courts and the gov-
ernment have insisted we should wait to see how the network develops.
In the context of copyright, the response has been different. Pushed by an
army of high-powered lawyers, greased with piles of money from PACs,
Congress and the courts have jumped into action to defend the old against
the new. They have legislated, and litigated, quickly to assure that control of
the old is not completely undermined by the new.
Ordinary people might find these priorities a bit odd. After all, the record-
ing industry continues to grow at an astounding rate. Annual CD sales have
tripled in the past ten years.39 Yet the law races to support the recording in-
dustry, without any showing of harm. (Indeed, possibly the opposite: when
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 200

2 0 0 L a w r e n c e L e s s i g

Napster usage fell after the court restricted access, album sales fell as well.
Napster may indeed have helped sales rather than hurt them.)40
At the same time, it can’t be denied that the Net has reduced the ability
that parents have to protect their children. Yet the law says, “Wait and see,
let’s make sure we don’t harm the growth of the Net.” In one case—where
the harm is the least—the law is most active; and in the other—where the
harm is most pronounced—the law stands back.
Indeed, the contrast is even stronger than this, and it is this that gets to the
heart of the matter.
The Internet exposes much more copyrighted content to theft than in the
world that existed before the Internet. This much of the content holders’
claim is plainly true.
But as I’ve argued, the Internet does two other things as well. First, the
Internet makes it possible (if the proper code is deployed) to control the use
of copyrighted material much more fully than in the world before the Inter-
net. And second, the Internet opens up a range of technologies for produc-
tion and distribution that threaten the existing concentrations of media
power.
In responding to the shock that the Internet presents to copyright law, it is
of course important to account for the increased exposure to theft. But the
law must also draw a balance to assure that this proper response to an in-
creased risk of theft does not simultaneously erase the important range of
access and use rights traditionally protected under copyright law. If the Net
creates an initial imbalance, the response by Congress should not create an
equal and opposite imbalance, where traditional rights are lost in the name
of perfect control by content holders.
That was my argument in Code. But now we should add a second con-
cern to that same story: The response by Congress should also not be such
as to permit this concentrated industry of today to leverage its control from
the old world into the new. Artists deserve compensation. But their right to
compensation should not translate into the industry’s right to control how
innovation in a new industry should develop.
Control, however, is precisely Hollywood’s and the recording labels’ ob-
jective. In the context of copyright law, the industry has been very clear: Its
aim, as RIAA president Hilary Rosen has described it, is to assure that no
venture capitalist invests in a start-up that aims to distribute content unless
that start-up has the approval of the recording industry.41 This industry thus
demands the right to veto new innovation, and it invokes the law to support
its veto right.42
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 201

T H E F U T U R E O F I D E A S 2 0 1

Michael Robertson of MP3.com agrees that this is the aim and effect.
“[T]his litigation,” Robertson told me, “is as much about straddling the
competition as anything else.”43 And it has had its effect.

[W]hat they’ve done very successfully is dried up the capital markets for
any digital music company. [W]e went public a little over a year ago [and]
raise[d] $400 million from going public. Today, if you took a digital music
company business plan, you couldn’t get a buck and a half from a venture
capital company.44

This is the reality that the current law has produced. In the name of pro-
tecting original copyright holders against the loss of income they never ex-
pected, we have established a regime where the future will be as the
copyright industry permits. This puny part of the American economy has
grabbed a veto on how creative distribution will occur.
One could quibble about whether current law is properly interpreted to
give existing interests this control. Some see these cases (in particular the
MP3.com and Napster cases) as simple; I find them very hard. But whether
they are simple or hard, the underlying law is not unchangeable. Congress
could play a role in making sure that the power of the old does not trump
innovation in the new. It could, that is, intervene to strike a balance be-
tween the right of copyright holders to be compensated and the right of in-
novators to innovate.
The model for this intervention is something we’ve already seen: the
compulsory license.45 For recall, as I described in chapter 4, the first real
Napster-type case: cable television. It, like Napster, made its money by
“stealing” the content of others. Congress in remedying this theft required
that the cable companies pay content holders compensation. But at the
same time, Congress gave cable television companies the right to license
broadcasting content, whether or not the copyright holder wanted to.
Congress’s aim in part was to assure that the cable industry could develop
free of the influence of the broadcasters. The broadcasters were a powerful
industry; Congress felt (rightly) that cable would grow more quickly and in-
novate more broadly if it was not beholden to the power of broadcasters. So
Congress cut any dependency that the cable industry might have, by assur-
ing it could get access to content without yielding control.
Compensation without control.46
The same solution is available today. But the recording industry is doing
everything it can to keep Congress far from this solution.47 For it knows that
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 202

2 0 2 L a w r e n c e L e s s i g

if it has the absolute right to veto distribution that it can’t control, then it can
strike deals with companies offering distribution that won’t threaten the la-
bels’ power. The courts, whether rightly or not, have handed the labels this
veto power; Congress, if it weren’t flustered by the emotion of the recording
industry, could well intervene to strike a very different balance.
We find that balance by looking for a balance—not by giving copyright
interests a veto over how new technologies will develop. We discover what
best serves both interests by allowing experimentation and alternatives.
But this is not how the law is treating copyright interests just now. Instead,
they are in effect getting more control over copyright in cyberspace than
they had in real space, even though the need for more control is less clear.
We are locking down the content layer and handing over the keys to Holly-
wood.
The costs of this lockdown are great enough without the Internet; the
Internet makes them much more significant. Before the Internet, as I de-
scribed in chapter 7, production was concentrated in the hands of the
few. With the Internet, this production could be widespread. But to the
extent that content remains controlled, to the extent the Alice Randalls
or Eric Eldreds must seek permission to use or build upon other as-
pects of our culture, these controls create barriers to new creativity. They
block the potential for innovation, by adding protections for existing in-
terests.

okay, time for a politics check. I know what you’re thinking: These are
just the ravings of a rampant leftist. But as writer Siva Vaidhyanathan argues,
“There is no ‘left’ or ‘right’ in debates over copyright. There are those who
favor ‘thick’ protection and those who prefer ‘thin.’ ”48 The argument in
favor of balance is not a liberal vs. conservative argument. The argument is
old vs. new.
The credentials of at least some conservatives in this debate cannot be
questioned. Circuit judge Richard Posner—father of much in law and eco-
nomics, and perhaps the most prolific and influential judge of the last
hundred years—has written persuasively about the complexity in finding
balance in copyright law. As I’ve described, the property right of copyright is
incomplete. As Posner writes:

Since the property right is incomplete, one might suppose that literature is
being underproduced and therefore copyright protection should be ex-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 203

T H E F U T U R E O F I D E A S 2 0 3

panded in both scope and duration—perhaps made comprehensive and
perpetual. The matter is not so simple.49

Not simple—indeed, quite complex. The complexity is just what we’ve
been considering throughout this book. Intellectual property is both an
input and an output in the creative process; increasing the “costs” of intel-
lectual property thus increases both the cost of production and the incen-
tives to produce. Which side outweighs the other can’t be known a priori.
“An expansion of copyright protection,” Posner argues, “might . . . reduce
the output of literature . . . by increasing the royalty expense of writers.”50
Thus the ideal mix cannot be found simply by increasing the power of copy-
right holders to control.
Other conservatives are a bit more colorful about the point. Consider, for
example, one of the brightest stars of the Ninth Circuit Court of Appeals,
Judge Alex Kozinski.
Kozinski is an immigrant. His family suffered at the hands of Romanian
communism; they fled Romania when he was twelve.51 In 1985, he was ap-
pointed by President Reagan to the federal bench. He has since then been
the darling of the Federalist Right. He is an extraordinarily talented and in-
sightful judge, who has little patience for the paternalism of the liberal Left.
But the extremes of copyright drive him mad, and there is no better an
opinion describing his view of limited copyright terms than a dissent he
wrote to an opinion upholding the right of Vanna White to control the use
of images that would remind the public of her.
At issue in the Vanna White case was whether intellectual property law—
in particular, a state-created right of publicity—would permit Vanna White
of Wheel of Fortune fame to control all images that suggest her, including in
this case any advertisement that “evoke[s] the celebrity’s image in the pub-
lic’s mind.”52
The Court of Appeals for the Ninth Circuit—or, as that circuit includes
California, the Court of Appeals for the Hollywood Circuit, as Kozinski puts
it53—upheld White’s right to control the use of this image. Kozinski sharply
dissented. As he wrote:

Something very dangerous is going on here. Private property, including in-
tellectual property, is essential to our way of life. It provides an incentive
for investment and innovation; it stimulates the flourishing of our culture;
it protects the moral entitlements of people to the fruits of their labors. But
reducing too much to private property can be bad medicine.54
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 204

2 0 4 L a w r e n c e L e s s i g

Why? For the same reasons we’ve been tracking throughout this book.

Private land . . . is far more useful if separated from other private land by
public streets, roads and highways. Public parks, utility rights-of-way and
sewers reduce the amount of land in private hands, but vastly enhance the
value of the property that remains.55

The state must therefore find a balance, and this balance will be struck
between overly strong and overly weak protection.

Overprotecting intellectual property is as harmful as underprotecting it.
Creativity is impossible without a rich public domain.56

But is that unfair? Is it unfair that someone gets to profit off the ideas of
someone else? Says Kozinski, No.

Intellectual property law assures authors the right to their original expres-
sion, but encourages others to build freely on the ideas that underlie it.
This result is neither unfair nor unfortunate: It is the means by which in-
tellectual property law advances the progress of science and art. We give
authors certain exclusive rights, but in exchange we get a richer public do-
main.57

This balance reflects something important about this kind of creativity:
that it is always building on something else.

Nothing today, likely nothing since we tamed fire, is genuinely new: Cul-
ture, like science and technology, grows by accretion, each new creator
building on the works of those who came before. Overprotection stifles the
very creative forces it’s supposed to nurture.58

This balance is necessary, Kozinski insists, “to maintain a free environ-
ment in which creative genius can flourish.”59 Not because “flourish[ing]”
innovation is the darling of the Left, but because innovation and creativity
were the ideals of our founding republic.

my story so far has been about copyright and, indirectly, its cousin, trade-
mark law.60 I have argued that these two bodies of rights will together be
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 205

T H E F U T U R E O F I D E A S 2 0 5

used by the old to protect themselves against the threat of the new. This pro-
tection is not necessary; there is nothing in our tradition that compels it. But
it is pushed not by those with the most to lose, but by those without the most
to win. And I have argued that we should be skeptical about just this sort of
protectionism.
But now I want to describe a second form of protectionism—perhaps
more threatening to the promise of the Internet’s future. This threat too is
the product of state intervention into Internet space. And this intervention
is even harder to justify.61

the issue here is patent law.62 A patent is a form of governmental regula-
tion. It is a state-backed monopoly granting exclusive rights to an “inventor”
for an invention deemed useful, novel, and nonobvious.
The argument favoring patents is as old as the hills. If an inventor can’t
get a patent, then he will have less incentive to invent. Without a patent, his
idea could simply be taken. If his idea could simply be taken, then others
could benefit from his invention without the cost. They could, in other
words, free-ride off the work of the inventor. If people could so easily free-
ride, fewer would be inventors. And if fewer were inventors, then we would
have less progress in “science and useful arts.”
Getting more progress is the constitutional aim of patents. So the ques-
tion that must always be asked of any patent regime is whether we have good
reason to believe that patents have that effect. As Harvard law professor
Stephen Shavell has written, “there is no necessity to marry the incentive to
innovate to conferral of monopoly power in innovations.”63 So is there any
evidence that it does any good?
In some cases, the evidence is good.64 For some kinds of innovations,
patents are extremely likely to induce more innovation. In particular, in
theory, where innovation is independent, or noncumulative (meaning one
invention is essentially separate from another), then economists predict that
patents will clearly benefit innovation.65 Likewise, even where innovation is
cumulative, if the use of the patent is clear, then in principle, the original
patent holder will have a strong incentive to license a patent to follow-on in-
novators.66 But here, economists have an important qualification: If we
don’t know which direction an improvement is likely to take, then licensing
may not occur, and patents here may actually do harm.67 Thus, for econo-
mists, at least, the theory suggests contexts in which innovation will be
helped by patents as well as contexts where it will be harmed.68
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 206

2 0 6 L a w r e n c e L e s s i g

The empirical evidence is less encouraging.69 The strongest conclusion
one can draw is that whatever benefit patents provide (except in industries
such as pharmaceutics),70 it is small. As economist Adam Jaffe concludes,
“[T]he value of patent rights might still be too small relative to overall costs
and returns to have a measurable impact on innovative behavior.”71 And as
he concludes more broadly:

There is a widespread unease that the costs of stronger patent protection
may exceed the benefits. Both theoretical and, to a lesser extent, empirical
research suggest this possibility. Economists have long understood that, at
a theoretical level, technological competition can lead to a socially exces-
sive level of resources devoted to innovation. The empirical literature is
convincing that, for the research process itself, the externalities are clearly
positive on balance (Griliches, 1992). But to the extent that firms’ atten-
tion and resources are, at the margin, diverted from innovation itself
towards the acquisition, defense and assertion against others of property
rights, the social return to the endeavor as a whole is likely to fall.72

Other commentators increasingly agree. As The Economist recently sum-
marized a broad range of research: “Do firms become more innovative
when they increase their patenting activity? Studies of the most patent-
conscious business of all—the semiconductor industry—suggest they do
not.”73
This skepticism has been with us from the start of the patent system. Ben
Franklin thought patents immoral.74 Some of the greatest inventors of our
history have refused to patent most of their inventions.75 Science has tradi-
tionally resisted patents.76 And even Bill Gates, no patsy when it comes to
intellectual property protections, expressed skepticism about software
patents. As he wrote in a memo to Microsoft executives in 1991:

If people had understood how patents would be granted when most of
today’s ideas were invented and had taken out patents, the industry would
be at a complete standstill today.77

The first patent commissioner himself—Thomas Jefferson—was also ex-
tremely skeptical about these forms of monopoly. Commenting upon the
proposed Constitution, with its proposed provision for granting monopolies
to cover writings and inventions, Jefferson wrote that he wished the draft
would be amended to eliminate any monopolies. As he wrote:
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 207

T H E F U T U R E O F I D E A S 2 0 7

I sincerely rejoice at the acceptance of our new constitution by nine states.
It is a good canvas, on which some strokes only want retouching. What
these are, I think are sufficiently manifested by the general voice from
north to south which calls for a bill of rights. It seems pretty generally
understood that this should go to juries, habeas corpus, standing armies,
printing, religion and monopolies. . . . The saying there shall be no mo-
nopolies lessens the incitement to ingenuity, which is spurred on by the
hope of a monopoly for a limited time, as of 14 years; but the benefit even
of limited monopolies is too doubtful to be opposed to that of their general
suppression.78

Jefferson’s views about patents were not his alone. From the beginning of
the Supreme Court’s interpretation of the law of patent, it has affirmed that
patents are no natural right; that the scope of patent rights is just as far as
Congress extends it. And Congress should extend it only when Congress has
reason to believe the monopolies it extends will do some good.

in the first two hundred–plus years after Congress first enacted a patent
statute, the duration and scope of patent law were fairly stable. The Framers
set a term of four years; they quickly extended that to fourteen; and that term
is close to the current term of twenty. And from the start, patents were not
granted for just anything; invention was required. So too today, when an in-
vention must be novel, nonobvious, and useful.79
But in the past twenty years, an important shift has occurred. The limits
to the reach of patent law have been eroded by a number of expansions in
patent law doctrine. “These changes,” Adam Jaffe writes, “were not brought
about primarily by Congressional action, but rather by the . . . Patent Of-
fice.”80
The expansions I want to focus on here are those relating to cyberspace.
And these include the patenting of software inventions and business meth-
ods.
Before the 1980s, software inventions in the United States were not
subject to patent protection. The reasons were tied to the nature of pro-
gramming (programs were considered algorithms, and algorithms were tra-
ditionally not protected), but the arguments in favor of not making software
patentable were more pragmatic. Since software is often distributed without
its source, it is often extremely hard to understand how it is in fact achieving
its effect. On the surface, functions could be implemented in any number
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 208

2 0 8 L a w r e n c e L e s s i g

of ways. When you sort a list of addresses within an address book program,
in principle, the algorithm that sorts the list could be one of a million such
programs. (There’s more than one way to skin a cat.) When you display a
picture, how the picture is displayed is nothing that is obvious to the devel-
oper or user.
But beginning in the 1980s, courts started recognizing software in-
ventions as patentable inventions. And by the early 1990s, these patents
had taken off. Patent applications for software-related patents went from 250
in 1980 to 21,000 in 1999, and the number granted has increased eight- or
ninefold.81
What was most striking about this explosion of law regulating innovation
was that the putative beneficiaries of this regulation—coders—were fairly
uniformly against it. As Richard Stallman put it, “We did not ask for the
change that was imposed upon us.”82 And this attitude was not limited to
free software advocates. When the U.S. Patent Office began explaining this
new benefit it would be providing software developers, key developers from
a range of software industries were frantic in avoiding the benefit. As Doug-
las Brotz from Adobe Corporation said in 1994:

I believe that software per se should not be allowed patent protection. I
take this position as the creator of software and as the beneficiary of the re-
wards that innovative software can bring in the marketplace. . . . [Adobe
and I] take this position because it is the best policy for maintaining a
healthy software industry, where innovation can prosper.83

Oracle took the same position.84 The system wasn’t broken, these coders
said. It certainly didn’t need Washington to fix it.
But Washington was not to be deterred, and the push for software patents
did not go away. Quite the opposite. Over time, the push was for even
broader patent protection—this time to cover business processes as well as
software inventions.
A software-implemented business process patent is a patent for a process
of doing business, sufficiently novel and nonobvious to earn the U.S. Patent
and Trademark Office’s favor.85 Most thought such processes beyond the
reach of patent law. This was not because patent law never covered
processes—it plainly did. But the expectation was that it would not cover
business processes because adequate return from the process itself would
create a sufficient incentive to invent.86
In 1998, however, the United States Court of Appeals for the Federal Cir-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 209

T H E F U T U R E O F I D E A S 2 0 9

cuit put this idea to rest. The patent law reached business processes just as
any other, and patents for business methods were, the court held, not in-
valid because of the subject matter.87
The case in which this issue arose was one where a financial services
company had developed a new kind of mutual fund service, one that would
manage a pool of mutual funds through a software-based technology. The
court upheld both the software patent and the patent on the business
method. Both, the court said, were inventions that the patent law could
reach. This decision, in turn, gave birth to an explosion of business method
patent applications. And by 1999, many were beginning to be approved in a
way that surprised the industry. Applications for computer-related business
methods jumped from about 1,000 in 1997 to over 2,500 in 1999.88 High on
that list was the Amazon 1-Click patent, but also on the list were Price-
line.com’s reverse auction patent, and British Telecom’s claim that it owned
the invention of hypertext links (and hence the World Wide Web!).89
In all these cases, the question the monopoly-granting body asked was
simply this: Was this sort of “invention” sufficiently like others that were the
subject of patents? If so, then the patent was granted for this field of innova-
tion.
Economists, however, are likely to ask a much different question. While
it is clear that patents spur innovation in many important fields, it is also
clear that for some fields of innovation, patents may do more harm than
good.90 While increasing the incentives to innovate, patents also increase
the costs of innovation. And when the costs outweigh the benefits, patents
make little sense.
How could this be? The answer links to an argument we’ve seen in many
different contexts before. The ordinary argument for a strong patent right is
a kind of prospecting theory. First advanced by Edward Kitch, the prospect
theory says there is good reason to hand out broad, strong patents because
then others will know with whom they should negotiate if they want to build
upon a certain innovation.91 This in turn will create incentives for people to
invent, and as information is a by-product of invention, it will induce
“progress” in the “useful arts.”92
The problem with this theory, however, is its very strong assumption (in
some contexts, at least) that the parties will know enough to properly license
the initial foundational invention, or that other issues won’t muck up the in-
centives to license.93
Both limitations on the ability to license are what economists would call
transaction costs.94 The transaction cost from ignorance is similar to the in-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 210

2 1 0 L a w r e n c e L e s s i g

sight the founders of the Net had when they embraced an end-to-end ar-
chitecture: rather than architecting a system of control from which changes
could be negotiated, they were driven by humility to a system of noncontrol
to induce many others to experiment with ways of using the technology that
the experts wouldn’t get.95
The transaction cost affecting incentives to license is in part a problem of
ignorance, but in part the problem of strategic behavior that we’ve seen in
many different contexts. It is the problem Christensen is discussing in The
Innovator’s Dilemma: the problem of nonneutral platforms that guided my
review in chapter 4 of open code projects.
My claim is not that these transaction costs are so high as to make patents
unadvisable in the Internet context. My point is simply that these consider-
ations, supported as they have been,96 at least raise a question.
So given this complexity, you might think that policy makers would be
eager to know whether the fields covered by software and business method
patents are the sorts where innovation is helped by patents or harmed. You
might think—given the extraordinary importance that these markets have
played in the recent economic boom—that before the government tries to
fix something through monopolies, it would check to see if anything is bro-
ken.
I had the chance to ask the government just this. In a debate in Wash-
ington, I was on a panel with Q. Todd Dickinson, patent commissioner in
the last days of the Clinton administration. In my part of the opening pre-
sentation, I suggested that it would be important to know whether patents
will help in these fields or harm.
Dickinson was impatient with the suggestion. As he said:

Some days I wish I was the professor and only had to think about these
things and not do the work. But I got an office to run. And I’ve got 1,500
applications coming in this year and I have to figure out what to do with
them. I don’t have the luxury to wait for five years for Congress to figure
out whether they will change the law or not.

Publisher and Net guru Tim O’Reilly was on the same panel. He had a
quick and devastating response. The head of the U.S. Patent Office,
O’Reilly said, has two roles in the administration. One is, as Dickinson had
just said, to run the office. But the other is to advise the administration about
what policy made sense. And where, O’Reilly asked, following up on my
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 211

T H E F U T U R E O F I D E A S 2 1 1

own question, was the policy analysis that justified this extraordinary change
in regulation?
I remember thinking, Where are the Republicans when you need them?
Here was critical new regulation that would significantly affect innovation
in cyberspace. Where was the regulatory impact statement? Here was a gov-
ernment official overseeing a radical expansion in patent regulation, within
a field that had been the most important component of growth in the
United States’ economy in the past twenty years. Yet the government didn’t
have time to learn whether its patent policy would do any harm or good?
Regulate first, ask questions later.
There’s good reason to wonder whether patents are necessary in a field
such as this. Patent law is designed to create a barrier against idea theft, so
that inventors have an incentive to invent and use their ideas. The term of
this protection is not to be overly long: patents are monopolies; monopolies
raise prices. The term should be long enough to give enough incentive,
without being so long as to raise prices unnecessarily.
But a patent isn’t the only device that might protect the innovator against
inefficient copying. Being first to market in a network economy creates a
first-mover advantage without imposing the costs of a patent.
And other incentives are often sufficient to induce innovation without a
patent. Jeff Bezos, for example, said of the 1-Click patent that Amazon.com
would have developed the 1-Click technology whether or not there was a
patent system.97 The reason is obvious: The system helps sell more books,
and the profit from those additional sales of books is enough of an incentive
for the invention of new technology.
Either one of these reasons, plus a host of others suggested by legal and
economic scholars, would lead a rational policy maker to ask whether mo-
nopoly is needed here.98 But this question has not been asked about patents
affecting cyberspace.

so why is Washington doing it? What reason could there be for the gov-
ernment to allow this launch of regulation to occur without even a hearing
about whether the regulation will do any good?99
The answer is obscure, but we can identify a number of causes. First is
the patent bar itself. Dickinson is not an evil man; his heart is certainly in
the right place. But he is a political figure, who feels the pressure of the in-
terest group that is most affected by the decisions of his department. That
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 212

2 1 2 L a w r e n c e L e s s i g

interest group is the patent bar—a group of lawyers who like a world where
their market increases dramatically. What interest do they have asking
whether this increased regulation does any good?
Second is our general way of thinking about patents. Most of us don’t
think about patents as a form of regulation. Most consider patents property
in the same sense that my car is property. In that same debate, patent king
Jay Walker was also on the panel. He argued that the question in this debate
about patents was whether you were for property or against it—and in his
view, the pro-property view was “beyond reproach.”100
But again, this is just silly. Patents are no more (and no less) “property”
than welfare is property. Granting patents may make sense, as providing
welfare certainly makes sense. But the idea that there is a right to a patent is
just absurd. From the very beginning, our tradition has self-consciously
understood that the only question guiding whether to issue a patent or not
is whether or not patents will do society any good. As conservative economist
Friedrich von Hayek put it:

It seems to me beyond doubt that in [the fields of patent and copyright] a
slavish application of the concept of property as it has been developed for
material things has done a great deal to foster the growth of monopoly and
that here drastic reforms may be required if competition is to be made to
work.101

Rather than reason, what governs the current patent debate is bias—bias
in favor of a system that seems right just because it seems old.102 But the
relevant system is not old—it is being expanded in ways that would shock
lawyers of a generation ago. And something is right not because it is old, but
only if it does some good. But we will never know whether or not it does any
good if we accept this never-ending expansion without limit. We will never
know what benefit this regulation provides until we begin to demand that
the regulation prove itself.
For the harms from this regulation are not hard to identify, and for the
cynical, or conspiratorial, the harms are not surprising. (On the margin, the
costs of a patent system will harm small inventors more than large; negoti-
ating a patent system is easier for IBM than for the garage inventor.)103 And
the harms from an expanded American patent system will harm foreign in-
ventors more than American. (It is easier to hire American law firms locally
than from a distance.) Thus this expansion in patent protection will shift the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 213

T H E F U T U R E O F I D E A S 2 1 3

competitive field away from the small, non-American inventor in favor of
the large, American inventor.104
The harms are even more pronounced, however, for open code projects.
Tim Berners-Lee has noticed its effect on Web development already. (“De-
velopers are stalling their efforts in a given direction when they hear rumors
that some company may have a patent that may involve the technology.”105
One example is the development of P3P, which may enable better protec-
tion of privacy on the Web.) And open code proponents—like software de-
velopers generally—have been among the strongest opponents to patents in
this field. As Richard Stallman writes, “The worst threat we face comes from
software patents, which can put . . . features off-limits to free software for up
to twenty years.”106 Red Hat chairman Bob Young thinks much the same:
“[S]oftware patents [are] an evil, or at least [a] very damaging encroachment
on the efficacy of the software programming industry.”107
The reason patents harm open code in particular is not hard to see.
Think about the mechanics of licensing a patent when you are licensing for
anyone working on an open code project. Who knows who they are? How
many users need to be sanctioned? As Peter Wayner writes, “[T]hese ques-
tions are much easier to answer if you’re a corporation charging customers
to buy a product.”108 Thus patents tilt the process to harm open code devel-
opers.
The problem is exacerbated with software patents because though the
patent system was designed to induce inventors to reveal their invention to
the public, there is no obligation that a software inventor reveal his source
code to get a patent. “The single most revealing symptom” of the failure of
the existing system, Professor Brian Kahin writes, “is that the software pro-
fessionals do not read patents.”109 As Bob Young analogizes it, “It’s like that
ceramic guy, producing a new kind of ceramic and [patenting it] without
ever telling anyone how he made the extra hard ceramic. So in software
you’re saying ‘I’m patenting software that has this look and feel, but I don’t
actually have to tell people how I achieved that look and feel.’ The source
code remains a secret.”110
And then there is the expense of patents, which is borne more sharply by
smaller inventors than larger. The costs include the costs of securing a
patent, but those in the end are trivial. The real costs are borne by those who
would challenge a patent. If the U.S. Patent Office makes a mistake, and a
patent is granted that shouldn’t be granted, then it costs on average $1.5 mil-
lion (for each side) to take a patent dispute to trial.111
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 214

2 1 4 L a w r e n c e L e s s i g

Finally, there is the obvious “hold-up” problem—where an innovator is
about to release a product and is discovered to be violating a patent. As
Berkeley economist Carl Shapiro describes it:

The hold-up problem is worst in industries where hundreds if not thou-
sands of patents, some already issued, others pending, can potentially read
on a given product. In these industries, the danger that a manufacturer
will “step on a land mine” is all too real. The result will be that some com-
panies avoid the minefield altogether, i.e., refrain from introducing cer-
tain products for fear of hold-up. Other companies will lose their
corporate legs, i.e., will be forced to pay royalties on patents that they
could easily have invented around at an earlier stage, had they merely
been aware that such a patent either existed or was pending.112

As Shapiro concludes, “[T]his ‘hold-up’ problem is very real today,
and . . . should [be considered] a problem of first-order significance in the
years ahead.”113
This may be an unintended consequence of this recent expansion in pro-
tection. I am, for example, quite certain this would not have motivated the
different courts that have contributed to this expansion. But it may well ex-
plain why there is little passion from those who fund lobbyists to find a way
to cut back on the expansion. By letting things go as they are, this change
may well give them a competitive advantage over the innovator who can’t
fund a legal team or isn’t from the United States. Again, as Bill Gates of
Microsoft told his senior management:

A future start-up with no patents of its own will be forced to pay whatever
price the giants choose to impose. That price might be high: Established
companies have an interest in excluding future competitors.114

this story about the potential danger of patents in a field where innova-
tion is sequential and complementary (where one builds on another, and
the second complements the value of the first) gets additional support from
an ingenious argument that Michigan law professor Michael Heller initially
made and that economist James Buchanan has now followed up on.115
Heller introduces the concept of an “anticommons.” If a commons is a re-
source where everyone has a right to use the resource (and therefore some-
times overuse the resource), an anticommons is a resource where many
have the right to block the use of a resource by others (and therefore many
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 215

T H E F U T U R E O F I D E A S 2 1 5

more underuse the resource). Heller gives the example of formerly state-
owned buildings in post-Soviet Russia: Because of the many claims that
could be made on them, the buildings were never developed. Too many bu-
reaucrats could veto any project, and thus insufficient effort at innovating in
the use of these buildings was made.
Nobel Prize–winning economist James Buchanan has expanded this idea
to the problem of regulation generally.116 He points to the problem of
patents in particular as an example where multiple and overlapping patent
protection may create an anticommons, where innovators are afraid to in-
novate in a field because too many people have the right to veto the use of
a particular resource or idea. This potential for strategic behavior by these
many rights holders makes it irrational for an innovator to develop a par-
ticular idea, just as the possibility of veto by many bureaucrats may leave a
particular piece of real property underdeveloped.
These ideas map directly onto the argument we’ve considered in this
book. Control, when complex, can often increase the costs of using a re-
source; increasing those costs can easily chill innovation. Recall the ex-
treme of AT&T’s control over innovation in the telecommunications
system: Who would waste his or her time developing for that system, when
any development would require convincing so many quasi bureaucrats
before it could even be tried?
The complexity in these rights to exclude creates this anticommons prob-
lem. And the more severe the problem, the more it will stifle new innovation.

i’ve told a story about intellectual property in two critical competitive
contexts.117 In both contexts, the emerging regime will have a significant
regulatory effect. In both contexts, the regime will shift protection from the
new to the old. The law in both cases will, on the margin, protect the old
against the new. RIAA president Hilary Rosen was clear about this objective
in the context of copyright law: No new ideas should be allowed unless the
old system of distribution okays it. And this will be the certain, if unin-
tended, consequence of the patent system as well. Those most likely
to be displaced by new innovation will have the power, through these
government-backed monopolies, to check or inhibit this innovation.
This power is the product of government-backed monopolies that in the
ordinary case raise little trouble. I am not against copyright law (I agree with
Hollywood: if you have simply copied the whole of this book, you are a
thief); in the ordinary case, the scope of its monopoly ought to be respected.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 216

2 1 6 L a w r e n c e L e s s i g

But when we, as a society, undergo a radical technological shift—which the
Internet revolution certainly is—then we should reexamine the scope of the
monopoly power we extend and ask once again whether that power makes
any sense. Is it necessary? Is there reason to believe it will do some good?
The tradition before the Internet had favored massive increase in the
scope of copyright law and a significant increase in the reach of patents. Es-
sentially anything you could attribute to a creative work, you had to respect
by getting the permission of this creative work before using it.
In a world like the world I described as the dark ages, this may not be a
terrible thing. When all publishers are largish corporations, who really cares
if creative energies must be licensed? The licensing process is an ordinary
cost of doing business, just like paying sales tax or filing statements with the
SEC. It may, on the margin, inhibit a bit, but not a terribly significant
amount.
But when the world of creativity shifts outside the largish corporation—
when individuals and smaller groups are much more enabled to do this cre-
ative activity—then this system of exclusive licenses for every derivative use
of a creative work begins to tax the creative process significantly. The op-
portunity cost, as economists would describe it, of this system of control is
higher when, without this system of control, much more creative activity
would go on.
Thus, when we have a massive shift in opportunity, we should be reevalu-
ating how necessary these systems of control are. We should be asking
whether control is necessary, or at least how far control is required. And if
we don’t have a good reason for extending these systems of government-
backed control, then we shouldn’t. If we have no good reason to believe a
government-backed monopoly will help, then we have no good reason to es-
tablish government-backed monopolies.

at the end of chapter 7, I argued that the control of media in the dark ages
may well be a product of economic constraints. That as long as economics
constrains, then this system of concentration and control may be inevitable.
The constraints I identified are not to be imagined or ignored away. They
are real and unavoidable.
But the constraints that I have described in this chapter are different.
They are not “real” in the same sense. The constraints of IP are constraints
we build. We create regimes of IP, and then the regimes we have built yield
the control I have identified. No doubt these regimes are in large measure
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 217

T H E F U T U R E O F I D E A S 2 1 7

justified. No doubt in the main they promote progress. But often (in copy-
right for sure, and possibly with patents as well) the regime expands beyond
its initial justification. The restrictions it imposes are artificial, in the sense
that they don’t promote progress; they simply benefit one person at the ex-
pense of another.
This then presses the fundamental question of this book: If the extremes
of these constraints are not necessary, if there is no good showing that they
do any good, if they limit the range of creativity by virtue of the system of
control they erect, why do we have them?
For this is a change. The content layer—the ability to use content and
ideas—is closing. It is closing without a clear showing of the benefit this
closing will provide and with a fairly clear showing of the harms it will im-
pose. Like the closing of the code layer described in chapter 10, this closing
of the content layer is control without any showing of a return. Mindless
locking up of resources that spur innovation. Control without reason.
This closing will not be without cost. Making it harder for innovations to
enter, making resources more universally controlled—this will drive new
competitors off the field, leaving the field once again safe for the old.
And more important, this closing does not occur without a purpose. As I
suggested at the end of the last chapter, our greatest fear should be of di-
nosaurs stopping evolution. More precisely, we should be most concerned
when existing interests use the legal system to protect themselves against in-
novation that might threaten them. The commitment of a society open to
innovation must be to let the old die young. The law should resist becoming
a tool to defend against the new; when change is on the horizon, it should
allow the market to bring about that change.
This is just what is not happening in the field of intellectual property.
The state is being pushed to defend expanded intellectual property rights in
the name of protecting the way the world was.
As in chapter 10, we are allowing an idea about “property” to overrun the
balance that grants access. Because we don’t see that balance, or don’t see
the place for balance, we are quick to follow the arguments that favor con-
trol.
Again, this idea in the background—the sanctity of perfect control—
blinds us. We in turn render blind the opportunities for innovation. When
the only innovation that will be allowed is what Hollywood permits, we will
not see innovation. That lesson, at least, we have already seen.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 218

C o n t r o l l i n g W i r e - l e s s

( a n d H e n c e t h e P h y s i c a l L a y e r )

O n the wall in my office is a multicolored poster. The poster is large
(maybe 30 by 42 inches), and it is titled, in beautifully retro typewriter
font, “United States Frequency Allocations—The Radio Spectrum.” To the
left are thirty-three colored boxes, listing the legends for the poster. Thirty
list “radio services.” Three list “activity codes.” Among the activity codes are
“government exclusive,” “government/non-government shared,” and “non-
governmental exclusive.” (Appropriately enough, government exclusive is
red, while nongovernmental exclusive is green.)
If you could tilt this poster and give it a bit of a 3D look, it might remind
you of the famous New Yorker cartoon maps, where everything close is de-
tailed and significant, while everything far is wide open and unimportant.
So it is with spectrum as well. At the highest frequency (30–300 gigahertz),
the allocations are a patchwork of tiny colored boxes, sometimes four
deep; but as you move down the frequency range, the allocations get wider
and less precise. The largest swath is AM radio.
This map, however, doesn’t mark out any physical space. It marks the al-
location of radio spectrum. The map says what kind of use will be permitted
at what range of radio spectrum in any particular part of the territorial
United States. It does not say by whom.
As I described in chapter 5, the “by whom” part is determined by a com-
plex set of federal regulations. The FCC makes a decision about who gets to
use what spectrum when, and under what conditions. These “licenses” are
not really licenses to spectrum. As Thomas Hazlett describes them, they are
simply permissions to use certain kinds of equipment at certain times for
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 219

T H E F U T U R E O F I D E A S 2 1 9

certain purposes. Their effect is therefore not so much to regulate a re-
source (spectrum) as it is to determine who has the rights to engage in cer-
tain kinds of businesses, where. To say that company X has an FCC license
is to say that the government has given company X the right to engage in a
certain kind of business (say, radio broadcasting) using certain equipment
tuned to certain radio frequencies.1
The manner in which this allocation of rights to use spectrum is made
has changed, and it changes still. As Eli Noam describes, in the first era of
spectrum use, spectrum was allocated on a first come, first served basis. This
was before the federal government entered the field. After 1912, Noam’s
“second era,” it was the government that chose who got what spectrum. This
invited predictable biases: existing owners bought the favor of regulators,
and regulators in turn protected them. The examples are many, and extraor-
dinary (at least to those who live outside D.C.): Hazlett has cataloged the
cases where favored interests have succeeded in using their power over regu-
lators to resist new technologies;2 as Noam writes, “[I]n the early 1950s, only
newspaper companies that had editorially endorsed Eisenhower for Presi-
dent had a chance at getting a TV license.”3
In the third era (now), the right to use spectrum is increasingly allocated
through auctions. The government sells the right to the highest bidder
(subject to a scad of typically governmentlike, mainly silly, conditions).
That bidder uses the spectrum as the auction specifies or, in a small set of
cases, the bidder is then free to reassign the right to others.
Politicians from the Left and the Right just love auctions. For the Left,
auctions promise more money for the government to spend; for the Right,
auctions sound like markets, and markets are always good.
But as we saw in chapter 5, both auctions and government assignments
ignore a fundamentally different way to “allocate” spectrum—namely, not
allocating it or, more realistically, not allocating all of it. Rather than as-
signing rights to use the spectrum resource ex ante, this alternative would
allow users to share the resource when the need to use it arose. This sharing
would be policed either by a market (in Noam’s conception) or by other
technological devices designed to deal with congestion.4 Like the Internet,
on this latter model, the system would find a technological means to deal
with undercapacity. Spectrum could then be shared, and a range of differ-
ent technologies says how.
How much? How completely? Could shared spectrum govern every-
where?
The optimists in this story say that shared spectrum could service all of
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 220

2 2 0 L a w r e n c e L e s s i g

our spectrum needs: that we could replace allocated spectrum in one fell
swoop and free spectrum for the use of all.5 Others argue that shared spec-
trum could serve only a small part of our spectrum needs and that we will
always need to have some spectrum that is allocated or controlled through
a market.6
Whether all or not, however, it is clear that shared spectrum could serve
a great deal of our spectrum needs, which means that not all spectrum
needs to be allocated for spectrum to be usable. Even under the most con-
servative estimates of what shared spectrum might offer, shared spectrum
could serve a large and important part of our spectrum demand.
Which raises an important question that by now will have become famil-
iar: To the extent spectrum could be shared, what justifies the extent of
spectrum allocation that we now see?

to bureaucrats and legislators, this kind of question will seem odd.
What justifies it? “What justifies it is that we’ve always done it like this. Any
change must be as we permit.” Thus, the FCC is moving slowly to open the
spectrum that it can, just as fast as it believes is right in the face of any lob-
bying or opposition we might see.
But to courts, the question of justification is at the core of what they do.
And when one asks what justifies a particular system of allocation, “We’ve
always done it like this” is not an answer. The answer is that it is justified
only if it is.
So is this system of allocation justified?
When the government allocates a speech resource like spectrum, its
decisions are tested according to a well-defined standard. The question is
not whether the regulation meets strict scrutiny—as would a regulation
that said, for example, that only Republicans may use the spectrum. The
question instead is whether the regulation meets intermediate scrutiny—
whether “[1] it advances important governmental interests unrelated to the
suppression of free speech and [2] does not burden substantially more
speech than necessary to further those interests.”7
If a court addressed that question now, my sense is that it would clearly
decide that the FCC’s system for allocating spectrum is just fine. But it
would decide that mainly because the alternatives are not yet developed or
understood. We don’t have many great examples of how spectrum could
otherwise be regulated. Most of the mature examples of how spectrum can
be used are examples that rely upon this system of allocated spectrum.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 221

T H E F U T U R E O F I D E A S 2 2 1

But the catch-22 is that the ability of these alternatives to demonstrate
their success depends upon the FCC’s opening up spectrum for alternative
use. The more it opens, and leaves to private market experimentation, the
easier it will be to demonstrate that the shared spectrum model works. And
likewise with the contrary. The less the FCC leaves open for experimenta-
tion, the fewer incentives there will be in the market for innovators to de-
velop new innovation. Thus, the critical need right now is a broad range of
spectrum where these alternative uses might demonstrate themselves.
In a sense, this is the telephone network all over again, though this time
the wires are ether, and the control is imposed through law alone rather
than through the law backing up the control of AT&T’s technology. Until
innovators are free to use a communications resource (now spectrum, be-
fore the wires), innovation will be slowed. Yet in this case, one would ex-
pect, the claim to free use is even stronger than with the wires. No investor
or corporation built the radio spectrum. This resource was given to us pre-
built by Mother Nature. Thus, the claim to free access is simply a claim that
the government not get in the way of experimentation by innovators.

at the very minimum, this possibility suggests a strategy for government
regulators (if those regulators were not effectively captured by existing spec-
trum users). The strategy builds on what we know: that government control
over spectrum use has stifled innovation; that it will continue to stifle inno-
vation as long as existing users have a political channel through which they
can defend their existing privilege. We therefore should move—as quickly
as possible—to a regime where the right to innovate does not depend upon
the permission of someone else.
Essentially, two such regimes are possible. As I described in chapter 5,
one follows Coase, and the other is a commons. Under the first, spectrum
would be propertized and sold on a market. The buyer would be free to
manage his spectrum however he saw fit. “Band managers” would control
the chunks of spectrum that they own; users wanting to “use” that spectrum
would license that right from a wide array of spectrum owners. Assuming
the supply of spectrum would be great and the number of competitors large,
this system would produce a strong competition in spectrum supply. No sin-
gle supplier could control innovation any more than a single supplier of
paper can control what books get written.
Under the second regime, spectrum is held in a commons and shared in
real time by smart technologies for sharing. The rights are not allocated up
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 222

2 2 2 L a w r e n c e L e s s i g

front; as with the Internet, the demand is managed by protocols as it arises.
Here, too, assuming the supply of spectrum is great and the protocols neu-
tral, there would be no one who could stymie innovation. As with the Inter-
net, the system would have no intelligence to discriminate against one form
of spectrum use in favor of another.
There are advantages and disadvantages to both regimes. The property
regime would produce great competition in spectrum use, and if competi-
tion were sufficiently “perfect,” then, as I’ve noted before, the regime would
produce the feature of the commons that is most salient here: that strategic
action by the resource owner would not be possible. The only costs to the
property regime are the burdens of any property regime—the costs imposed
on the market by the need to negotiate and secure rights to access.8 And
if the ability to “share” spectrum becomes central to efficient spectrum
management, then the costs of securing this right to “share” through private
contract could become quite prohibitive.
The commons regime, too, would produce great competition in spec-
trum use. But the danger with the commons is overuse: that the free re-
source of spectrum would produce more demand than supply; and that
important uses of spectrum would be shut down by congestion.
But in this choice between regimes, the mistake is to assume that a single
solution is necessary. There is no reason to embrace either the market or the
commons completely. Instead, the best strategy for now would be to em-
brace both solutions vigorously—to mark off significant chunks of spectrum
for sale, while leaving significant chunks of spectrum open in a commons.
Alongside auctioned space, we should have broad swaths of unowned space.
And the government should then assure innovators that the unowned would
remain so for a good long time—say, the length of a patent or, better yet, a
single copyright (just to get some political energy on the side of reducing
the term of copyright).9
This way the market could experiment with technologies and different
spectrum uses. Some spectrum uses may need reserved space; let the mar-
ket provide that. Other spectrum uses may be more flexible—as use of the
Internet today is. But the opportunity for broad and creative use of the spec-
trum would inspire many to develop technologies that they wouldn’t other-
wise build. As entrepreneur Alex Lightman puts it:

We need to have some slack in the system. We need to have a certain
amount of the spectrum not be in the category where it’s owned.10
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 223

T H E F U T U R E O F I D E A S 2 2 3

This proposal is relatively neutral. It doesn’t take sides in the technology
of the future; it instead structures the competitive environment to allow a
range of technologies to flourish. It acknowledges that many different solu-
tions are possible; it recognizes the radical change in technologies for using
spectrum that we are already seeing; and it simply embraces a strategy for
getting the most out of these innovations. But it does say that for the mo-
ment, “these things need to be held in common, and they need to be held
for the future. Because we don’t know what the future will hold.”11

so is this proposal anything close to what the government is actually
doing? The answer is yes in words, but no in reality. At the level of high
theory, the government remains committed to developing these alternative
uses. But the devil is in the details, and the dungeons of Washington are
well detailed.
If there is one thing that is certain about governments and innovation, it
is that those who are threatened by new innovation will turn first to the gov-
ernment for help. Spectrum policy is not, and has never been, different.
Every new idea is a threat to those who depend upon old ways of doing busi-
ness. As David Hughes puts it, “[B]ig corporations have always marched
to government to lock in their profits.”12 So we might well expect that
those whose way of doing business depends upon the comfortable life of
government-backed monopolies over spectrum will do what they can to
make sure that those monopolies are not threatened by this new, free way of
using spectrum.
We might expect it, and if we look at what’s happening at the FCC, we
would also observe it. For as quickly as innovators can develop new ways of
using spectrum, incumbents are finding ways to make this innovation
harder.
As I’ve already argued, this is nothing new. The surprise is how blatantly
this protectionism continues. Consider, for example, the effort of former
FCC chairman William Kennard to license low-power FM radio stations.
This was a good move from the standpoint of increasing competition and di-
versity in speech. Kennard was committed to finding a way to free spectrum
resources to enable a broader range of speakers. And there was very good
technical evidence that these low-power radio stations—which might sup-
port a community action center or a local school—would create no techni-
cal interference with existing radio stations.13
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 224

2 2 4 L a w r e n c e L e s s i g

The aim of the FCC was to enable local community broadcasts while as-
suring the broadcasts would not interfere with existing radio stations. The
technical staff of the FCC conducted tests to determine how low-power the
stations would have to be to assure no interference; the rules the FCC even-
tually proposed were more conservative than the technical staff recom-
mended. “As a result of the FCC’s conservatism, community groups in large
urban centers with many incumbent broadcasters would find it difficult, if
not impossible, to operate. But it would have enabled over 1,000 commu-
nity organizations, churches, and schools to create a new medium for local
discourse.”14
But the existing stations balked. At first they complained to the FCC.
When the FCC concluded that their evidence of interference was not sub-
stantiated, the broadcasters went to Congress. Congress didn’t care much
about these low-power stations (not many campaign dollars, after all, come
from them). It did care about the broadcasters who were threatened. So
Congress passed a law to restrict low-power broadcasters.15 Large FM sta-
tions were protected from increased competition; that protection was ef-
fected through a law that silenced other speakers. So much for the First
Amendment’s demand that “Congress shall make no law . . . abridging the
freedom of speech.”
An example closer to the technology at the core of this chapter is the case
of “AirPorts” in “airports.” The AirPort is a wireless device sold by Apple
Computer. It uses the 802.11b protocol to enable a computer to connect to
a network at very fast speeds—11 megabits per second is the maximum for
802.11b, which is about twice the current DSL or cable speed. The device
uses spread spectrum technology within one of the three swaths of spectrum
that the FCC has allocated for “unlicensed use” for data.
Apple was a pioneer in pushing this form of technology. But the AirPort
connects not just Apples. Any computer with an 802.11b wireless card can
connect to an AirPort. And other companies, too, are building the equiva-
lent of AirPort modems. Indeed, a whole sector is growing up around this
possibility of wireless network access.
Some got the idea of putting AirPorts (or their equivalent) in airports—
enabling travelers to connect to their Internet while sitting in an airport
lounge. But soon local airport authorities started to complain: wireless
modems, they argued, would interfere with air traffic controllers. They
would also reduce the usage of pay phones.16
Now, I don’t doubt that interference is possible for some of these new
technologies. It is important that we be certain that new technologies don’t
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 225

T H E F U T U R E O F I D E A S 2 2 5

damage important pieces of the existing infrastructure—at least those parts
that we want to keep. But this complaint about air traffic was just silly.
There is more interference caused by a hair dryer than by an Apple AirPort
modem. And the notion that airport authorities should be able to stop
progress to protect their telephone revenue is absurd.17
What’s needed in contexts like this is a balanced way to evaluate these
claims of interference to resolve whether they are real or just pretext. More
generally, what’s needed is a commitment to progress in the use of spectrum
resources.
Instead the politicians have done just the opposite. Claims of technical
interference are not credibly evaluated. Indeed, the FCC has placed the
burden on new technologies not to harm existing use at all.18 As Hazlett
puts it, the “system is booby-trapped against new rivals, an irresistible ‘at-
tractive nuisance’ to anticompetitive constituencies.”19 Thus, for example,
amateur radio operators are allowed to veto new spectrum uses if they inter-
fere at all with existing ham operations. And this pork is not just because of
a special favor to amateur operators. Any new use that interferes with any
old use must step aside.20
These restrictions on the use of amateur radio spectrum are particularly
ironic. The Amateur Radio Service (ARS) defines a range of spectrum that
is allocated to amateurs, but the members of the service “share” the spec-
trum in a commons mode. Any amateur can use any portion of that spec-
trum at any time and with just about any modulation technique now
known. Within the prime “beachfront” spectrum property (from 30 mega-
hertz to 3 gigahertz) there is a great deal allocated to the ARS. But armed
with the FCC’s veto rule, amateurs are effectively able to veto new and dif-
ferent users of “their” spectrum—despite the obligation imposed by the
FCC’s regulation to enhance “the value . . . to the public as a voluntary
noncommercial communication service.”21
These are technical rules that protect the old from the new. So too are
there political rules that achieve the same end. Among these, none is more
significant than the commitment and eagerness of the government to sell
rights to spectrum, in the way spectrum rights are sold now.22 This form of
auction essentially entrenches the use of spectrum for particular businesses.
The license “does not yield the right to deploy spectrum in alternative
uses.”23 It entrenches a way of speaking of spectrum that is resistant to mod-
ern sharing technologies: that the spectrum is “my property.” Once that
property is established, it will be harder to deploy technologies that “share”
other people’s “property.” (This despite the fact that an applicant for an
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 226

2 2 6 L a w r e n c e L e s s i g

FCC license must first certify that it will not assert any propertied interest in
radio spectrum.)24 As Dave Hughes has asked:

[W]hy should AT&T, who is offering a wireless service, . . . consent to
competitors in their own area? . . . [I]t’s not just a question of interference
now. Now it becomes . . . opening the door by their consent to competi-
tion. And the last damn thing big companies want is competition.25

the danger in selling spectrum or, more precisely, in not experimenting
broadly with unlicensed spectrum is that existing spectrum users will be
able to use purchased spectrum to resist changes in spectrum policy that
might threaten their business models.26 By selling spectrum now, before al-
ternative uses can be developed, we create a world where the resources for
these new alternatives are held by those with the strongest incentive to stop
them. As Eli Noam puts it, it is like “having the old AT&T auction off the
right to compete against it. Under such a system, MCI would not have
emerged.”27
The concern is not just about spectrum owners; it is also about the nature
of the existing spectrum uses. The dominant and fastest-growing spectrum
use right now is mobile telephone systems. These systems are architected in
just the way the old telephone network was—intelligence is located not at
the ends, but instead in the network itself. The cellular phone companies
retain control over how the cellular technology develops; if you want a new
application for your (increasingly powerful) phone, you will get it only if the
telephone company wants you to.
This architecture for a wireless system creates the obvious protectionist
risks. And as Charmed Technologies CEO Alex Lightman puts it, we are al-
ready seeing these risks mature into protectionist practices.

[T]here is a nice little cozy ménage à trois between the companies that are
providing infrastructure and the companies that provide the handsets, and
the monopoly carriers or the oligopoly carriers.28

By selling the spectrum, the carriers have a strong incentive to assure re-
turns sufficient to recover the investment in spectrum. These returns are
best assured (or at least it seems to the companies that are best assured) if the
companies husband the market power that is carried over from the non-
competitive telephone world (recall: much of the action here is inter-
national, where competitive phone systems don’t yet exist). Thus, the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 227

T H E F U T U R E O F I D E A S 2 2 7

willingness of the existing players to open up their spectrum to a wildly dif-
ferent form of use—one that would be much more competitive—is unlikely
at best. Recall the words of AT&T executive Jack Osterman in 1964:
“[W]e’ll be damned if we allow the creation of a competitor to ourselves.”29
A policy from the FCC that does not create a strong opportunity for an al-
ternative to develop is designed to protect existing interests. To not encour-
age or permit wide-scale experimentation, to not set aside much broader
unlicensed spectrum, to protect existing uses against any interference—
these are policies designed to preserve the old against the new. They are just
what we would expect from government regulation of spectrum; they are
much less than we should demand after the experience of the Internet.
The government’s role should be to induce investment where there is a
great deal of social value to be created. This is precisely the opportunity with
unlicensed spectrum.30

opponents of this mixed strategy are of two sorts: those who think it is
unnecessary—that the market will get us to the right answer without the ex-
periments with open spectrum; and those who think it unwise—that open
spectrum is a terrible way to allocate spectrum resources.
The first group thinks that the market has a sufficient incentive to find the
optimal use of spectrum. As with any resource, these market mavens argue,
privatization will give owners the strongest incentive to commit the re-
sources that they hold to the highest and best use. Thus, if there is an inno-
vative new way to use spectrum, then someone who owns the spectrum has
the best incentive to find it and deploy it. Markets deal best with scarcity and
choice about innovation. Hence we should be pushing to strengthen the
market.
Peter Huber is a good example of this kind of optimist. Huber is a brilliant
polymath who, while working full-time as a lawyer, has written some of the
most important policy and academic work about government regulation in
general and telecommunications policy in particular.31 In his book Law and
Disorder in Cyberspace, Huber describes both the market model for allocat-
ing spectrum—where spectrum rights are auctioned off up front—and the
commons model for allocating spectrum, promoted most strongly by an ally
of Huber’s, George Gilder. As I described in chapter 5, Gilder argues
strongly that we should allocate spectrum as a commons. Auctions, Gilder
argues, will simply entrench existing uses; free or common spectrum would
create a strong incentive for new uses.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 228

2 2 8 L a w r e n c e L e s s i g

Huber does not reject Gilder’s predications. He argues instead that we
could get to Gilder’s world of spectrum as a commons by first auctioning off
all the spectrum and then allowing the market to “reassemble” the rights if
that proves efficient.

Markets find ways of reassembling private pieces into public spaces when
that is the most profitable thing to do. They may take more time than an
omniscient central authority, but finding omniscient central authority
takes even longer. For now, the thing to do is to get the spectrum out of
government hands, however it can be done, and leave it to the market to
re-create the public commons. It will, if the economics are there.32

This is an empirical claim that begs for some evidence. I doubt there is a
single example of “private pieces” “reassembling” into “public spaces.” Cer-
tainly there are examples of small landowners selling their land to develop-
ers; and we might imagine developers selling their land to other developers.
But I can’t begin to imagine the process by which this buying and selling of
rights eventually leads to the spectrum commons that Gilder has described.
There are too many sirens of strategic behavior on the way to imagine the
private property system working itself pure.
If Huber’s model were correct, we should expect that at least in some
places the construction of commons has been left solely to a market. But at
least within our tradition, however distasteful this is to strong libertarians,
the most important commons have been supported by state intervention.
Not the intervention of nationalization—these are commons, not state
property. Instead, the intervention of a legal system that protects certain
resources as open and neutral. Roads are not built through the “reassem-
bling of private pieces” into a national highway grid. They are built by self-
conscious commons constructions.33
Even if one were optimistic in the way Huber is, a more obvious question
is this: Why start with the burdens of establishing a property system if there
is good reason to believe that the better solution would be a commons? Why
not start the other way around—with a commons that might, in certain cir-
cumstances, be privatized where needed?
This default of a commons gains support when one considers the kind of
spectrum use that is increasingly preferred by researchers. To the extent re-
search shows that a more efficient manner of organizing spectrum is not to
narrow the bandwidth allocated while increasing the power of the transmit-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 229

T H E F U T U R E O F I D E A S 2 2 9

ter, but instead to broaden the bandwidth used with much less power, the
model of propertized spectrum makes much less sense.34
It is here that the open spectrum model confronts a different critic—one
who argues that a spectrum commons would be a wildly inefficient method
of allocating spectrum. And here, Thomas Hazlett is again the strongest
voice. In a recent paper addressing this idea of the spectrum commons,
Hazlett rightly ties the idea to the Internet. Writes Hazlett:

The spectrum commons idea is motivated by analogy to the Internet. Yet
the architecture of the Internet . . . seriously misallocates scarce band-
width. Because data cannot easily be prioritized, or billed, within the ex-
isting Internet protocols, tragedy of the commons appears frequently. High
value communications are jammed in congested arteries with massive vol-
umes of data of only marginal significance. Classically, the brain surgeon
cannot read the life-or-death CT-scan because the Internet backbone is
clogged with junk e-mail.35

But the response to Hazlett’s example is not to criticize the Internet. The
response is to ask who is this “brain surgeon” reading a CT scan over the
Internet? And how does her ability to use the Net determine whether
the Net “seriously misallocates” resources? Hazlett offers no data to support
the claim that the “tragedy of the commons appears frequently.” In fact, ca-
pacity has consistently outstripped demand.36
More significant, Hazlett ignores the advantages to innovation that I
have identified throughout this book. Let’s assume that the Internet “mis-
allocates” bandwidth relative to the model Hazlett has. Does Hazlett really
believe that the very same innovation (or better) would have been realized
had the Internet been architected to “properly allocate” bandwidth from
the start?
For there is no conceptual reason why we couldn’t have auctioned off the
Internet’s resources at the start of the Internet. We could easily have imag-
ined protocols to charge and prioritize being implemented from the very
start. If we had, would we have produced the same kind of innovation?
Across the same range? Had the network been architected to give the net-
work owners control, would we have produced the Internet that we did?
Clearly, in my view, the answer is no. But I’m just a lawyer; I haven’t the
skill to model this counterfactual. My point is simply the part that
Hazlett has not yet accounted for. He must at least show us why the oppor-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 230

2 3 0 L a w r e n c e L e s s i g

tunity for innovation that the original Net created was not actually benefi-
cial. And he must likewise show that it would not be beneficial in spectrum,
either.
Hazlett has his biases; I have mine. He’s enamored of perfect pricing and
perfect control; I’m still surprised (lawyer that I am) by the extraordinary in-
novation that comes from imperfect pricing and leaky control. If we had to
take a poll among neutrals—people biased neither as I am nor as Hazlett
is—I suspect most would be skeptical about whether a control architecture
in the Internet would have produced the same innovation that the com-
mons architecture did. But whatever a poll would indicate, it is here that the
debate should occur. Given the creativity and innovation that the original
Internet produced, and given how different that innovation is relative to
other computer networks and other telecommunications systems, my bet for
spectrum would be that an architecture modeled on the Internet would not
be so bad.

the ideal mix in the short term would be a regime that had both a com-
mons and a property component, with the property component subject to
an important caveat. There would be broad swaths of spectrum left in the
commons; there would be broad swaths that would be sold as Hazlett pro-
poses. But in light of the emerging technologies for sharing, even the spec-
trum sold as property would be subject to an important qualification: Other
users would be free to “share” that spectrum if they followed a “listen first”
protocol—the technology would listen to see whether a certain chunk of
spectrum were being used at a particular time, and if it weren’t, it would be
free for the taking.
I recognize that idea is jarring—that “my property” would be free for the
taking just because I was not using it. But do you recognize why the idea is
jarring? The assumption that fuels the dissonance about property “free for
the taking” is that the taken property is exhaustible. I may not be using my
car at the moment, but that doesn’t mean you should have the right to take
it, since your use of my car will, to some degree, deplete the property I have.
Cars are exhaustible resources.
Spectrum is not. When I use a bit of spectrum at a particular moment in
time, that spectrum is just as good after I’m finished as it was before. My use
in no way exhausts the resource. And more important, when spectrum is not
used, its value as a resource is not saved. Unused spectrum, like an empty
seat on an airplane, is a resource that is lost forever.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 231

T H E F U T U R E O F I D E A S 2 3 1

Thus, if we adjust our intuitions about what spectrum is—recogniz-
ing that even if under some protocols spectrum is rivalrous (in some sense,
we both can’t use the “same” spectrum at exactly the same time), it is
inexhaustible—then a property rule that presumes an opportunity for
sharing begins to make sense. The implicit “use it or lose it” require-
ment, while costly for some sorts of rights, is not costly where the social con-
sequence of not using a particular resource is that that resource is lost forever.
Wouldn’t there still be a conflict if I’m “sharing” your spectrum and you
then decide you want to use it? In principle, but not in practice. For the
“sharing” rule would require that the sharer use the spectrum for an ex-
tremely short period of time. Thus, the owner may be delayed, but the delay
would not be significant.
This compromise simply recognizes an important limitation in current
understanding of how spectrum might optimally be used.37 It may well be
that the market model makes most sense, and that in the long run, a market
for spectrum will govern all spectrum. But that doesn’t mean we need to
embrace the market fully now. For as Eli Noam has demonstrated, a market
structure could be layered onto open spectrum without embracing the ex
ante allocation that auctions envision.38 Like the Ethernet network de-
scribed in chapter 5, at the time the system needs it, the system would make
a request for a reservation. The only addition would be a system for charg-
ing for that token of reservation. Noam likens that system to the subway’s
method of charging for ridership.
The advantage of Noam’s solution is that it keeps the cost of spectrum use
down. As Noam writes, an auction is simply “a tax on the communications
sector and its users.”39 Given the size of the bids currently being offered for
this resource, it will tend, as Noam argues, to encourage oligopoly. “An auc-
tion payment that must be paid in advance is a barrier to entry, unless capi-
tal markets are perfect, which they are not.”40
We are far from the moment, however, when it would make sense to layer
this market onto open spectrum. Just as the National Park Service began
charging entrance fees late in its life, so too should we begin without en-
trance fees and layer them on, neutrally, as needed. This mode of regulat-
ing a large chunk of the resource of spectrum would inspire the widest
range of spectrum use. And a strong commitment by the government to sup-
port open spectrum would convince venture capitalists to invest in this
alternative use.

* * *
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 232

2 3 2 L a w r e n c e L e s s i g

but while theorists favoring the market fight theorists pushing the com-
mons, our government is pursuing neither policy well. Instead, our govern-
ment pursues policies that are precisely contrary to freeing the spectrum,
either through the competition of a well-functioning market or through a
commons. It gets pushed by politicians, so it pushes to sell off rights to spec-
trum without fully embracing the property model. The reason is not ideal
spectrum policy. The reason is that it auctions off this form to best protect
existing businesses. As Gilder puts it, this push to sell off spectrum is simply
a “legal infrastructure and protectionist program for information smoke-
stacks and gas guzzlers.”41 The effect won’t be to inspire new ways of using
spectrum; instead, the effect will be to entrench the old ways against new
uses.
While auctions “seem preferable to the agency’s previous policy of simply
giving [spectrum] away to ‘worthy’ applicants,” they will not inspire new
uses, and they may tend, Gilder argues, “to produce a winner’s curse.”42
The expected revenue from these sales will exceed $50 billion; that revenue
will put pressure on the incumbents to earn supracompetitive returns. The
best way to earn supracompetitive returns is to continue the noncompetitive
architecture in broadcasting that has earned them profits in the past. Selling
spectrum will give the incumbents the means and the motive to make
certain that the spectrum does not become (as we as a society should want)
a commodity product—constrained by strong and broad competition.

gilder’s argument assumes a kind of irrationality within firms that
economists are quick to attack. The money spent on spectrum is a “sunk
cost.” The rational business would ignore the sunk costs already spent and
focus instead on the optimal way to get a return from the assets it has.
But Gilder’s fear is not just a result of irrationality; the fear is that power-
ful actors can work to slow innovation that harms them. The long-run ad-
vantages of FM radio didn’t stop AM broadcasters from working to kill
FM.43 The technical arguments in FM’s favor, powerful and unrefuted,
were impotent in the face of existing broadcasting interests. And the same
danger continues to exist about spectrum management policies. There are
too many places for the devil to find details that will effectively kill impor-
tant new technologies.44
There is an opportunity here for a crucial layer of the communicative
architecture to be opened and made free. Opening it would reduce the
pressure on other channels of Internet access. Keeping it free would en-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 233

T H E F U T U R E O F I D E A S 2 3 3

courage a wide range of innovation around how that resource is used. But
instead of encouraging the use of this resource, instead of expanding it
broadly, we are quickly whittling away the opportunity this commons would
create. Without justification beyond the knee-jerk bias of our day, we are
swallowing the idea that control is better than freedom.

here again, an idea about property is doing all the work—but this time
the idea is at its most attenuated. We don’t yet have a full property regime
for allocating and controlling spectrum. Yet we are still being driven to em-
brace this single view. We are racing to deny the opportunity for balance,
pushed (as we always are) by those who have the least to gain from a world
of balance. The possibility of a commons at the physical layer is ignored;
even the chance to experiment with the commons is denied. Instead, policy
makers on the Right and on the Left race to embrace a system of perfect
control.
So strong is this idea of property, so unbalanced is our understanding
of its tradition, that we embrace it fully, without limitation, even when it
doesn’t yet exist, and even when the asset being assigned a property right is
not—like the wires of AT&T’s cable or the creative genius behind Disney’s
Mickey Mouse—something anyone has created. We are racing to assign
property rights in the air, because we can’t imagine that balance could do
better.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 234

W h a t ’ s H a p p e n i n g H e r e ?

I n the early 1970s, RCA was experimenting with a new technology for
distributing film on magnetic tape—what we would come to call video.
Researchers were keen not only to find a technology that could reproduce
film with high fidelity; they were also keen to find a way to control the use
of the technology. Their aim was a technology that could control the use of
film distributed on video, so that the owner of the film might maximize its
return from the distribution.
The technology eventually chosen was relatively simple. A video would
play once, and when finished, the film would lock into place. If a renter of
the video wanted to play the video again, he or she would have to return the
video to the video store and have the tape unlocked. In this way, the owner
of the film could assure that it was being compensated for every use of the
copyrighted material.
RCA presented this technology to the Disney Corporation in the early
1970s. In a room with just five of the senior executives from Disney, a young
RCA executive, Pat Feely, demonstrated RCA’s device. The executives were
horrified. They would “never,” Feely reports their saying, permit their con-
tent to be distributed in this form. For the content, however clever the self-
locking tape player was, was still insufficiently controlled. “How could they
know,” a Disney executive asked Feely, “how many people are going to be
sitting there watching” a film? “What’s to stop someone else coming in and
watching for free?”1

* * *
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 235

T H E F U T U R E O F I D E A S 2 3 5

as the cost of digital filmmaking falls, educators are experimenting
with filmmaking as a way of learning. There’s something different about ex-
pression in film. As John Seely Brown, chief scientist at Xerox PARC, de-
scribes it:

[As] you move into the role of film . . . you get something that most people
overlook. [I]n text we can always add the parenthetical comment: Paren,
dot, dot, dot, comma, if, sorta. . . . And so you string qualifier after quali-
fier. [But w]hen you do a sketch of an idea there are no qualifiers. . . .
[Q]ualifiers don’t hold. You’ve got to decide what the kernel idea is and
then you sketch that idea.2

This means learning through filmmaking is different. And in a group of
California schools, a number of filmmakers have been experimenting with
giving students the tools to make film as a kind of writing. These experi-
ments let the students draw upon a wide range of existing film, which they
recombine in new and creative ways and then supplement with new scenes
that the students shoot on their own.
This “changes the thinking process,” Brown describes. It produces a
“completely different experience from writing an essay.” And though these
educators are just at the start of this experiment, they have already seen the
changes it can make in the students it touches. This is a new kind of think-
ing, enabled by this emerging digital technology.
The product of this creativity, however, can’t be displayed publicly. The
films these students produce are housed on a private, password-protected
network. Lawyers for the university supporting this work have advised the
educators that putting the content on the Web would subject the teachers to
liability. Thus this creativity is bottled up in a private network, unavailable
for other students to view or learn from. The lesson the students learn is that
sharing this creativity is not to be allowed.
But why? John Seely Brown asks.

To me, this is where education . . . is going. [This is how] students who grow
up digital think and want to learn. [But] we are building a [legal] system
that completely suppresses the natural tendencies of today’s digital students.

The law says that their creation can’t be shared and the technologies that
we’ve seen in chapter 11 will take away the very ability to draw upon film
content and mix it with something different.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 236

2 3 6 L a w r e n c e L e s s i g

Here we have an educational crisis in our country and we have this
incredible platform where these students are discovering on their own.
[But this platform] is progressively going to be closed down instead of
opened up.

“Technology makes it possible,” Brown concluded, “but the law is going
to come in . . . and knock it out.”

the story of this part is easy to summarize. In one sense, each of the
changes I describe in these three chapters is very different. Modifications to
the broadband environment are different from modifications to patent law;
changes in spectrum rules don’t quite track the motives of Hollywood.
Yet in another sense, each of these examples is motivated by a common
idea or common attitude. In each, an attitude of control, perfected by an
idea about property, is in tension with a system that protects a commons.
And in each, the idea about property prevails. We race to empower networks
to discriminate (after all, they are “their computers”); we race to empower
owners of copyright to control new modes of distribution; we race to
develop property in the air. Our single, overriding view of the world is
that only property matters; our systematic blindness is to the lesson of our
tradition—that property flourishes best in an environment of freedom, both
freedom from state control and freedom from private control. That a com-
mons can have value greater than the same assets would if enclosed.
The consequence in each of these contexts is a change in the environ-
ment within which innovation occurs. That change has been, or threatens
to be, a shift from a world where the commons dominates to a world where
control has been reclaimed. The shift is away from the open resources that
defined the early Internet to a world where a smaller number get to control
how resources in this space are deployed.
This change will have consequences. It will entrench the old against the
new. It will centralize and commercialize the kinds of creativity that the Net
permits. And it will stifle creativity that is outside the picture of the world
that those with control prefer.
If there were a reason why this change was necessary—if there were a rea-
son to believe the Net could not advance without it or would be harmed
without it—then I would support this shift, however reluctantly.
But no good reason has been given. We are marching backward, undoing
the architecture—both the legal and the technical—of the original Net
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 237

T H E F U T U R E O F I D E A S 2 3 7

without anyone demonstrating why this change is needed. We are moving
resources from the commons into a system of control without an argument
about why control will help or why the commons will fail. We are jumping
in the name of an ideology without any consideration of the facts of these
past ten years.
So why are we making these changes?
In part, there is a dark story to be told. Change threatens existing interests.
Channeling change is often the best strategy for preserving that threatened
power. Those whose position is threatened by the change the Internet rep-
resents have a strong interest in trying to channel that change.
So, too, do they have the means. It is an iron law of politics that the or-
ganized beat the unorganized and that the vested have interests that get
organized over the unknown. In Washington, decisions are made by repre-
sentatives conferring with people whose interests are affected by changes in
the Net. But who represents the innovations not yet made? Who demands
that the platform be kept open for them?
No one. Some of those interests can’t afford the negotiation with existing
interests; some interests don’t even yet exist.
The result is that the pressure in the existing system is biased in favor of
the old. Policies get made to favor the old; the interest of the new simply has
no voice.
But the larger story here is not about dark forces. It is about a blindness
that affects our political culture generally. We have been so captured by the
ideals of property and control that we don’t even see the benefits from re-
sources not perfectly controlled. Resistance to property is read as an en-
dorsement of the state. The challenge to extreme propertization is read as
the endorsement of nationalization.
In the context of intellectual property, the general problem is magnified
by another blindness, the error induced by thinking of intellectual property
as property. By simplifying the nature of the rights that IP law protects, by
speaking of it as property, just like the ordinary property of cars and homes,
our thinking is guided in a very particular way. When it is viewed as prop-
erty, we see endless arguments for strengthening IP and few for resisting that
increase.
This is not conspiracy. It is a cultural blindness. We have forgotten what
the Framers of the American Constitution knew about the nature of IP, and
hence we have lost the balance our Framers had in protecting IP.
The consequence of these three stories is a massive push in cyberspace
right now to reestablish systems of control. This push is happening at all
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 238

2 3 8 L a w r e n c e L e s s i g

three of Benkler’s communications layers: it is happening at the physical
layer—as the architecture for wired access gets pushed to an architecture of
control and as the spectrum carrying wireless content gets sold into a system
of control; it is happening at the code layer, as the legal system through
patents favors closed code over open and as content providers build the ar-
chitecture that will enable them to have more perfect control over content
on the Net; and it is happening at the content layer, as the rules, both tech-
nical and legal, for facilitating distribution of content increasingly favor
control of that distribution over the free flow of the original Net.
This shift is a step back, to a system of creativity and distribution that is
largely within the control of large commercial actors. But this time, this sys-
tem of control does not have the relatively neutral justification of econom-
ics behind it. The constraints that make this control necessary are not the
constraints built into the nature of real-space scarcity. Here, instead,
the scarcity is largely artificial. The expansion of IP rights creates a certain
scarcity, but to the extent that expansion is beyond what is needed to induce
progress, it is an unnecessary and unjustified handout to existing interests.
To the extent the architecture gets built to reestablish the power to control
distribution, and thereby innovation, it is a constraint that is not demanded
by economics. It is a constraint that simply favors some interests over others.
Because our bias is to ignore the choice between the free and the con-
trolled, we ignore the costs of a system of control over a system that remains
free. We fail to see the benefits from freedom because we assume that free-
dom is not possible. We assume that creativity and innovation and growth
will occur only where property and markets function most strongly.
Against this ideology, I offer the Internet. Against this bias, I submit a tra-
dition that has understood balance better. The past decade has demon-
strated the value of the free; that freedom came from the Net’s architecture.
The changes we are now seeing simply ignore the value of the free; they get
implemented by changing the architecture of the original Net.
In Code, I argued that the original Net protected fundamental aspects of
liberty—free speech, privacy, access to content, freedom from excessive
regulation; I argued that those freedoms flowed from the architecture of the
Net; but, I argued, that architecture was not fixed; nothing guaranteed it
would survive unchanged; and, in fact, this architecture was being changed
in ways that took away some of these fundamental liberties.
I’ve now told the same story about innovation and the Net. The original
Net protected fundamental aspects of innovation. End-to-end meant new
ideas were protected; open code meant innovation would not be attacked;
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 239

T H E F U T U R E O F I D E A S 2 3 9

free distribution meant new ways of connecting would be assured. These
protections were largely architectural. This architecture is now changing.
And as it changes, as with the threats to liberty, there is a threat here to in-
novation.
In both cases, the conclusion is the same. We can architect this space in
any number of ways. Some architectures protect liberty, others do not.
Some architectures protect innovation, others do not. But nothing assures
that the first version of the Net’s architecture will survive through its matu-
rity. Indeed, in both cases, there is all the pressure in the world to say that
the first version will not survive.
As the old Net gets replaced by the new, as old interests succeed in pro-
tecting themselves against the new, we face a fundamental choice. We can
embrace this return to the architecture of creativity that has defined modern
American life—perpetual control by homogeneous corporations of a system
for creativity focused primarily on a mass audience. Or we can embrace the
architecture the Net was. This is a choice we cannot avoid, and in the next
chapter, I offer points of resistance to the trend we now see.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 240

A l t . C o m m o n s

T here are changes that could be made. This march backward is neither
necessary nor complete. We still have the time to point policy in a differ-
ent direction. The question is whether we have the will. Are we willing to
set principles that will guide the next stage of the Internet’s evolution, or
will we allow those who have interests inconsistent with those principles to
exercise their control?
My aim in this chapter is to outline some of these changes. My list is nei-
ther complete nor certain. But it should be a start of a conversation about
returning the Net to the conditions that let innovation flourish.
I divide these proposals into the frame of Benkler’s three layers. Changes
in some might make changes elsewhere unnecessary, but changes in all of
them would make the situation much better.

T H E P H Y S I C A L L AY E R

the physical layer would seem the least likely for reform, since this layer
lives within real space only, and as I’ve argued, the constraints of real space
are, well, real.
But there are a number of places where a change might do some good.
And the most important follows from the argument considered in chap-
ter 12.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 241

T H E F U T U R E O F I D E A S 2 4 1

Free Spectrum

the physical layer includes technology upon which, or in which, or
over which the network lives. It includes the computers that connect to the
Net, the wires they connect to, the routers that feed those wires, and the
spectrum that substitutes for the wires.
As I have described, most of these elements are owned—and with one ex-
ception, I think properly so. Computers are private property, whether the
government’s (in the NSA or libraries) or individuals’. The wires, whether
copper or fiber, linking these computers to routers to other computers are
privately owned. Massive investment laid them; even greater investment has
been needed to bring them up to date.
These private investments deserve the reward of private property. With
one qualification that I will offer in the next section, the owners of this prop-
erty should be free to use it as they wish. No one should have the right to sit
at my machine. Access to my machine, and the wires of AT&T, should not
be free. If it were free, then those who buy the machines and those who lay
the wires would lose lots of the reason to buy the machines and lay the
wires. If access were free, the incentives to build the Net out would largely
be lost.
But the same virtues from control can’t be said of spectrum. Or at least
they can’t be asserted with the same confidence. No one builds spectrum;
no investment from AT&T has made it possible. No entitlement justifies
special control over spectrum, and the owners of Ryder Truck Rental and
Leasing should not have a monopoly right to control the highways just be-
cause they purchased expensive equipment to use it. Thus, any monopoly
control over spectrum should be allowed only if it can be shown that mo-
nopoly control is needed.
My argument is not that exclusive control is not needed. It may well be
that Hazlett is right. The congestion in the airwaves may push us to build
out a property system. Spectrum auctions—either in advance or in real
time—may turn out to be needed to use the spectrum in the best possible
way.
But we don’t know that yet, and we certainly don’t know enough yet to
know how spectrum will be used. Thus, rather than architecting the space
exclusively for control, we should begin, as much as possible, as we be-
gan with the Internet: by building a regime that by design leaves a signifi-
cant part of these resources in the commons. And once we see how that
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 242

2 4 2 L a w r e n c e L e s s i g

commons gets used, we can then change how that commons gets con-
trolled.
This argues for a dual strategy. We should be setting aside broad swaths of
spectrum as a commons, intermixed with spectrum as property. The market
should be assured that both models will survive for the next chunk of time.
And regulators should assure that the devices using the commons are smart
enough not to spoil the space.1
What would this idea look like in detail? First, spectrum is not all the
same. The spectrum the AM radio uses can’t do what the spectrum that
radio astronomy uses does. Thus, we can’t simply carve off a single chunk of
spectrum for open, unlicensed use. We must instead set off significant bands
at each spectrum level, to assure that innovation for different uses of spec-
trum would be possible.
Second, we should force the government to give up its obscenely waste-
ful hoarding of spectrum. When radios were stupid and clear channels were
necessary, this hoarding made sense. But the government is not using this
spectrum with stupid radios. The most advanced work being done in
“software-defined radios”—radios that would, like chameleons, change
their character to fit the protocol in the context that works best—is being
done by the same group that gave us the Internet—the Defense Advanced
Research Projects Agency (DARPA). DARPA is researching software-defined
radios that share spectrum smartly. It is, in other words, building the Inter-
net in the air.2
I’m a great fan of DARPA’s work, but I don’t believe the government
should have a monopoly on innovation around networks. A sensible policy
would divide spectrum into many swaths, some controlled, some free. Users
of the spectrum—meaning the machines that use the spectrum—would
then decide which chunks work best for them. Indeed, with software-
defined radios, they could make that decision many times over. Out on the
highway, a cell phone could connect to a network in the best of several dif-
ferent ways. Some of those ways might use the free space of unlicensed spec-
trum. Some might log on to controlled spectrum property. But the decision
would be a function of which choice made most sense at the moment.
Third, there is no doubt that a sensible spectrum policy for the future will
require changes in the way spectrum is used today. Steel mills didn’t get a
permanent waiver from pollution laws just because their technology was
old.3 And pollution is precisely the way we should think about old uses of
spectrum: large and stupid towers billow overly powerful broadcasts into
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 243

T H E F U T U R E O F I D E A S 2 4 3

the ether, making it impossible for smaller, quieter, more efficient uses of
spectrum to flourish. Why should these smokestack technologies get pro-
tection, when the steel mills did not? Why not force them to improve their
technology—to reduce the pollution they spew forth into the ether—so that
others could innovate in yet unimagined ways?
These changes may well force familiar uses of the spectrum off of spec-
trum completely. Broadcast television, for example, is an extraordinary
spectrum guzzler; in most contexts it would be best moved from the air to
wires. This is just another instance of what has been called “the Negroponte
switch”—that everything that used wireless spectrum would move to wires,
and everything using wires would move to the air. Such movement will im-
pose costs on some—just as rebuilding smokestacks imposed costs on Pitts-
burgh. But these costs in the short run would be easily outweighed by
benefits in the long run. By establishing a dual system of free and controlled
spectrum, each with an equivalent opportunity to demonstrate its own suc-
cess, we could ensure future spectrum innovation.
Existing spectrum users will resist these changes. It is unfair, they will
argue, to change the rules now. They paid good money, they will insist, for
the monopoly control they have purchased. It is wrong for the government
to go back on promises it made.
But this argument is absurd. First, technically, this change would not
breach any promise. From the very beginning of spectrum allocation, the
recipients have been required to affirm that they asserted no property inter-
est in the spectrum rights they got.4 From the very beginning, the express
understanding has been that the licenses granted were limited in time and
that the risk that spectrum would be allocated differently was a risk these
businesses should take into account. For corporations now to claim that,
these statements notwithstanding, they expected to have their monopolies
in perpetuity is just absurd.
That this argument is being made, however, does signal the importance
of the government’s acting soon. Rather than racing to auction off more
spectrum in the way it has to date, and thereby increasing the sense of entitle-
ment that these spectrum hoarders claim, the government should be mark-
ing off much wider areas of spectrum that will be dedicated, for a significant
period, to the commons. The government could free existing holders of
spectrum to sell their leaseholds to others—thereby facilitating the market
that Hazlett wants. But before it puts any more of this resource into a system
of control, it should assure that the commons in spectrum is properly secured.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 244

2 4 4 L a w r e n c e L e s s i g

Finally, the FCC should free up greater access to existing unlicensed
bands, including the Amateur Radio Service bands. Existing users here too
should not have the power to veto something new simply because it is a late-
comer.5
If there were a rich and developed physical layer of free spectrum
access—permitting many competitors to offer Internet access using this
final link of free spectrum—then the need for the government to worry
about other modes of access to the Net would be lessened. If a rich and pow-
erful channel is kept open and kept in the commons, then what owners of
other channels, not left in the commons, do is less of a concern. The key is
balance, but our practice now is at the extreme.

Free Highways

opening channels of spectrum would be one important expansion of
the commons. It isn’t, however, the only possible change. There are other
steps the government might take to open access at the physical layer—not
by becoming network owners, but by clearing the way for others to develop
and run networks.
The model here is the highway system. While we have grown skeptical
about the state’s role in many aspects of our life, there remains a strong be-
lief that the state has a place in the provision of basic services like roads. The
government has funded the construction of highways and local roads; these
highways are then used either “for free” or with the payment of a toll. In ei-
ther case, the highway functions as a commons.
In some cases there has been a move to privatize highways. Los Angeles,
for example, has experimented with private roads, promising lower conges-
tion in exchange. But these experiments exist alongside open, public roads.
The two go together, and where only one road is available, it is a public
one.
This is a sensible use of public resources. It builds a resource for ex-
change that all have access to. It balances resources controlled privately; it
sets a baseline against which private resources compete.
In my view, the same attitude should guide government policy with re-
spect to the physical infrastructure supporting the Net—in particular, the
fiber infrastructure and rights-of-way to deploy alternative access. Just as the
state has spent resources on building highways, so too might it need to
spend resources on building out the information superhighway.6
Many cities are already picking up on the idea. Chicago, for example, has
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 245

T H E F U T U R E O F I D E A S 2 4 5

started funding the deployment of “dark fiber”—meaning simply fiber optic
cable that is not immediately connected to any particular service.7 The city
does not get into the business of wiring the links or running the routers; the
city’s job is simply to fund the laying of the cable and let competitors con-
nect to the fiber to sell Internet access.8
The advantage in this way of building the basic physical infrastructure
upon which the network is built is that there is no need to strike a bargain
with the monopoly devil to finance the deployment of this fiber. But with
cable, the government decided to fund the build-out with the grant of mo-
nopolies to the cable owners. That strategy proved costly, as the cable own-
ers became invested with eagerness to defend their monopoly. The aim of a
monopolist is always to protect his monopoly, and if he’s rational, he’s will-
ing to spend the net present value of his monopoly to defend it.
But if the state were funding the building of at least some of this basic
infrastructure, and if it were kept far from the content that gets played across
this infrastructure, then there would be no actor in this system with an in-
centive to discriminate. Government would lay the pipes; private industry
would use the pipes to serve access to local customers.
This point would not stop with dark fiber. Right now the most expensive
part of the connection is the “last mile” to the consumer’s home. That’s ex-
pensive in part because the technology for converting from glass to elec-
tronics is itself expensive and in part because of the cost of trenches. If every
house had to be rewired (or fibered), the cost of running fast Internet to
every house would be very high.
One alternative would be to run the fiber to a wireless broadcasting sta-
tion that then would beam Internet service to many users in the neighbor-
hood. (This would be the inverse of the system that gave us cable TV.) But
to do this, providers would have to have access to telephone poles or other
places where broadcasting stations might be built.
Here again the state could play a role—either by granting access to state
property or by purchasing access to private property. This access would
amount to permission for a private company to build a broadcasting service;
it would not be an invitation for the state to get into the ISP business. And
because the capacity of these fibers is so great, the potential for a bottleneck
would be much less.
In both cases, the role the state plays is to assure that bottlenecks not
become opportunities for exercising market power. The state builds a
competitive environment where there is incentive to behave as a competi-
tor. This commitment will require resources; it would be less pressing if the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 246

2 4 6 L a w r e n c e L e s s i g

FCC freed spectrum generally. But it is an appropriately translated role for
the state when superhighways carry bits rather than trucks.

T H E C O D E L AY E R

the code layer is the heart of the Internet; its particular architecture is
what was special. It would make no sense, however, to say that we should fix
ourselves to the particular architecture of the network at any one time. The
point is not a blind originalism; the point instead is to preserve the values ex-
pressed by that original architecture. Some ways to do that are considered
here.

Neutral Platforms

the critical layer to protect if we are to protect innovation on the Net
is the code layer—the space where code decides how content and applica-
tions flow, and where code could control how innovation develops. It is at
this layer that the Internet originally embraced the principle of end-to-end.
That principle assured that control was bottom up; that what would succeed
would succeed because users demanded it; and that what users demanded
would be free to flow to them.
It is the compromise of this principle that threatens the greatest harm to
innovation. And the pressure to compromise comes from those who would
use their power over architecture to protect a legacy monopoly. The danger
exists when control over the platform can translate into the power to protect
against new innovation.
We’ve seen this power in two different contexts. The claims the govern-
ment made against Microsoft were the clearest example of this danger:
Microsoft, the government argued, used its power over the Windows plat-
form to protect the Windows platform from innovation that would threaten
it. Likewise in the context of broadband cable: the danger was that by re-
stricting the number of ISPs, the network owner could exercise control over
the ISPs to assure that its content, or business model, would not be threat-
ened by certain uses of the Net.
This power, whether exercised over the operating system or over the net-
work within which Internet traffic flows, threatens innovation. The risk of a
strategic response by the platform owner reduces the expected benefit from
innovating in certain status quo–threatening ways. Thus, if innovation is
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 247

T H E F U T U R E O F I D E A S 2 4 7

our goal, our policy should be to minimize the threat of this strategic be-
havior. A number of strategies to that end are obvious.
First, the government should encourage the development of open code.
Open code, as I’ve argued, risks none of the dangers of strategic behavior
that closed code, or controlled networks, do. If open code is used strategi-
cally, then the resources to counter that strategic action are always available.
Innovators can rely upon the promise of open code in their innovations.
They need not worry that what they develop will be swallowed by the plat-
form they develop for.
This encouragement should not be coercive. There’s no reason to ban or
punish proprietary providers. People should be free to develop code how-
ever they wish. But a government has its own interests, and closing its re-
sources to others is not one of them. If the federal government develops a
system to handle welfare claims, what reason does it have for hiding the
code for that system from the states? Why not let the states take that code
and build upon it? And if the states, then so, too, with the universities. In
each case, the aim should be to expand the reach of these powerful and
valuable resources, not to contract and hoard them when no value to the
hoarding exists.
Likewise with the government’s choice of operating systems. What reason
does the government have for supporting closed code, when open code is
as powerful and the externalities from using open code would benefit oth-
ers? If the PCs that the government owned ran something other than Win-
dows, then the market for these alternative platforms would be wildly
expanded. And if the market for alternatives were strong, then the benefits
from building for these alternatives would be strong as well.
Again, such a strategy does not flow from animus against proprietary
providers. Any environment is richer when built upon a diversity of life. As
the various viruses that have plagued the Internet have shown, we are in-
creasingly vulnerable the more concentrated the strain. Opening the code
that constitutes cyberspace in as many contexts as possible will enable a
flourishing of innovation that need not depend upon a single platform.
Microsoft may not like this, but a policy favoring this diversity is not a policy
against Microsoft.9
Second, the government should continue to ensure that no major player
in the Internet space is able to architect the Internet space to empower its
own strategic behavior. If the cable companies want to build a cable tele-
vision system where they have complete control over the content that flows
across their cable, more power to them. Control has been at the core of the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 248

2 4 8 L a w r e n c e L e s s i g

cable system from the start. But if cable wants to carry TCP/IP, then the val-
ues of the Internet should trump the control of cable. Any major network
that wants to piggyback on the Internet’s success should piggyback with the
values of the Internet kept in mind.
How best can the government carry this role into effect? Historically, the
most successful strategy has been banishment. When the government, for
example, banished the telephone company from the game of providing
computer services, then the telephone company had little interest in play-
ing games among different providers of computer service. If it were simply
in the business of selling access to pipe, then those buying that access would
have an equal playing field for competing in providing network services.
Since the 1996 Telecommunications Act, the strategy has been different.
The government has required the telephone company to compete against
ISPs. The telephone company, for example, is permitted to sell you DSL,
but it must permit competitor DSL providers to offer you DSL service
as well. This is the “open access” requirement. Competition is assured by
regulating the Bell companies and by requiring them to unbundle local ac-
cess services. And this unbundling in turn assures that the telephone com-
pany can’t play any games.
In this way, the regulation assures that competitive pressures will exist for
every mode of broadband access, so that broadband providers won’t have
the incentive to exercise control in a strategic way. And until enough com-
petitors are on the competitive field, this may well be the best strategy for
keeping access to the Internet open.10
Alternatively, we might imagine a simpler regulatory strategy. If the con-
cern at stake is that network providers will leverage control over the network
into some control over content—if the concern is that they will have an in-
centive to compromise the principle of end-to-end—then rather than re-
quiring unbundling of services, the government could adopt a more direct
regulatory strategy: if you provide Internet services, then you must provide
them consistent with the principle of end-to-end.11
It is hard in the abstract to game that alternative. DSL would compete
with cable, fiber, and wireless service in providing Internet traffic. The lat-
ter would not be required to facilitate competition on their own facilities,
but the competition among facilities might be enough in most places to en-
sure that prices are kept low. As long as discrimination is not enabled (the
consequence of preserving end-to-end), the essential elements of a com-
mons would be preserved as well.
My bias is in favor of the least invasive regulatory response, but it is also a
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 249

T H E F U T U R E O F I D E A S 2 4 9

bias in favor of a guarantee of a regulatory response if regulation is needed.
The trends have not been toward fast networks that promise uncontrolled
access. The trends have been toward control.
One useful point of comparison here is our neighbor to the north,
Canada. The Canadians have required open access for broadband pro-
viders, and a recent Report of the National Broadband Task Force has
endorsed a “bill of rights” for broadband users that assures continued con-
sumer choice for any build-out of a network employing government funds.
This choice of policy has apparently not harmed Canadian access. Accord-
ing to a recent OECD report, broadband connections are twice as common
in Canada per capita as in the United States.12
Finally, and at a minimum, regulators should begin to evaluate changes
to the network in terms of the neutrality of end-to-end. We should begin to
think about the trade-offs between control and neutrality explicitly—not
because every trade-off would be a sin, but because otherwise, the loss will
be invisible.

T H E C O N T E N T L AY E R

the changes at the physical and code layers are significant. They will re-
quire a commitment that I am skeptical our politicians are capable of giv-
ing. But they are nothing compared with the changes that are required at
the content layer. For it is here that we have moved the furthest from sensi-
ble policy, and here where there is the strongest political power to resist.
The core idea that we, as a culture, must recapture is that control over
content is not to be perfect. Ideas and expression must to some degree be
free. That was the aim of copyright law initially—the balance between con-
trol and freedom. It was, “even twenty years ago . . . an article of faith [that
it] offer only circumscribed, porous protection.”13 But these balanced laws
now have an ally that threatens to destroy the balance: code.
Technology, tied to law, now promises almost perfect control over con-
tent and its distribution. And it is this perfect control that threatens to
undermine the potential for innovation that the Internet promises.
To resist this threat, we need specific changes to reestablish a balance be-
tween control and creativity. Our aim should be a system of sufficient con-
trol to give artists enough incentive to produce, while leaving free as much
as we can for others to build upon and create.
In setting this balance, there are a few ideas to keep in mind. First, we live
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 250

2 5 0 L a w r e n c e L e s s i g

in a world with “free” content, and this freedom is not an imperfection. We
listen to the radio without paying for the songs we hear; we hear friends
humming tunes that they have not licensed. We refer to plots in movies to
tell jokes without the permission of the director. We read books to our chil-
dren borrowed from a library without any payment for performance rights to
the original copyright holder. The fact that content at any particular time is
free tells us nothing about whether using that content is “theft.” Similarly,
an argument for increasing control by content owners needs more than
“they didn’t pay for this use” to back it up.
Second, and related, the reason perfect control is not our aim is that
creation is always the building upon something else. There is no art that
doesn’t reuse. And there will be less art if every reuse is taxed by the earlier
appropriator. Monopoly controls have been the exception in free society;
they have been the rule in closed societies.
Finally, while control is needed, and perfectly justifiable, our bias should
be clear up front: Monopolies are not justified by theory; they should be
permitted only when justified by facts. If there is no solid basis for extending
a certain monopoly protection, then we should not extend that protection.
This does not mean that every copyright must prove its value up front. That
would be a far too cumbersome system of control. But it does mean that
every system or category of copyright or patent should prove its value up
front. Before the monopoly should be permitted, there should be reason to
believe it will do some good—for society, and not just for monopoly holders.
With these ideals in mind, here are some first steps to freeing culture:

the trend in copyright law has been to increase copyright’s scope and
duration, while making the right easier to secure and keep. While the origi-
nal copyright statutes put a great burden on copyright owners to register
their work, make deposits of their work to the government, and renew the
copyright after an initial term, now copyright affixes automatically, it ex-
tends for the life of the author plus seventy years without any effort by the
copyright owner, and the copyright owner need make no effort at all to con-
tinue to enjoy this government-granted monopoly.
This shift is bizarre. We have been pushed to this “no effort” monopoly
handout by the view that “technical” requirements should not interfere with
the right of an author to his or her copyright. That argument sounds good
until one considers the other side of the bargain—the public. Copyright
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 251

T H E F U T U R E O F I D E A S 2 5 1

owners should not be denied legitimate copyright protection for technicali-
ties, no doubt; but assuring that the reach of state-backed monopolies over
speech is not broader than necessary is not a “technicality.” If welfare recipi-
ents can be denied their benefits because they fail to complete a benefits
form properly, then I can’t see the unfairness in requiring those who de-
mand state support to defend their monopoly similarly by filling out a regis-
tration form.
I would go even further.14

f i v e - y e a r r e n e wa b l e t e r m s

authors and creators deserve to receive the benefits of their creation.
But when those benefits stop, what they create should fall into the public
domain. It does not do so now. Every creative act reduced to a tangible me-
dium is protected for upward of 150 years, whether or not the protection
benefits the author. This work thus falls into a copyright black hole, unfree
for over a century.
The solution to this black hole of copyright is to force those who benefit
from copyright to take steps to protect their state-backed benefit. And in the
age of the Internet, those steps could be extremely simple.
Work that an author “publishes” should be protected for a term of five
years once registered, and that registration can be renewed fifteen times. If
the registration is not renewed, then the work falls into the public domain.
Registration need not be difficult. The U.S. Copyright Office could run
a simple Web site where authors could register their work. That Web site
could be funded by charges for copyright renewals. When an author wants
to renew the copyright, the system could charge the author a renewal fee.
That fee might increase over time or depend upon the nature of the work.
This registration site could also take deposits of certain kinds of work. For
archive purposes, the site could collect digital copies of all the works copy-
righted. And for certain kinds of work—software in particular—those de-
posits would be required for the protection to be secured. Given how easy
the Net has made such transfers, these costs would be relatively small.
“Unpublished works” would be different. If I write an e-mail and send it
to a group of my friends, that creativity should be treated differently from
the creativity of a published book or recorded song. The e-mail should be
protected for privacy reasons, the song and book protected as a quid pro quo
for a government-backed monopoly. Thus, for private, unpublished corre-
spondence, I think the current protection is perfectly sensible: the life of the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 252

2 5 2 L a w r e n c e L e s s i g

author plus seventy years, automatically created, with no registration or re-
newal requirements.
One of the strongest reasons that the copyright industry has raised for the
elimination of this renewal requirement is the injustice that comes from a
family’s or author’s losing copyright protection merely because of a techni-
cality. If “technicality” means something like the registration was lost in the
mail or was delivered two hours late, then the complaint is a good one.
There is no reason to punish authors for slips. But the remedy for an overly
strict system is a more relaxed system, not no system at all. If a registration is
lost or a deadline missed by a short period of time, the U.S. Copyright Of-
fice should have the power to forgive.
A change in the copyright term would have no effect on the incentives for
authors to produce work today. There is no author who decides whether or
not to write a book depending upon whether he or his estate will receive
money three-quarters of a century from now. The same with a film pro-
ducer: Hollywood studios forecast revenues a few years into the future, not
ninety-five. The effect on expected income from this change would there-
fore be tiny.
But the benefit for creativity from more works falling into the commons
would be large. If a copyright isn’t worth it to an author to renew for a modest
fee, then it isn’t worth it to society to support—through an array of criminal
and civil statutes—the monopoly protected. But the same work that the origi-
nal author might not value could well be used by other creators in society.
Even more significant, this repository of data about what work is pro-
tected would lower the costs of licensing copyrighted works significantly.
Because the database would be relatively fresh—with a requirement that
contact information be kept current—creators who want to license other
creators’ work would have a simple tool to do so.

s o f t wa r e c o p y r i g h t

software is a special case. The current protection for software is the life
of an author plus seventy years or, if a corporation, ninety-five years. This is
a parody of the Constitution’s requirement that copyright be for “limited
times.” When Apple Macintosh’s operating system falls into the public do-
main, there will be no machine that could possibly run it. The term of
copyright for software is effectively unlimited.
Worse, the copyright system protects software without getting any new
knowledge in return. When the system protects Hemingway, we at least get
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 253

T H E F U T U R E O F I D E A S 2 5 3

to see how Hemingway writes. We get to learn about his style and the tricks
he uses to make his work succeed. We can see this because it is the nature
of creative writing that the writing is public. There is no such thing as lan-
guage that doesn’t simultaneously transmit its words.
Software is different. As I’ve described, software is compiled; the com-
piled code is essentially unreadable; but to copyright software, the author
need not reveal the source code. Thus, while an English Department gets
to analyze Virginia Woolf’s novels to train writers in better writing, the
Computer Science Department doesn’t get to examine Microsoft’s operat-
ing system to train its students in better coding.
The harm from this system of protecting creativity is greater than this loss
to computer science education. While the creative works from the sixteenth
century can still be accessed and used by others, the data used by software
programs from the 1990s are already inaccessible. Once a company pro-
ducing a certain product goes out of business, there is no simple way to
uncover how its products encoded data. The data are thus lost, and the soft-
ware is unusable. Knowledge has been destroyed.
The reason copyright law doesn’t require the production of source code
is that it is believed that that would make the software unprotectable. The
open code movement might throw that view into doubt, but even if one be-
lieves it, the remedy (no source code) is worse than the harm. There are
plenty of ways for software to be protected without the protection of law.
Copy protection systems, for example, give the copyright holder plenty of
control over how and when the software is copied.
If society is to give software producers more protection than they other-
wise would get through technology, then we should get something in re-
turn. And one thing we could get would be access to the source code after
the copyright expires. Thus, I would protect software for a term of five years,
renewable once. But that protection would be granted only if the author
submitted a copy of the source code to be held in escrow while the work was
protected. Once the copyright expired, that escrowed copy would be pub-
licly available from the U.S. Copyright Office server.15

protecting innovation

the single most striking feature of copyright law is its ability to give own-
ers the power to control innovation in the context of the Net. The power
to issue injunctions against technologies for distributing content is an ex-
traordinary power for controlling what new technologies can be created.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 254

2 5 4 L a w r e n c e L e s s i g

This is the power that Hollywood has used to control the development of
new distribution technologies. Without any showing of harm, the industry
has been able to exercise control over new modes of distribution.
No one industry should have the power to veto the Net’s development.
No single set of interests should be able to decide what innovations are best.
This is especially true when no showing of harm is necessary for this veto to
be imposed. And this is precisely the reality with respect to copyright and
the Net.
Congress should limit this reactive character of copyright law. While in
the ordinary case the copyright holder should not have to prove harm before
enforcing a copyright, in a context of significant technological change, a de-
fendant should at least have the opportunity to show that the copyright
holder will suffer no harm.16

protecting music

the net has created a world where content is free. Napster is the most
salient example of this world, but it is not the only one. At any time a user
can select the channel of music he or she wants. A song from your child-
hood? Search on the lyrics and find a recording. Within seconds you can
hear any music you want.
This freedom the recording industry calls theft. But they don’t call it theft
when I hear an old favorite of mine on the radio. They don’t call it
theft when they are recording takeoffs of prior recorded music. And they
don’t call it theft when they make a new version of “Jingle Bells.” They
don’t, in other words, call it theft when they are using music for free in ways
that have been defined by the copyright system as fair and appropriate uses.
The issue we must confront is whether this free distribution should con-
tinue to be free. And the solution to that question is to keep an important
distinction in mind: As we’ve seen, there is a distinction between music
being “free” and music being available at zero cost. Artists should be paid,
but it doesn’t follow that selling music like chewing gum is the only possi-
ble way.17
Here, too, a bit of history helps. As I have described, there have been
many contexts where Congress had to balance the rights of free access
against the rights of control. When the courts said piano rolls were not
“copies” of sheet music, Congress balanced the rights of composers against
the rights to mechanically reproduce what was composed. It balanced these
rights through a compulsory license that enabled payment to artists while
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 255

T H E F U T U R E O F I D E A S 2 5 5

assuring free access to the work produced. The same is true in the context
of cable TV. As we saw in chapter 7, the Supreme Court twice said that
cable TV providers had a right, under existing law, to free TV. Congress fi-
nally changed those rights, but again, in a balanced and sensible way. Cable
providers got access to television broadcasts, but broadcasters and copyright
holders had a right to compensation for that access. This compensation
again was set by a compulsory licensing term. Congress protected the au-
thor, but not through a property right.
The same solution is possible in the context of music on the Net.18 But
here, rather than balance, the rhetoric is about “theft” and “crime.” But was
it “theft” when cable TV took television broadcasts?
Congress should empower file sharing by recognizing a similar system of
compulsory licenses. These fees should not be set by an industry intent on
killing this new mode of distribution. They should be set, as they have al-
ways been set, by a policy maker keen on striking a balance. If only such a
policy maker were somewhere to be found.

rebuilding the creative commons

these changes would affect works produced in the future. They would
have no effect on works already produced and protected by these extensive
terms. Nor could these changes affect works already produced—the Con-
stitution limits Congress’s power to take away property already granted, as
well it should.
But there are other ways that Congress might act to create an incentive to
build out the creative commons. One way would be to create incentives for
holders of copyright to donate their holdings into a public conservancy. I’ve
worked with others to build such a conservancy, but ours is not the only pos-
sible one. If Congress gave tax benefits to donors of IP to parallel the tax
benefits given to donors of art, then there would be a much greater incen-
tive to donate works to the general weal.19
One context in particular where incentives could do some good is that of
orphaned software. Companies often decide that the costs of developing or
maintaining software are higher than the benefits. They therefore orphan
the software, by neither selling it nor supporting it. They have little reason,
however, to make the source code for that software available to others. The
code therefore simply disappears, and the products become useless.
If Congress created an incentive for these companies to donate their code
to a conservancy, then others could build on the earlier work to produce up-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 256

2 5 6 L a w r e n c e L e s s i g

dated or altered versions. This in turn could improve the software available
by avoiding the loss of knowledge that was built into the original code. Or-
phans could be adopted by others who saw their special benefit.
There are other steps that Congress could take as well. One problem that
often plagues creators is the claim that a work is copyrighted when it is not.
It is a common practice for publishers, for example, to claim copyright
when under the law they plainly do not have such a right. Sheet music pub-
lishers, for example, will often put copyright notices on public domain
works.
This practice is in violation of existing copyright law. It is a crime to say
something is copyrighted when in fact it is not. But the only punishment
that the law provides for this crime is an action brought by a U.S. attorney.
Not surprisingly, U.S. attorneys have better things to do; no one has ever
been prosecuted for violations of this.20
This is a perfect claim for private attorneys general. Professor David
Lange suggests that “claims . . . so extravagant in relation to the reality from
which . . . they . . . spring . . . ought to be made the subject of a serious
counterclaim for punitive damages rooted in some sort of tort . . . to be
termed ‘unconscionable overreaching.’ ”21 Congress could authorize pri-
vate citizens to bring suits against false copyright claims. If successful, the
plaintiffs in these suits should earn a bounty, plus their costs. And if publi-
cized, such suits should change the behavior of publishers.

limits on code

as i’ve described, the interests that copyright law protects can be protected
by technology as well as law. As the DVD case discussed in chapter 11
makes clear, this protection by technology can often reach far beyond the
protection of law. Copyright protection systems can limit fair use or extend
the term of protection beyond the copyright term. If there is a reason for the
balance in copyright law, then there is a reason to be concerned about the
imbalance created by this code.
How one thinks about the code on its own is a hard legal question. We
don’t ordinarily think that there is an affirmative wrong in these private sys-
tems of protection. But where the law backs up this code, then there is clear
reason to be concerned. Then the protection that the copyright gets is not
just the private protection of code; it is also the protection of law. So here,
in my view, there is a clear reason to limit the protections of law. Congress
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 257

T H E F U T U R E O F I D E A S 2 5 7

should expressly require that any law protecting copyright protection sys-
tems not protect those systems beyond the reach of copyright law itself.
Only those code protection systems, in other words, that preserve adequate
room for fair use would get the protection of Congress’s law. (This is a ver-
sion of what has been called the “Cohen Theorem,” named after law pro-
fessor Julie Cohen: that one has a right to hack copyright protection systems
in order to secure fair use.)

limits on contract

the same point can be made about contract. Often a copyrighted work is
sold or licensed subject to a set of terms imposed in a license. Sometimes
the terms imposed by the license are inconsistent with the balance that
copyright law aims for. If the balance in copyright law is important, then it
should not be undermined by a different kind of law—contract law. While
not every license is in conflict with copyright law, many licenses are in con-
flict with the limited protection copyright law is to give.
State laws in particular that give copyright holders greater power than the
balance copyright law was to set should be resisted. Among these, the most
troubling is a uniform law making its way through the states called the “Uni-
form Computer Information Transactions Act”—or UCITA.
UCITA is designed to facilitate transactions in computer information. Its
aim is to make on-line contracting easier. But as many have argued (quite
convincingly, in my view), the balance that UCITA strikes between the
seller and the consumer is not adequate to protect the balance the law in-
tends between copyright and the commons.22
The opposition to UCITA has been strong; the need to adopt a uniform
code right now has not been demonstrated. We have very little understand-
ing of what on-line contracts could be. Yet UCITA is marching across the
states with the aim to settle a range of issues up front.
This is a mistake. The premise of state uniform laws is that they are to re-
flect and codify mature and settled understandings of the law. They are not
to be leaders; they are simply to clarify and to make uniform where unifor-
mity matters. In this case, the law has been used to expand the rights of one
side of a deal relative to another. The process has been captured. The
proper response to this capture is to reject the proposal now and wait to see
what’s needed later.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 258

2 5 8 L a w r e n c e L e s s i g

limit commercial exploitation

professor jessica Litman has taken these recommendations one step
further. As she rightly observes, following the work of Professor L. Ray Pat-
terson,23 copyright was originally simply a restriction on commercial enti-
ties. It regulated “publishers” and those who “vend” “maps, charts, and
books.” Because the law slipped into using the term copy in 1909, it has now
extended its reach to every act of duplication, by printing press or computer
memory. It now therefore covers actions far beyond the “commercial” ex-
ploitation of anything.
Litman therefore argues that we take copyright back to its roots. “[W]hy
not,” she writes, “start by recasting copyright as an exclusive right of com-
mercial exploitation? Making money (or trying to) from someone else’s
work without permission would be infringement, as would large scale inter-
ference with the copyright holders’ opportunities to do so.”24 Thus, she
argues, we should redraw the border between commercial and noncom-
mercial exploitation, giving authors strong control over the “pirating” of
their work by commercial entities, but leaving noncommercial actions out-
side the reach of the law.
There is a great merit to this idea and, in my view, good reason to explore
it extensively. The strongest reason to be skeptical is that the Net itself has
now erased any effective distinction between commercial and noncommer-
cial. Napster no doubt is a commercial activity, though the sharing that
Napster enables is not. The law might well regulate Napster, but then sub-
sequent p2p technologies would enable the same sort of sharing, just with-
out commercial links. If there were a harm from Napster (and I’m not
certain there is), it’s not clear what benefit we gain from merely pushing the
control underground.
This line-drawing problem reinforces my own view that the better solu-
tion is simply to go back to the Framers’ notion of limited terms. The great
benefit of a technology such as Napster is the ability to get access to music
that is no longer made available commercially—songs from the 1930s or
1940s, for example, that still hang in copyright, but that the copyright hold-
ers don’t make available commercially. If copyright were returned to a
meaningfully “limited time,” then we wouldn’t need to worry so much
about drawing commercial vs. noncommercial distinctions. For five or
maybe ten years, commercial entities would hold these rights exclusively.
Beyond that, the music, like culture generally, would be freely available.
Litman’s suggestion does hint at a different limitation on the copyright.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 259

T H E F U T U R E O F I D E A S 2 5 9

We might call this the “use it or lose it” restriction.25 Once a work is pub-
lished, if a holder of a copyright does not continue to make it available com-
mercially, then others should have the right to exploit the work. Technically,
we could accomplish this balance by giving anyone a right, after a brief pe-
riod of exclusive control, to license the work under compulsory terms. The
terms of such licenses can’t be set here. Which would work best depends
upon lots of things we can’t know in the abstract. But the basic idea is that
once a limited monopoly right has been granted, there is no further reason
to allow a rights holder to hold up the content. This, like the need to renew,
would assure that work was quickly pushed into the public domain.

Patents

the urgency in the field of patents is even greater. Here again, patents
are not evil per se; they are evil only if they do no social good. They do no
social good if they benefit certain companies at the expense of innovation
generally. And as many have argued convincingly, that’s just what many
patents today do.
Our response should be empirical. Congress should demand of the U.S.
Patent Office that it perform a regulatory review of its patent regulation and
produce from that review a regulatory impact statement. In particular, it
should be required to perform an economic study to justify the most con-
troversial extensions of patent right now—business method and software
patents. If these forms of innovation regulation can’t at least meet the bur-
den of demonstrating that they are more likely to aid innovation than harm
it, then Congress should withdraw this form of monopoly protection.
In the meantime, there are smaller changes that Congress might make,
all designed to lessen the harm patents generally, and bad patents in par-
ticular, might cause.

moratorium

congress should enact a moratorium on the offensive use of these
questionable patents until this review is complete. While this study occurs,
innovators should be free to file for these patents. If someone tries to enforce
a patent, then a company should be allowed to defend against that enforce-
ment with patents secured in the ordinary way. But until there is a showing
of the good this system does, we should allow the courts to intervene in the
regulatory process.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 260

2 6 0 L a w r e n c e L e s s i g

damages

the greatest harm that patents create may well be independent of the
patents; the greatest harm may well be the harm caused by its enforcement.
The real power a patent holder has—whether he or she holds a valid or in-
valid patent—is the threat of an injunction to stop the use of the patented
technology, when damages would be an adequate remedy. Ordinarily, the
use of someone else’s “property” without negotiation is deemed adequate
justification for an injunction; but there is no reason we should have to con-
ceive of patents in this context as “property.” Inventors could get an ade-
quate return from the guarantee of reasonable royalties; users of patents
could be assured that their innovation would not be blocked by a guarantee
of a compulsory license.
This is not an easy idea to sketch briefly. To the extent that technologies
are rightfully patented, granting only a compulsory right can, in many con-
texts, defeat the important incentives that patents produce. Thus, it would
be a mistake to give up property protections for patents in all cases.
But you don’t have to give up property protections for patents in all cases
for it to make sense to adjust patents in the context of the Net. Here, where
innovation is sequential and complementary, giving users greater flexibility
would reduce the hold-up problem created by patent law.26

reform

finally, there’s a range of reform that has been pushed on the U.S.
Patent Office, much of it extremely valuable.27 Software patents are (rela-
tively speaking) new. At least, they are newer than software itself. For many
years, software could not be patented, which means that, for many years, the
U.S. Patent Office did not collect data about prior art in patents.
This fact combines with another to make patents in this field particularly
uncertain. When a person seeking a patent files the patent application, he
or she is to include in the application a list of known “prior art”—earlier in-
ventions that might be related to the invention for which a patent is sought.
As the rules are now, however, the applicant need report only what the ap-
plicant actually knows. This creates an ostrich incentive: if you’re responsi-
ble only for what you know, then you have an incentive to do very little
research.28
The law deals with problems like this in many contexts. The ordinary re-
sponse is a “negligence standard”: the applicant must file what he or she
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 261

T H E F U T U R E O F I D E A S 2 6 1

knows or should have known. This creates a strong incentive for the appli-
cant to discover relevant prior art. And it would help the U.S. Patent Office
make a judgment about whether a patent should be granted.
If Congress determines that business method patents are justified, it
should also consider the proposals of Jeff Bezos and Tim O’Reilly to grant
patent protection for business methods for only a very short period. Bezos
proposes five years, but an even shorter period may make sense.29 Network
technologies move so quickly that a longer period of protection is never
really needed; and whatever distortions this system might produce, they
could be minimized by shorting the period of protection.
Congress should also, and most obviously, radically improve funding for
the U.S. Patent Office and mandate fundamental improvements in the
functioning of that office.30
Finally, Congress should consider the proposals of Congressman Berman
to deny patents to business methods that are simply translations from real-
space inventions to cyberspace.31 Scholars at first thought these inventions
would be denied patent protection because of lack of inventiveness.32 But a
clear rule marking the lack of inventiveness has not been shown. 33

these changes are just beginnings, but they would be significant be-
ginnings if done. They would together go a great distance in assuring that
the space for innovation remains open and that the resources for innovation
remain free. They would commit us to an environment that would preserve
the innovation we have seen.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 262

W h a t O r r i n U n d e r s t a n d s

O rrin hatch is a conservative. The senior senator from Utah, former
chair of the Judiciary Committee, he is a critical force in practically every
sphere of the Senate. He was a candidate for president in 2000. And he is
admired by most, especially on the Right, as a principled politician and a
decent man.
But there’s something funny about Hatch. He betrays “policy
anomalies”—positions that can’t quite be explained on a simple left/right
scale. Some of the things that he believes in most are puzzles to many con-
servatives. And puzzles in a politician are trouble. Unpredictability is not an
asset in a political world where results cost lobbyists millions to buy.
Two of Hatch’s anomalies are at the core of this book. The first is his con-
cern about the market power and behavior of the Microsoft Corporation.
And the second is his affection for emerging technologies like Napster.
Hatch was a strong supporter of the Justice Department’s investigation into
Microsoft’s behavior; he is a strong skeptic of the power that music labels
have over innovation in the arts.
The pundits think they have an explanation for Hatch’s resistance to
Microsoft: Corel Corporation, which purchased WordPerfect. WordPerfect
had been the dominant word processor. It was a Utah-based company. As
with many leading technologies, WordPerfect fumbled the move to GUI
interfaces. Microsoft picked up the ball and ran far. Many attribute Hatch’s
skepticism about Microsoft to these sour grapes.
Hatch’s views on Napster are explained in a different way. Hatch is a mu-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 263

T H E F U T U R E O F I D E A S 2 6 3

sician. He has written and recorded many Christian songs. But you don’t
find the senator’s CDs in record stores; the recording labels were not much
interested in recruiting the senator from Utah. Thus, Hatch again may have
a motive to resent the labels. Therefore, when a new technology comes
along that threatens the power of the labels, it is Schadenfreude, not con-
cern, that drives the senator.
It is hard to believe that any politician does what he does for a reason of
principle. We live in an era when principled politicians are characters in
TV dramas; real politicians are something very different. Thus, the idea that
a successful senator would do something that might harm him politically
because of ideals strikes us as a fantasy. The stuff of Hollywood, perhaps, but
not of Washington, D.C.
But as this book has made clear, there is a principle that would explain
Hatch’s stand. And while I am no friend of Hatch’s, or of many of the poli-
cies that he has pushed, I do believe that what pushes Hatch to both posi-
tions is a matter of principle. Concentrations of power worry conservatives
like Hatch; and in both of these anomalies, concentration of power is at
stake.
In the Microsoft case, the fear is that this dominant controller of the plat-
form will be able to use its power to direct evolution. Power over the
platform will mean the ability to direct how the platform develops. And the
ability to direct how the platform develops is a dangerous power for any
single company to hold. It would be awful for the FCC to decide what
technologies should look like in the future, then force those technologies
on us through the power of law. But likewise, while it wouldn’t be as awful,
it is still fairly bad that any single company, whether by virtue of the law
or because of its control over a platform, could control how technology
should develop. Hatch is a believer in the diverse, decentralized market that
allows consumers to choose the future. Thus, though he is among the old-
est members of the Senate, his spirit is among the closest to what makes the
Net run.
The same can be said about the production of culture. Obviously, the
government has no legitimate role in controlling how our culture should
evolve. What music people listen to and what art they find compelling are
matters of private, not public, choice. But even if not as bad as it has been,
the world we now have controlling media in our country is worse than the
world that Hatch would want. The concentration of power that Hollywood
has permits Hollywood a power that Hatch would rather it not have. A bet-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 264

2 6 4 L a w r e n c e L e s s i g

ter system is less concentrated, less controlled, more diverse and decentral-
ized. As Hatch has written:

[I]f those digital pipes through which the new music will be delivered are
significantly narrowed by gatekeepers who limit access to or divert fans to
preferred content, a unique opportunity will be lost for both the creators of
music and their fans. That is why I think it is crucial that policymakers be
vigilant in keeping the pipes wide open.1

As I have argued throughout this book, the architecture that keeps the
“pipes wide open” is simply the original architecture of the Net. And a com-
mitment to keeping these pipes open is a commitment to preserving the
Net.
In both of these contexts, the senator sees something that ideologues
miss: that the greatest lesson of our history is the strength that comes from
our economic and cultural diversity. That concentration in either threatens
innovation in both—not because concentration alone is necessarily bad,
but because concentration gives the concentrated the power to steer evolu-
tion as it benefits them.
That power is not within our tradition. It is not what has built the
America we admire. And whether you’re from the Right or the Left, there is
a lesson in what this conservative preaches. We make choices, Hatch shows
us, that affect how easily the concentrated can direct the future. We should
make choices, Hatch insists, that make it less easy for the future to be di-
rected. Decentralized, diverse, nominated: this is the tradition that Hatch
defends; this is the architecture of the original Net.

as i write the last pages of this book, the threat to those values grows. A
court has just effectively shut down Napster, thereby assuring that the
recording industry gets to choose what kind of innovation in the distribution
of content will be allowed. Another court has ruled against Eric Eldred’s
challenge to copyright’s bloating, finding that “copyrights are categorically
immune from challenges under the First Amendment.”2 Though the Con-
stitution speaks of “limited times,” Congress is free to give Hollywood “per-
petual copyright on the installment plan.”3 And streaming across my
computer as I write these final paragraphs, judges from the D.C. Circuit
Court of Appeals are asking skeptical questions of lawyers for the govern-
ment defending the judgment against Microsoft. Commenting on the gov-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 265

T H E F U T U R E O F I D E A S 2 6 5

ernment’s defense of Java technologies, one judge has just said, “We are
going to replace one monopoly with another . . . right?”4
Though we’ve seen the new only when it has been freed from the old,
that lesson is lost on the Napster court. And though our Framers saw as
clearly as we can today that free content fuels innovation, that lesson was
forgotten by the court that decided Eric Eldred’s case. And though the
clearest lesson of the past twenty years is that innovation flourishes best
when it flourishes freely on a neutral platform, the judges deciding the Mi-
crosoft case cannot even imagine the value of a neutral platform. Is one
monopoly really just as good as another?
Alexander Hamilton promised that the judiciary would be “the least dan-
gerous branch.” The early history of the Net confirmed Hamilton’s predic-
tions. The Court in Reno v. ACLU spoke of the values in a free Net. It
resisted the popular efforts by Congress to regulate it quickly, even if Con-
gress was regulating in the name of important social values.
But the most significant governmental actions affecting the Net in the
twenty-first century so far are instances of judges intervening to protect
the old against the new. Rather than “wait and see,” the law has become the
willing tool of those who would protect what they have against the innova-
tion the Net could promise. The law is the instrument through which a
technological revolution is undone. And since we barely understand how
the technologists built this revolution, we don’t even see when the lawyers
take it away. As activist and technologist John Gilmore has put it, in a line
that captures the puzzle of this book: “[W]e have invented the technology
to eliminate scarcity, but we are deliberately throwing it away to benefit
those who profit from scarcity. . . . I think,” Gilmore continues, “we should
embrace the era of plenty, and work out how to mutually live in it.”5

late in the afternoon of one of California’s inevitably beautiful days,
Marc Andreessen was driving along one of California’s inevitably over-
crowded highways. More fitting the traffic than the weather, Andreessen’s
mood was dark. He was a twenty-nine-year-old computer science graduate
who had become one of the most successful entrepreneurs of his genera-
tion. Coauthor of an early browser for the World Wide Web (Mosaic),
founder of the first company to make the World Wide Web go (Netscape),
Andreessen was nonetheless down on the future.
“Innovation,” in Andreessen’s mind, is what the Web produced. As he
told me:
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 266

2 6 6 L a w r e n c e L e s s i g

When I came to Silicon Valley, everybody said . . . there’s no way in hell
that you could ever fund another desktop software company. That’s just
over. And then in 1995, 1996, 1997, and 1998, all those developers who
previously worked on desktop software said, Ah-hah, we’re upgrading to a
brand-new platform not controlled . . . by anybody—the Internet. [A]ll of
a sudden there was an explosion of innovation, a huge number of applica-
tions, and [a] huge number [of] companies.6

Innovation “resumed” just at the time when the platform for innovation
was neutral and, in the sense that I’ve described, free: when many different
actors were able to bring new ideas to the Net; when they knew that this
neutrality meant the old could not control how the new would behave;
when the new could behave however the market demanded.
But this innovation, Andreessen said, “is slowing once again. . . . Appli-
cation lock-in . . . [has] actually gotten stronger.” The opportunity to inno-
vate outside of the dominant players has again evaporated. We are back to
where we were before this revolution began. As control shifts back to the
large, the powerful, and the old, and as that control is ratified by the judges
in black robes, the opportunity that drew Andreessen from cold but traffic-
less Illinois disappears. The chance for something different is lost. The in-
novation age, Andreessen says, “is over.” And we are back to a world where
innovation proceeds as the dominant players choose.
Andreessen’s story is the fear of this book. An “explosion” of innovation
grew upon a neutral platform; that explosion is burning out quickly as the
platform is increasingly controlled.7 Whether through changes in the physi-
cal, or code, or content layers, the change Andreessen worries about is the
shift that I have described.
There is little to stop the transformation that worries Andreessen; there is
everything to push it along as fast as it can go. This book will be published
just as Microsoft’s .NET and Hailstorm initiatives hit the network. They
promise to integrate an extraordinary range of functionality into the core op-
erating system that Microsoft owns. Emboldened by an expected victory at
the court of appeals, Microsoft has expanded the bundling that the govern-
ment attacked to include a range of services never imagined by government
prosecutors. Authentication, instant messaging, e-mail, Web services—all
these will be bundled into the core operating system of the next generation
of Windows. Anyone who wants to compete in the provision of these
services will face as strong a barrier as Netscape faced against a bundled
Internet Explorer.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 267

T H E F U T U R E O F I D E A S 2 6 7

Microsoft is simply responding to another, very different nonneutral
platform—the emerging and dominant platform of America Online. After
its merger with Time Warner, AOL and its loyal members are another huge
and powerful force influencing the future of the Internet. AOL is not an op-
erating system, but for almost a majority of those who use the Internet, it is
in effect an operating system. Functionality is served in the AOL suite of
software; functionality beyond that is not.
These two companies—AOL Time Warner and Microsoft—will define
the next five years of the Internet’s life. Neither company has committed it-
self to a neutral and open platform.8 Hence, the next five years will be radi-
cally different from the past ten. Innovation in content and applications will
be as these platform owners permit. Additions that benefit either company
will be encouraged; additions that don’t, won’t. We will have re-created the
network of old AT&T, but now on the platform of the Internet. Content and
access will once again be controlled; the innovation commons will have
been carved up and sold.
This is the future of ideas. It could be different, but my sense is that it
won’t be. If we were more like Hatch, more skeptical of “gatekeepers,”
whether private or public; if we were less like Jay Walker, eager to view every
government-granted privilege as a God-given property right; if we were
more like Richard Stallman, committed to a principle of freedom in knowl-
edge and to a practice that assures that the power to control is minimized; if
there weren’t so few Paul Barans, willing to struggle for many years to force
a monopoly to face itself—if all this were so, there would be reason for hope.
But we are not. We are a democracy increasingly ruled by judges. We
elect a Congress that is increasingly chained by lobbyists. And we are a cul-
ture that deep down believes in this counterrevolution: that strangely thinks
that this increase in control makes sense.
As commentator Gordon Cook writes:

The Internet revolution has come and gone. It has created a tremendous
burst of innovation[—a] burst that now looks to have been misman-
aged. . . . [T]he people who did the least to advance the new technologies
seem most likely to control them. We are left not with the edge-controlled
intelligence of the [end-to-end] network but with the central authoritarian
control of the likes of AOL Time Warner.9

The irony astounds. We win the political struggle against state control so
as to reentrench control in the name of the market. We fight battles in the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 268

2 6 8 L a w r e n c e L e s s i g

name of free speech, only to have those tools turned over to the arsenal of
those who would control speech. We defend the ideal of property and then
forget its limits, and extend its reach to a space none of our Founders would
ever have imagined.
We move through this moment of an architecture of innovation to, once
again, embrace an architecture of control—without noticing, without resis-
tance, without so much as a question. Those threatened by this technology
of freedom have learned how to turn the technology off. The switch is now
being thrown. We are doing nothing about it.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 269

/ / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

N o t e s

Note: All URLs given as citations are subject to change.

PREFACE
1 Andrew L. Shapiro, The Control Revolution: How the Internet Is Putting Individuals
in Charge and Changing the World We Know (New York: PublicAffairs, 1999).

CHAPTER 1
1 Telephone interview with Davis Guggenheim, November 15, 2000. The law on the
books in this area (as distinct from what directors and lawyers working for directors think)
is “by no means certain.” See Melville B. Nimmer and David Nimmer, “Nimmer on
Copyright,” §13.05[D][3], at 13-222 (2001).
2 For these cases and others, see the extraordinarily helpful site http://www.

benedict.com/visual/visual.htm. See also Daniel B. Wood, “Hollywood Loves an Origi-
nal Idea,” Christian Science Monitor, December 15, 1997, http://www.csmonitor.com/
durable/1997/12/15/us/us.4.html; Woods v. Universal City Studios, Inc., 920 F. Supp. 62
(SDNY 1996) (12 Monkeys).
3 See Matthew C. Lucas, “The De Minimis Dilemma: A Bedeviling Problem of

Definitions and a New Proposal for a Notice Rule,” Journal of Technology Law & Policy
4, no. 3 (2000): 2 (http://journal.law.ufl.edu/~techlaw/4-3/lucas.html).
4 Jessica Litman, Digital Copyright (Amherst, N.Y.: Prometheus Books, 2001),

244–45.
5 Telephone interview with Davis Guggenheim.
6 Niccolò Machiavelli, The Prince, 2nd ed. (London and New York: W. W. Norton,

1992), 17.
7 In 1710, the English Parliament passed the Statute of Anne, which, to the horror of

its original supporters, was amended to limit the term of copyright to twenty-eight years.
In 1774, the House of Lords finally upheld the limit, permitting the works of Shake-
speare to fall into the public domain for the first time. Donaldson v. Becket, English Re-
ports 98 (House of Lords, 1774), 251, overturning Millar v. Taylor, Burroughs 4 (1769):
2303, 2308. See Mark Rose, Authors and Owners (Cambridge, Mass.: Harvard Univer-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 270

2 7 0 N O T E S

sity Press, 1993), 97. Had the work of Shakespeare not fallen into the public domain, it
would not have been protected in the United States, because foreign copyrights were not
protected in the United States until 1891. T. Bender and D. Sampliner, “Poets, Pirates,
and the Creation of American Literature,” New York University Journal of International
Law & Politics 29 (1997): 255. Americans were free to copy English works without the
permission of English authors and were free to translate foreign works without the per-
mission of foreign copyright holders.
8 For an introduction, see “The Future of Digital Entertainment” (Special Report),

Scientific American 283 (2000): 47.
9 Apple, of course, means something a bit narrower by the term mix. See http://www.

apple.com/imac/digitalmusic.html: “Because iTunes is really about freedom. The free-
dom, first and foremost, to play songs in the order you want, not the order they were first
recorded on CD. The freedom to mix and match artists and musical categories as it suits
you. The freedom to create your own music CDs. And the freedom to put more than a
hundred MP3 songs on a single CD.”
10 The relationship to low-cost production is no accident with some modern music. As

John Leland describes it, “The digital sampling device has changed not only the sound
of pop music, but also the mythology. It has done what punk rock threatened to do: made
everybody into a potential musician, bridged the gap between performer and audi-
ence.” Siva Vaidhyanathan, Copyrights and Copywrongs: The Rise of Intellectual Prop-
erty and How It Threatens Creativity (New York: New York University Press, 2001), 138.
By keeping the cost low, and hence the distance between creator and consumer short,
the genre aspires to keep the range of creators as broad as possible. That aspiration for
music could, I argue in this book, become an aspiration for creativity more generally.
11 This is the character of Caribbean music as well. “Every new version will slightly

modify the original tune,” but then, obviously, draw upon and copy it. Ibid., 136.
12 Richard Stallman has been likened to the Moses of what I will call the “open code”

movement. The likeness is indeed striking. As I describe in chapter 4, below, Stallman
began the movement to build a free operating system. But as with Moses, it was another
leader, Linus Torvalds, who finally carried the movement into the promised land by fa-
cilitating the development of the final part of the OS puzzle. Like Moses, too, Stallman
is both respected and reviled by allies within the movement. He is an unforgiving, and
hence for many inspiring, leader of a critically important aspect of modern culture. I
have deep respect for the principle and commitment of this extraordinary individual,
though I also have great respect for those who are courageous enough to question his
thinking and then sustain his wrath.
Stallman insists that those who would advance the values of the free software move-
ment must adopt the language “free” rather than “open.” This seems to me an unpro-
ductive debate. To the extent Stallman believes that people dilute the insights of the free
software movement by minimizing its connection to fundamental values, he is correct.
The importance of free and open source software is much more than business, or effi-
cient code. But the remedy to narrowness is not magic words—especially when the
magic words tend to confuse rather than clarify. I am partial to the term open—as in
open society; I believe it is properly a reference to values as well as the licenses under
which code is distributed; and by “open code” I mean to refer to the values across both
technical and legal contexts that promote a world where governing structures—code—
are fundamentally free.
For an exceptional study of free and open source software, and the incentives that
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 271

N O T E S 2 7 1

are behind it, see Working Group on Libre Software, “Free Software/Open Source:
Information Society Opportunities for Europe?,” Version 1.2 (April 2000), http://eu.
conecta.it/paper.pdf.
13 In the terms of legal theory, there are two distinct ways in which a resource could be

“free” in the sense I mean. Either no one would have any entitlement to the resource, or
if someone did have an entitlement, the resource would be protected by a liability rather
than a property rule. See Guido Calabresi and Douglas Melamed, “Property Rules, Lia-
bility Rules, and Inalienability: One View of the Cathedral,” Harvard Law Review 85
(1972): 1089. See also Robert P. Merges, “Institutions for Intellectual Property Transac-
tions: The Case of Patent Pools,” in Expanding the Boundaries of Intellectual Property,
Rochelle Cooper Dreyfuss and Diane Leenheer Zimmerman, eds. (Oxford: Oxford
University Press, 2001), 123, 131 (“The essence of this Framework is this: Calabresi and
Melamed assign all legal entitlements to one of two rules, ‘property rules’ and ‘liability
rules.’ The former are best described as ‘absolute permission rules’: one cannot take
these entitlements without prior permission of the owner. The rightholder, acting indi-
vidually, thus sets the price. Most real estate fits this description. By contrast, liability
rules are best described as ‘take now, pay later.’ They allow for nonowners to take the en-
titlement without permission of the owner, so long as they adequately compensate the
owner later. In the Calabresi-Melamed Framework, ex post adequate compensation is
deemed ‘collective valuation.’ ”).
14 Carol Rose, “The Comedy of the Commons: Custom, Commerce, and Inherently

Public Property,” University of Chicago Law Review 53 (1986): 711, 712 .
15 I don’t mean that all such barriers will be properly considered “neutral.” Sometimes

barriers are merely a tool for raising rivals’ costs. Some of those nonneutral barriers will
be remedied through antitrust law. But I’m assuming the general case is benign in real
space.

CHAPTER 2
1 For useful analyses of “the commons,” see generally Anarchy and the Environment:
The International Relations of Common Pool Resources, J. Samuel Barkin and George E.
Shambaugh, eds. (Albany: State University of New York Press, 1999); Managing the
Commons, 2nd ed., John A. Baden and Douglas S. Noonan, eds. (Bloomington, Ind.: In-
diana University Press, 1998); Susan J. Buck, The Global Commons: An Introduction
(Washington, D.C.: Island Press, 1998); Privatizing Nature: Political Struggles for the
Global Commons, Michael Goldman, ed. (London: Pluto Press, in association with
Transnational Institute, 1998); Local Commons and Global Interdependence: Hetero-
geneity and Cooperation in Two Domains, Robert O. Keohane and Elinor Ostrom, eds.
(London and Thousand Oaks, Calif.: Sage Publications, 1995); The Political Economy
of Customs and Culture: Informal Solutions to the Commons Problem, Terry L. Anderson
and Randy T. Simmons, eds. (Savage, Md.: Rowman and Littlefield Publishers, 1993);
Making the Commons Work: Theory, Practice and Policy, Daniel W. Bromley, ed. (San
Francisco: ICS Press, 1992); Commons Without Tragedy: Protecting the Environment
from Overpopulation—A New Approach, Robert V. Anderson, ed. (London: Shepheard-
Walwyn; Savage, Md.: Barnes & Noble, 1991); Glenn G. Stevenson, Common Property
Economics: A General Theory and Land Use Applications (Cambridge, England, and
New York: Cambridge University Press, 1991); Elinor Ostrom, Governing the Commons:
The Evolution of Institutions for Collective Action (Cambridge, England, and New York:
Cambridge University Press, 1990).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 272

2 7 2 N O T E S

The traditional commons is distinct from what Henry Smith calls a “semicommons.”
See Henry E. Smith, “Semicommon Property Rights and Scattering in the Open
Fields,” Journal of Legal Studies 29 (2000): 131 (defining a semicommons as a mix of
common and private rights where both are significant and interact).
While the standard view is that a commons induces overuse, some argue that it may
inspire underuse. Compare Richard A. Posner, Economic Analysis of Law, 4th ed.
(Boston: Little, Brown, 1992), 32, with Frank I. Michelman, “Ethics, Economics and
the Law of Property,” in Ethics, Economics, and the Law, J. Roland Pennock and
John W. Chapman, eds. (New York: New York University Press, Nomos XXIV [series],
1982), 25–27.
2 The OED was compiled from the volunteer efforts of thousands of people sending

examples of usage to editors. See Simon Winchester, The Professor and the Madman: A
Tale of Murder, Insanity, and the Making of the Oxford English Dictionary (New York:
HarperCollins Publishers, 1998).
3 The Oxford English Dictionary, 2nd ed., vol. 3, prepared by J. A. Simpson and E. S. C.

Weiner (Oxford: Clarendon Press; New York: Oxford University Press, 1989), 567.
4 The observant will notice that a resource offered in conditions of perfect competi-

tion comes close to the conditions that I described obtain with a commons, even though
the resource is “owned.” Perfect competition constrains the owner in similar ways,
though the right not to sell at all distinguishes the two sets of constraints.
5 This is the essence of an entitlement protected by a property rule. See Guido Cala-

bresi and Douglas Melamed, “Property Rules, Liability Rules, and Inalienability: One
View of the Cathedral,” Harvard Law Review 85 (1972): 1089, 1092. Not all liability rule
cases will eliminate this core of discretion. But the system is structured to avoid it.
6 Virgil, Virgil in English Rhythm, 2nd ed., Robert Corbet Singleton, trans. (London:

Bell and Daldy [imprint], 1871), iii–iv (“There is a common of language to which both
poetry and prose have the freest access.”).
7 Garrett Hardin, “The Tragedy of the Commons,” Science 162 (1968): 1243. The

idea of congestion externalities of course predates Hardin. See Posner, “Economic
Analysis of Law,” 32–34, citing Frank H. Knight, “Some Fallacies in the Interpretation of
Social Cost,” Quarterly Journal of Economics 38 (1924): 582.
8 Hardin, 1244 (emphasis added).
9 Ostrom, ch. 3; Robert C. Ellickson, Order Without Law: How Neighbors Settle Dis-

putes (Cambridge, Mass.: Harvard University Press, 1991). See also Making the Com-
mons Work: Theory, Practice and Policy, Daniel W. Bromley, ed. (San Francisco: ICS
Press, 1992), part 2 (describing case studies).
10 Elinor Ostrom has been the most effective in demonstrating the “fallacy” of the

“tragedy of the commons.” As she writes:
What makes these models so dangerous—when they are used metaphorically as the
foundation for policy—is that the constraints that are assumed to be fixed for the purpose
of analysis are taken on faith as being fixed in empirical settings, unless some external au-
thorities change them.
Ostrom, 6–7. For example, in an article in 1988, The Economist asserts about fisheries
that “left to their own devices, fishermen will overexploit stocks” and that “to avoid di-
saster, managers must have effective hegemony over them.” Ibid., 8. But that assumption
cannot be made. It depends upon the social system within which the commons exist;
often, as Ostrom has demonstrated, control is possible without either state intervention
or a system of private property.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 273

N O T E S 2 7 3

11 Obviously, many have noticed the “commonslike” character of the Internet. See,

e.g., Douglas S. Noonan, “Internet Decentralization, Feedback, and Self-Organization,”
188–89, in Managing the Commons, 2nd ed., John A. Baden and Douglas S. Noonan,
eds. (Bloomington, Ind.: Indiana University Press, 1998) (“The Internet is a commons
[or several commons]. . . .”). See also James Boyle, “A Politics of Intellectual Property:
Environmentalism for the Net?,” Duke Law Journal 47 (1997): 87. Not all have been as
convinced that its commons nature accounts for its innovation. See, e.g., Noonan,
“Internet Decentralization,” 198 (“In spite of its commons nature, or perhaps because of
it”).
12 See Yochai Benkler, “From Consumers to Users: Shifting the Deeper Structures of

Regulation,” Federal Communications Law Journal 52 (2000): 561, 562–63 (“These
choices occur at all levels of the information environment: the physical infrastruc-
ture layer—wires, cable, radio frequency spectrum—the logical infrastructure layer—
software—and the content layer. . . .”).
13 I’m simplifying vastly from both the OSI seven-layer model of network design, see

Douglas E. Comer, Internetworking with TCP/IP, 4th ed. (Upper Saddle River, N.J.:
Prentice-Hall, 2000), 181–95; Pete Loshin, TCP/IP Clearly Explained, 2nd ed. (Bos-
ton: AP Professional, 1997), 12–18; and Berners-Lee’s four-layer description (trans-
mission, computer, software, and content) in Tim Berners-Lee, Weaving the Web: The
Original Design and Ultimate Destiny of the World Wide Web by Its Inventor (San Fran-
cisco: HarperSanFrancisco, 1999), 129–30.
14 Certain speech has always been regulated on the telephone. For example, in 1883,

a local telephone company terminated a subscriber’s service for using the word damned
with an operator. The company’s contract prohibited “profane, indecent or rude lan-
guage.” An Ohio court upheld the company’s actions, and the Supreme Court Com-
mission later affirmed the court’s decision. For this and other examples, see Peter W.
Huber, Michael K. Kellogg, and John Thorne, Federal Telecommunications Law, 2nd
ed. (Gaithersburg, Md.: Aspen Law & Business, 1999), 1275–76, 1288–92.

CHAPTER 3
1 Milton Mueller, Universal Service: Competition, Interconnection, and Monopoly in
the Making of the American Telephone System (Cambridge, Mass.: MIT Press; Washing-
ton, D.C.: AEI Press, 1997), 180. For other useful histories, see A History of Computing
Research at Bell Laboratories (1937–1975), Bernard D. Holbrook and W. Stanley
Brown, eds. (New York: The Laboratories, 1982); A History of Engineering and Science
in the Bell System, M. D. Fagen, ed. (New York: The Laboratories, prepared by members
of the technical staff, Bell Telephone Laboratories, 1975–85); Robert W. Garnet, The
Telephone Enterprise: The Evolution of the Bell System’s Horizontal Structure,
1876–1909 (Baltimore: Johns Hopkins University Press, 1985); Alvin Von Auw, Heri-
tage and Destiny: Reflections on the Bell System in Transition (New York: Praeger, 1983);
John Brooks, Telephone: The First Hundred Years (New York: Harper & Row, 1976);
George David Smith, The Anatomy of a Business Strategy: Bell, Western Electric, and the
Origins of the American Telephone Industry (Baltimore: Johns Hopkins University Press,
1985); Edwin M. Asmann, The Telegraph and the Telephone: Their Development and
Role in the Economic History of the United States: The First Century, 1844–1944 (Lake
Forest: Lake Forest College, 1980); Amy Friedlander, Natural Monopoly and Universal
Service: Telephones and Telegraphs in the U.S. Communications Infrastructure,
1837–1940 (Reston, Va.: Corporation for National Research Initiatives, 1995); John
Patrick Phillips, Ma Bell’s Millions (New York: Vantage Press, 1970).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 274

2 7 4 N O T E S

2 Mueller, 7.
3 Ibid., 148.
4 Ibid., 188–89.
5 Ibid., 104–13.
6 Ibid., 180–81. See also David F. Weiman and Richard C. Levin, “Preying for Mo-

nopoly? The Case of Southern Bell Telephone Company, 1894–1912,” Journal of the
Political Economy 102 (1994): 103 (proposing a “modified version of the predation hy-
pothesis” that shores up Mueller’s work).
7 Mueller, 51, 78–79.
8 The alternative, modern meaning for “universal service” is that every household has

access to the network. Mueller argues convincingly this was not the original meaning;
ibid., 163. For an argument that the funding obligations that support universal service
should be extended to ISPs, see Robert M. Frieden, “Universal Service: When Tech-
nologies Converge and Regulatory Models Diverge,” Harvard Journal of Law & Tech-
nology 13 (2000): 395.
9 Peter W. Huber, Michael K. Kellogg, and John Thorne, Federal Telecommunications

Law, 2nd ed. (Gaithersburg, Md.: Aspen Law & Business, 1999), 17.
10 Ibid.
11 Ibid.
12 Ibid.
13 Mueller, 162.
14 Interview with Paul Baran in Stanford, California, November 14, 2000. For other

interviews of Baran, see Stewart Brand, “Founding Father,” Wired (March 2001), avail-
able at http://www.wired.com/wired/archive/9.03/baran_pr.html; interview by David
Hochfelder with Paul Baran, electrical engineer, Newark, New Jersey (October 24,
1999), available at http://ieee.org/organizations/history_center/oral_histories/transcripts/
baran.html; interview by J. O’Neill with Paul Baran, Menlo Park, California (March 5,
1990); George Gilder, “Inventing the Internet Again,” Forbes (June 2, 1997), 106
(lengthy article about Baran); Katie Hafner and Matthew Lyon, “Casting the Net,” The
Sciences (September 1, 1996), 32.
15 American Telephone & Telegraphy Co., Telephone Almanac, foreword (1941).
16 Interview with Paul Baran.
17 Ibid.
18 Peter Huber, Orwell’s Revenge: The 1984 Palimpsest (New York: Free Press; Toronto:

Maxwell Macmillan Canada; New York: Maxwell Macmillan International, 1994),
268–69; Huber, Kellogg, and Thorne, 416.
19 And the decision was reversed by the D.C. circuit. Hush-a-Phone Corp. v. United

States, 238 F. 2d 266 (D.C. Cir., 1956).
20 The idea is developed in Kleinrock’s dissertation: Leonard Kleinrock, Message

Delay in Communication Nets with Storage (1962, unpublished Ph.D. dissertation, Mas-
sachusetts Institute of Technology), which was later published in a modified form. See
Leonard Kleinrock, Communication Nets: Stochastic Message Flow and Delay (New
York: McGraw-Hill, 1964). See also John Naughton, A Brief History of the Future: The
Origins of the Internet (London: Weidenfeld & Nicolson, 1999), 92, 118–19 (discussing
other earlier contributors to the Internet).
21 Baran attributes to him the discovery of the term. Interview with Paul Baran (“The

term ‘packet switching’ was first used by Donald Davies of the National Physical Labo-
ratory in England, who independently came up with the same general concept in No-
vember 1965.”).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 275

N O T E S 2 7 5

22 Baran confirmed this history to me in an interview. “So the first level of objections

was about technology—that I didn’t understand how the telephone system worked,
[and] that what I’m proposing could not possibly work.” Interview with Paul Baran.
23 Naughton, 107. Authors Katie Hafner and Matthew Lyon recount a similar resis-

tance in Where Wizards Stay Up Late: The Origins of the Internet (New York: Simon &
Schuster, 1996), 62–64.
24 Interview with Paul Baran.
25 Ibid.
26 See Steve Bickerstaff, “Shackles on the Giant: How the Federal Government Cre-

ated Microsoft, Personal Computers, and the Internet,” Texas Law Review 78 (1999): 1,
60–61 (explaining that “restrictions on the Bell System helped the personal computer
market to develop through diverse competition and innovation”); Peter W. Huber, Law
and Disorder in Cyberspace: Abolish the FCC and Let Common Law Rule the Telecosm
(Oxford and New York: Oxford University Press, 1997).
27 Naughton, 108–9.
28 Interview with Paul Baran.
29 See J. H. Saltzer et al., “End-to-End Arguments in System Design,” available at

http://web.mit.edu/Saltzer/www/publications/endtoend/endtoend.pdf; David P. Reed et
al., “Active Networking in End-to-End Arguments,” available at http://Web.mit.edu/
Saltzer/www/publications/endtoend/ANe2ecomment.html. For treatment of e2e, see
Cameron R. Graham, “Cable TV Law 2001: Competition in Video, Internet and Tele-
phony,” Practicing Law Institute: PLI Order No. G0–00LY, 642 (2001): 486; Rob
Frieden, “Does a Hierarchical Internet Necessitate Multilateral Intervention?,” North
Carolina Journal of International Law and Commercial Regulation 26 (2001): 361;
Robert M. Kossick Jr., “The Internet in Latin America: New Opportunities, Develop-
ments and Challenges,” Cyberspace Lawyer 6, no. 1 (2001): 11 (noting the use of end-to-
end principles in Brazil and Mexico); and Lawrence Lessig, foreword, “Cyberspace and
Privacy: A New Legal Paradigm?,” Stanford Law Review 52 (2000): 987.
30 National Research Council, The Internet’s Coming of Age (Washington, D.C.: Na-

tional Academy Press, 2000), 30.
31 Telephone interview with David P. Reed (February 7, 2001), who contributed to the

early design of the Internet protocols—TCP/IP—while a graduate student at MIT.
32 As I describe in this paragraph, the importance of architecture to the character of

the Internet was a theme of the early activism of Mitch Kapor. But the author who first
focused the importance of architecture on the issues I describe in this book is Kevin Wer-
bach. As Werbach wrote in Release 1.0:
Architecture matters. For the most part, today’s Net is open, decentralized and competi-
tive. It fosters innovation because it is a standards-based general-purpose platform. . . .
[But t]he people building the next generation of high-speed access pipes are trying to
change this model.
Kevin Werbach, “The Architecture of Internet 2.0,” Release 1.0 (February 19, 1999).
As François Bar puts it, describing the electronic marketplace: “The most fundamental
transformation of commercial activities [from cyberspace] is not primarily about effi-
ciency, but has to do with the market and industry structure. It is about architecture.”
François Bar, “The Construction of Marketplace Architecture,” Brookings & Internet
Policy Institute (forthcoming 2001): 12.
33 Lawrence Lessig, Code and Other Laws of Cyberspace (New York: Basic Books,

1999), 243, note 19 (citing Kapor).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 276

2 7 6 N O T E S

34 To computer scientists, my use of the term architecture is a bit misleading. Com-

puter scientists typically use “architecture” to refer to “the visible characteristics of only
the element actually performing instructions, that is, the processor of a computer sys-
tem”; Harold Lorin, Introduction to Computer Architecture and Organization, 2nd ed.
(New York: Wiley, 1989), 10, and analogously for a network. IBM’s System/360 is said to
be the first computer system to have had an “architecture.” According to its architects,
“the term architecture [means] the attributes of a system as seen by the programmer, i.e.,
the conceptual structure and functional behavior, as distinct from the organization of the
data flow and controls, the logical design and the physical implementation.” Carliss Y.
Baldwin and Kim B. Clark, Design Rules, vol. 1 (Cambridge, Mass.: MIT Press, 2000), 215.
I mean the term to be far more general—to refer to “both the Internet’s technical pro-
tocols (e.g., TCP/IP) and its entrenched structures of governance and social patterns of
usage that themselves are not easily changeable, at least not without coordinated action
by many parties.” Lawrence Lessig and Paul Resnick, “Zoning Internet Speech,” Michi-
gan Law Review 98 (1999): 395.
35 Network Working Group, “Request for Comments: 1958, Architectural Principles

of the Internet,” Brian E. Carpenter, ed. (1996), available at http://www.ietf.org/rfc/
rfc1958.txt.
36 Ibid, §2.1.
37 Ibid.
38 Tim Berners-Lee, Weaving the Web: The Original Design and Ultimate Destiny of

the World Wide Web by Its Inventor (San Francisco: HarperSanFrancisco, 1999), 99.
39 As background, see Peter Cukor and Lee McKnight, “Knowledge Networks, the

Internet, and Development,” Fletcher Forum of World Affairs (Winter 2001): 43, 46;
George Gilder, Telecosm: How Infinite Bandwidth Will Revolutionize Our World (New
York: Free Press, 2000), 70–71.
40 Telephone interview with David Isenberg, February 14, 2001.
41 Or at least this is an ideal. See Roger Feldman, “e2e vs. General Edison OnLine,”

PMA OnLine Magazine, http://www.retailenergy.com/feldman/0007flmn.htm.
42 Telephone interview with David Reed.
43 Berners-Lee, 208.
44 National Research Council, 138.
45 Ibid., 107.
46 Ibid., 36–37.
47 Ibid., 37.
48 Douglas E. Comer, Internetworking with TCP/IP, 4th ed., vol. 1 (Upper Saddle

River, N.J.: Prentice-Hall, 2000), 691 (HTTP stands for “hypertext transfer protocol” and
is “[t]he protocol used to transfer Web documents from a server to a browser”), 713 (TCP
stands for “transmission control protocol”), and 694 (IP stands for “Internet protocol).”
Together, TCP and IP allow data delivery between machines on the Internet. “The en-
tire protocol suite is often referred to as TCP/IP because TCP and IP are the two funda-
mental protocols.”).
49 Berners-Lee, 35.
50 See, e.g., Paul E. Ceruzzi, A History of Modern Computing (Cambridge, Mass.:

MIT Press, 1998), 301–2 (describing hypertext “inventor” Ted Nelson’s debt to Vannevar
Bush, quoting Bush: “The human mind . . . operates by association. With one item in its
grasp, it snaps instantly to the next that is suggested by the association of thoughts, in ac-
cordance with some intricate web of trails carried by the cells of the brain.”).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 277

N O T E S 2 7 7

51 See Robert M. Fano, “On the Social Role of Computer Communications,” Pro-

ceedings of the IEEE 60 (September 1972): 1249.
52 Berners-Lee, 46. See also James Gillies, How the Web Was Born: The Story of the

World Wide Web (Oxford and New York: Oxford University Press, 2000); Hafner and
Lyon, Internet Dreams: Archetypes, Myths, and Metaphors, Mark J. Stefik and Vinton G.
Cerf, eds. (Cambridge, Mass.: MIT Press, 1997).
53 Berners-Lee, 40 (describing Gopher and WAIS growing faster).
54 Ibid. (interconnect).
55 Ibid., 72–73.
56 Ibid.
57 Ibid., 74.
58 Ibid., 99.
59 See Barbara Esbin, “Internet over Cable: Defining the Future in Terms of the Past,”

Communications Law Conspectus (Winter 1999): 37, 46 (“ARPA began to support the
development of communications protocols for transferring data and electronic mail [e-
mail] between different types of computer networks.”); Needham J. Boddie II et al., “A
Review of Copyright and the Internet,” Campbell Law Review 20 (1998): 193, 196
(“Funding for the Internet comes from five federal agencies, various universities and
states, and private companies such as IBM and MCI.”).
60 Naughton, 83–85.
61 Ibid., 84.
62 See, e.g., http://www.asiapoint.net/insight/asia/countries/myanmar/my_spedev.htm.
63 Among other restrictions, AT&T was not permitted to get into the computer busi-

ness. This fact becomes quite important in explaining the birth of Unix. For a compre-
hensive and balanced account of the effect of these limitations, see Bickerstaff, 14–17.
64 Bickerstaff, 25.
65 For more on the history of the FCC’s Computer I and Computer II decrees, see

ibid., 1–37; Huber, Kellogg, and Thorne, §5.4.
66 National Research Council, 130–31, n. 18 (describing best efforts as consequences

of uniformity).
67 “Humans can tolerate about 250 msec of latency before it has a noticeable effect,”

http://www.dialogic.com/solution/Internet/4070Web.htm.
68 They are described in Mark Gaynor et al., “Theory of Service Architecture: How to

Encourage Innovation of Services in Networks” (working paper, 2000, on file with au-
thor), 14.
69 Indeed, George Gilder believes the increasing capacity of optical fiber will render

moot this debate about QoS. “Today on the Internet, the consensus claims that QoS will
be indispensable for voice and video. But with true bandwidth abundance, QoS com-
plexities are irrelevant—an ATM tax imposed on the vast bandwidth of fiber with its 10
to the minus 15 error rates, far better than the reliability of telephone circuits.” Gilder,
Telecosm, 80.
This point is made expressly with respect to quality of service and end-to-end in an im-
portant paper by Barbara van Schewick, “The End-to-End Principle in Network Design,
Its Impact on Innovation and Competition in the Internet, and Its Future in the Net-
work Architecture of the Next Generation Internet” (working paper, September 2000,
on file with author). Van Schewick distinguishes between two forms of QoS—
integrated services architecture and differentiated services architecture (6, 7)—and ar-
gues that the former puts more pressure on the end-to-end principle than the latter. She
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 278

2 7 8 N O T E S

argues that a more general solution to the QoS problem is overprovisioning (7), and mo-
tivates that argument with strong evidence of the incentives of network providers to dif-
ferentiate between different applications. This, she argues, reduces “the incentives for a
potential innovator to develop new applications.” Ibid., 7.
Van Schewick is the first researcher I know of to link QoS, overprovisioning, and end-
to-end, and I have drawn heavily upon her understanding in this context and throughout
this book.
Others are skeptical of the overprovisioning solution. For overprovisioning to work,
they argue, overprovisioning must exist at every point of the network. But that requires
coordinated action that is impossible in the context of the Internet. Thus, the solution is
an ideal without a mechanism to achieve it.
This skepticism is appropriate, but it does suggest another response. If the market
“naturally” would not choose overprovisioning, but overprovisioning would nonetheless
be a productive response, then we may well have identified a role for government. I de-
scribe this a bit more in chapter 12.
70 For examples of QoS technologies, and a discussion of QoS generally, see Joe

Lardieri, “Quality of Service: Which Flavor Is Right for You?,” Telecommunications
(August 1999): 53; Dave Kosiur, “Directing Traffic with a Touch of Class,” PC Week
(March 23, 1998): 91; David B. Miller, “Quality of Service: Directing Data Through the
Network,” ENT (October 6, 1999): 22.
71 See Gilder, Telecosm, 158–64.
72 The critical change necessary for this “fibered” network to achieve this level of

capacity is the emergence of optical routing technologies. The emerging bottleneck in
the network is the relatively slow speed of electronic switches. For an introduction, see
“The Ultimate Optical Network” (Special Report), Scientific American 284 (January
2001): 80.
73 See, e.g., Bill Frezza, “Telecosmic Punditry: The World Through Gilder-Colored

Glasses,” Internet Week (December 4, 2000): 47; Rob Walker, “The Gildercosm,” Slate
magazine (September 11, 2000), available at http://slate.msn.com/code/MoneyBox/
MoneyBox.asp?Show=9/11/2000&idMessage=6030 (Gilder runs “against the current
wisdom that sees bandwidth shortage as a problem”); Julian Dibbell, “From Here to In-
finity,” Village Voice, September 5, 2000, 65 (“It takes either profound sloth or transcen-
dent faith to persist in voicing such breathless sentiments.”). For a more favorable review,
see, e.g., Blair Levin, “Review, TELECOSM: How Infinite Bandwidth Will Revolution-
ize Our World,” Washington Monthly (September 1, 2000): 54.
74 E-mail from Timothy Wu to Lawrence Lessig, February 16, 2001.
75 This claim depends upon the assumption that the value of the network activity was

higher than the telephone activity it displaced. That assumption about social value
therefore has little relation to the actual costs.

CHAPTER 4
1 Technically, some of these functions could be provided if the program simply pro-
vided perfectly transparent representations of the “application program interfaces,” or
APIs. In some contexts this distinction will be important, but I ignore it here. A suffi-
ciently skilled programmer could also learn something about compiled code through re-
verse engineering. The success of this would depend upon how well structured the
original code was.
2 Ceruzzi, A History of Modern Computing, 108.
3 For a brief history of Unix, see William Shattuck, “The Meaning of UNIX,” in The
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 279

N O T E S 2 7 9

Unix System Encyclopedia, 2d ed.,Yates Ventures, eds. (Palo Alto, Calif.: Yates Ventures,
1985), 89, 93–94; Peter H. Salus, A Quarter Century of UNIX (Reading, Mass.: Addison-
Wesley Publishing Company, 1994), 5–61; Ronda Hauben, “The History of UNIX,”
available at http://www.dei.isep.ipp.pt/docs/unix.html (last visited June 12, 2001).
4 Robert Young and Wendy Goldman Rohm, Under the Radar: How Red Hat

Changed the Software Business—and Took Microsoft by Surprise (Scottsdale, Ariz.: Cori-
olis Group Books, 1999), 21; Donald K. Rosenberg, Open Source: The Unauthorized
White Papers (Foster City, Calif.: Hungry Minds, 2000), 9.
5 Ceruzzi, 94–95.
6 Ibid., 92.
7 Ibid., 283 (The result was that “UNIX was a godsend for university computer science

departments”).
8 Young and Rohm, 20.
9 See Richard Stallman, “The GNU Operating System and the Free Software Move-

ment,” in Open Sources: Voices from the Open Source Revolution, Chris DiBona, Sam
Ockman, and Mark Stone, eds. (Beijing and Sebastopol: O’Reilly, 1999), 53–54.
10 This commons would be built, in Stallman’s ideal world, with nonproprietary code.

Stallman is not opposed, however, to commercial software. As he explains, “Commer-
cial software and proprietary software are totally different concepts. ‘Commercial’ refers
to the financial arrangement of the software. ‘Proprietary’ refers to what the users are per-
mitted to do. Free software must have the freedom to copy, to modify, and to have the
source code. So proprietary software is mutually exclusive with free software, but there
can be commercial software that [is] free software.” Telephone interview by Hiroo Yama-
gata with Richard M. Stallman, August 8, 1997.
11 Peter Wayner, Free for All: How Linux and the Free Software Movement Undercut

the High-Tech Titans (New York: HarperBusiness, 2000), 36.
12 For a discussion of Stallman and the history of GNU/Linux, see ibid., 9, 34–36,

67–68; Stallman, 53–66; Mark Leon, “Richard Stallman, GNU/Linux,” InfoWorld (Oc-
tober 9, 2000): 62.
13 See, e.g., Linus Torvalds and David Diamond, Just for Fun: The Story of an Acci-

dental Revolutionary (New York: HarperBusiness, 2001); Pekka Himanen, Manuel
Castells (epilogue), and Linus Torvalds (prologue), The Hacker Ethic and the Spirit of
the Information Age (New York: Random House, 2001); Paula Rooney, “No. 11: The
Dark Horse,” Computer Reseller News, November 15, 1999.
14 Stallman: “Around 1992, combining Linux with the not-quite-complete GNU sys-

tem resulted in a complete free operating system. (Combining them was a substantial
job in itself, of course.) It is due to Linux that we can actually run a version of the GNU
system today.” Stallman, 65.
What to call this resulting OS—“Linux” or “GNU/Linux”—is a hotly contested issue.
Indeed, I’ve received more heat about this issue than about any other I’ve discussed in
this book. Supporters of the free software movement insist the product is GNU/Linux
and object that calling it “Linux” underplays the importance of Stallman and the
Free Software Foundation. Others emphasize the other great work that has spread this
free OS and insist that “Linux” is enough of a moniker. As Bob Young, chairman of Red
Hat, wrote me, “While Linus gets more credit than he deserves for the whole 800 MB
OSs that are known by their 16 MB Linux kernel, Richard gets less credit than he
deserves. But where Linus gets far more credit than he demands, Richard demands
more credit than he deserves (not that he does not deserve a great deal).” E-mail, June 8,
2001.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 280

2 8 0 N O T E S

It is an unfortunate feature of the current debate around open source and free software
projects that so much energy is devoted to what things are called—with great fury di-
rected at those who fail to speak properly (using “open” where “free” is to be used; using
“closed” where “proprietary” should be used). I’m guilty of some of those language sins
in this book, and only sometimes self-consciously. I will refer generally to the Linux OS
as “Linux,” though sometimes as “GNU/Linux” when the reminder helps. The impor-
tant point for our purposes is not who gets credit for what, but that the kernel of the
Linux OS is licensed under the Free Software Foundation’s General Public License
(GPL). That is, I believe, a critically important feature of this debate, and one that must
be kept clear.
15 For a recent and compelling history of the birth of Linux, see Glyn Moody, Rebel

Code: Linux and the Open Source Revolution (Cambridge, Mass.: Perseus Press, 2001).
16 Linus Torvalds, “The Linux Edge,” in Open Sources: Voices from the Open Source

Revolution, Chris DiBona, Sam Ockman and Mark Stone, eds. (Beijing and Sebastopol:
O’Reilly, 1999), 101.
17 Torvalds, “The Linux Edge,” 101.
18 Moody, Rebel Code, 130. See also Dorte Toft, “Open Source Group Forms Non-

profit,” Industry Standard (July 1, 1999), at http://www.thestandard.com/article/
0,1902,5377,00.html (visited on May 27, 2001).
19 See http://www.isc.org/products/BIND/bind-history.html.
20 Tim O’Reilly, “Hardware, Software, and Infoware,” in Open Sources: Voices from

the Open Source Revolution, Chris DiBona, Sam Ockman, and Mark Stone, eds. (Bei-
jing and Sebastopol: O’Reilly, 1999) 189, 191.
21 Wayner, 172.
22 See Carolyn Duffy Marsan, “Sendmail Adds Admin Console, IPv6 Support; New

Unix-Based Tool Can Manage Sendmail-Based E-mail Systems Across a Widespread
Organization,” Network World (February 7, 2000).
23 Alan Cox, “This Is How Free Software Works,” wideopennews 5, 8.6, http://

www2.usermagnet.com/cox/index.html.
24 There is a resistance to calling these “exclusive rights” “monopolies.” Frank Easter-

brook, “Intellectual Property Is Still Property,” Harvard Journal of Law & Public Policy
13 (1991): 108–09. But technically, of course, they are. That, however, does not entail
that a given exclusive right has any monopoly power. There is a purpose, however, in
calling these rights “monopolies”—to shift the burden of defending the rights to those
who claim they do some good. This was the strategy of McCauley in the mid–nineteenth
century as well. See Siva Vaidhyanathan, Copyrights and Copywrongs: The Rise of Intel-
lectual Property and How It Threatens Creativity (New York: New York University Press,
2001), 142.
25 Letter from Thomas Jefferson to Isaac McPherson (August 13, 1813), in The Writ-

ings of Thomas Jefferson, vol. 6 (H. A. Washington, ed., 1861), 175, 180.
26 Though GPL is often referred to as a “copyleft” license, copyleft depends funda-

mentally upon copyright. Without a copyright, there would be no way to ensure that
users of a bit of software code agreed to further obligations imposed upon the code by the
GPL. Software dedicated to the public domain, for example, would not have the ca-
pacity to ensure that users were bound by GPL. But with copyrighted code offered sub-
ject to the terms of GPL, the user either complies with the license or is an infringing user
of the code.
As I describe more below, GPL is not the only, or perhaps even the most important,
open code license. Other open code licenses differ from GPL by not requiring that sub-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 281

N O T E S 2 8 1

sequent derivative works open up their own code base. For an evaluation of GPL’s via-
bility under German law, see Axel Metzger and Till Jaeger, “Open Source Software and
German Copyright Law,” International Review of Industrial Property & Copyright Law
32 (2001):52.
27 See http://www.fsf.org/copyleft/gpl.html.
28 Stallman, 56.
29 See Moody, 168 (describing open source definition).
30 For a fuller discussion of various open code licenses and their implications, see

David McGowan, “Legal Implications of Open Source Software,” University of Illinois
Law Review (2001): 101, 113–19; Patrick K. Bobko, “Can Copyright Keep ‘Open
Source’ Software Free?,” AIPLA Quarterly Journal 28 (2000): 81; Ira V. Heffan, “Note:
Copyleft: Licensing Collaborative Works in the Digital Age,” Stanford Law Review 49
(1997): 1487; Daniel B. Ravicher, “Facilitating Collaborative Software Develop-
ment: The Enforceability of Mass-Market Public Software Licenses,” Virginia Journal of
Law & Technology 5 (2000): 11.
31 Wayner, 260.
32 The latest operating system from Apple, OS X, includes a portion called Darwin

that is licensed under an open source license. See http://www.apple.com/macosx/tour/
darwin.html.
33 For detailed analyses of this case, see Ken Auletta, World War 3.0: Microsoft and Its

Enemies (New York: Random House, 2001); David Bank, Breaking Windows: How Bill
Gates Fumbled the Future of Microsoft (New York: Free Press, 2001); John Heilemann,
Pride Before the Fall: The Trials of Bill Gates and the End of the Microsoft Era (New York:
HarperCollins, 2001).
34 The government isn’t the only one to make this allegation. As Red Hat chairman

Bob Young describes it, “Microsoft appeared to have gone so far as to dictate how fast
other companies could bring out new products—or if they could bring them out at all,
for that matter.” Young and Rohm, 7.
35 Some attempts were made at graphical user interface desktop environments before

Windows hit pay dirt, including VisiOn (by VisiCalc), Top View (by IBM), GEM (by
Digital Research), and Interface Manager (Microsoft’s precursor to Windows); none
fared well with consumers, however. Ceruzzi, 276.
36 In a well-known story, the company didn’t actually write DOS but simply licensed

it from Tim Paterson of Seattle Computer Products. Microsoft paid “about $15,000 for
the rights to use Seattle Computer Products’s work.” Microsoft later paid more for the
complete rights. Ibid., 270.
37 Stan Miastkowski, “A Cure for What Ails DOS,” BYTE (August 1990): 107.
38 See Caldera v. Microsoft, 72 F. Supp. 2d 1295, 1299 (D. Utah, 1999).
39 Jim Carlton, Apple: The Intrigue, Egomania, and Business Blunders That Toppled

an American Icon (New York: Times Books/Random House, 1997), 39–47.
40 Ibid., 28.
41 “Strategic behavior arises when two or more individuals interact and each indi-

vidual’s decision turns on what that individual expects the others to do.” Douglas G.
Baird, Robert H. Gertner, and Randal C. Picker, Game Theory and the Law (Cam-
bridge, Mass.: Harvard University Press, 1994), 1. As I use the term, I mean to restrict it
to cases where the resulting behavior raises questions under antitrust law.
As David Bank writes, the notion of a competitive strategy to protect Microsoft’s base
was at the core of Microsoft’s product design. This was referred to within Microsoft as the
“strategy tax.” As Bank writes:
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 282

2 8 2 N O T E S

Gates invariably pressed the same point: Be more “strategic.” He demanded hooks to
lock in customers. The goal was to drive interdependencies between various Microsoft
products and thus make it more painful for customers to switch to a competitor for any
individual feature.

Bank, 134. Gates placed the need for “strategic advantage” above concern for “cus-
tomers.” As Gates wrote in an e-mail titled “HTML ‘Openness’ ”:
People were suggesting that Office had to work equally well with all browsers and that we
shouldn’t force Office users to use our browser. . . . This is wrong and I wanted to correct
this.

Ibid., 73–74. Instead, as he went on to instruct, the product should be designed to be
“Office/Windows/Microsoft specific.” Ibid.
This “strategy tax” was “inherently demoralizing,” Bank writes. Ibid., 134. Software
teams wanted to “win on the merits,” ibid., not based on tricks built into the code. More
generally, Bank argues, many within the firm believed the strategy tax would eventually
harm Microsoft competitively. As Ben Slivka said to Gates, “I think your focus on ‘com-
mon code’ as a goal in itself has served to stifle innovation.” Ibid., 99. Slivka pushed “un-
fettered innovation,” free from “the constraints of the company’s historical franchises.”
Ibid. Another senior executive, Brad Silverberg, wrote, “I simply do not want to spend
my life in meetings struggling with the internal issues, getting pissy mail from Billg, [or]
hearing from people who want me to do unnatural and losing things to ‘protect’ Win-
dows.” Ibid., 175.
The idea that a company might succeed in this market without proprietary control—
without, in other words, a “strategy tax”—might seem counterintuitive. Companies such
as Cisco, however, demonstrate that success can come from something other than play-
ing games with platforms. Cisco has flourished despite the fact that the platform it builds
upon—the protocols of TCP/IP—are neutral and unowned. No doubt it faces fierce
competition, but it has succeeded nonetheless.
As I describe in chapter 14, Bank believes the strategy within Microsoft has now
changed to embrace neutral, rather than “taxing,” competition.
42 Elizabeth Corcoran, “Microsoft Deal Came Down to a Phone Call,” Washington

Post, July 18, 1994, A1.
43 Ibid.
44 As David Bank writes:

At 3 am on a rainy Monday in April 1995, Gates declared in an email to his top lieu-
tenants that the “Internet is destroying our position as the setter of standards. . . . [It] is
taking away our power every day and will have eroded it irretrievably by the time broad-
band is pervasive on the course we are on right now.[”]

Bank, 26. The core of the threat, Gates believed, was Netscape. “As Netscape’s stan-
dards became more important than Windows’,” Gates wrote, Netscape’s browser would
“commoditize the underlying operating system.” Ibid., 27.
Gates was equally concerned about Microsoft’s embrace of Java. As he wrote after
reading an analysis that pushed neutral development of Java, “This scares the hell out of
me. . . . It’s still very unclear to me what our [OSes] will offer to Java that will make them
unique enough to preserve our market position.” Ibid.
45 The government relied upon extensive evidence from Mr. Gates’s own e-mail and

statements. See United States v. Microsoft, 84 F. Supp. 2d 9, 59–60 (D.D.C. 1999).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 283

N O T E S 2 8 3

46 While it is a mistake to read too much into the loose use of the term leverage, it is

clear that Microsoft executives believed “leveraging” IE through the use of Windows was
the most effective way to gain market share. Bank, 70 (quoting Allchin). “We should
think first about an integrated solution—that is our strength.” Ibid., 69.
For a careful analysis of current economic thinking about tying and exclusive con-
tracts (at the core of the Microsoft case), see Michael D. Whinston, “Exclusivity and
Tying in United States v. Microsoft: What We Know, and Don’t Know,” Journal of Eco-
nomic Perspectives 15 (2001): 63.
47 Ibid., 11, 13.
48 Ibid., 212.
49 Ibid., 228.

And obviously, too, it can control more than the pace of innovation. My concern in
Code and Other Laws of Cyberspace was precisely the regulation that gets effected
through code—“Code is Law”—and the concern that the more code controls our life,
the more we should care about who the coders are. Denise Caruso has written power-
fully on the same idea. As she puts it, “[I]t is still true today that software—written by a
team of sleep-deprived programmers in some fusty cubicle—is the code that lays down
the absolute law by which we live our digital lives. We are not free to change that code;
our choice is to love it or leave it. In the Microsoft trial . . . the battle is really over whose
law is to be sovereign, software’s or the government’s.” “The Legacy of Microsoft’s Trial,”
New York Times, December 6, 1999, available at http://www.nytimes.com/library/tech/
99/12/biztech/articles/06digi.html. She adds, “[C]yberspace is already regulated by soft-
ware code, which by its nature decides whether and under what circumstances we will
have privacy or anonymity and who can or cannot have access and how. Though this fact
was not so terrifying when the software running the global Internet was built on open
standards, the situation has become significantly more ominous as the code has become
commercialized and thus private and not open to public scrutiny. . . . [W]hen commer-
cial code begins to determine the Internet’s architecture, it creates a kind of privatized
law that must be regulated if the public interest and public values are to be democrati-
cally represented.” Ibid.
50 One of the most important Unix clones was a project called BSD—the Berkeley

Software Distribution Project. The BSD project was born when Unix was still free. After
AT&T reclaimed control of Unix, it demanded that BSD cease its distribution. BSD re-
fused, and lawsuits ensued. After protracted litigation, the right of BSD (with certain
code rewritten) to distribute its version of Unix free of AT&T was sustained through a
settlement. For more on the BSD lawsuit and BSD generally, see Wayner, 38, 49–54
(discussing AT&T’s lawsuit against BSD), and 92–95 (discussing the development of
BSD); Cheryl Garber, “USL v. Berkeley,” UNIX Review 10 (1992): 32.
BSD gave its code away under a very weak license. Anyone could do anything with the
code, as long as he kept the copyright notices clear. But as groups developed to support
this free code, divisions among the troops developed as well: from this common base of
BSD, the project forked to include FreeBSD, NetBSD, and OpenBSD, among others.
Wayner, 98–99.
51 There is an interesting but unresolved debate about whether GPL (the strictest of

the open code licenses) does more to stem the risk of forking than more permissive open
code licenses. Compare Young and Rohm, 180 (arguing that GPL minimizes forking),
with Wayner, 221 (arguing the contrary).
52 Ibid., 210.
53 Eric S. Raymond, The Cathedral and the Bazaar: Musings on Linux and Open
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 284

2 8 4 N O T E S

Source by an Accidental Revolutionary (Beijing and Sebastopol, Calif.: O’Reilly, 1999),
151.
54 Wayner, 289.
55 “IBM Extends Software Leadership on Linux,” IBM Press Release, December 8,

2000.
56 Economists have only begun to examine the incentives that might affect open code

projects. Josh Lerner and Jean Tirole have summarized some of the benefits, relative to
closed code projects, as follows: (1) lower costs due to (a) familiarity of the code from,
e.g., university training, and (b) customization/bug-fixing advantages; and (2) higher
benefits, especially due to signaling from (a) better performance measurement (easier to
demonstrate skill), (b) “full initiative” (because no supervisory involvement), and (c)
greater labor market fluidity. See Josh Lerner and Jean Tirole, “The Simple Economics
of Open Source” (NBER Working Paper No. 7600, December 2000).
James Bessen has offered a far more ambitious model of the benefits from open code
projects. In a paper not yet published, Bessen argues that open code provision is superior
to closed source provision when the public good (i.e., software) is a complex public
good. The intuition is that the costs of debugging complex software projects become pro-
hibitively high within a closed code environment. As Bessen writes, “[I]f a product had
100 independent features and each combination took only one second to test, then
the job could not be finished before the sun is predicted to swallow the earth even if
every human currently alive spent every second until then testing.” James Bessen,
“Open Source Software: Free Provision of Complex Public Goods” (working version,
April 2001), http://researchoninnovation.org. Bessen concludes that “strong property
rights . . . may actually limit provision of complex [public goods],” and hence that open
code projects have a competitive advantage over closed.
57 This argument is related to a point made by Carl Shapiro and Hal R. Varian in In-

formation Rules: A Strategic Guide to the Network Economy (Boston, Mass.: Harvard
Business School Press, 1999), 256–59. As they argue there, in a market heavily depen-
dent upon standards, there is often an incentive for competitors to contribute to public
standards rather than to make the standard proprietary. The battle over the standard
HTML is an example of this. For a while, both Netscape and Microsoft tried to develop
extensions to HTML that were peculiar to their own servers and Web development soft-
ware. Their aim was to split the common standard base to induce the market to follow
their own design. Later, however, they both apparently decided that this effort to divide
the standard would not be profitable; better if there were more on a common platform
than fewer on a proprietary platform. Varian and Shapiro demonstrate the conditions
under which that behavior could be rational. See also Varian, “Buying, Sharing, and
Renting Information Goods,” Journal of Industrial Economics 48 (2000): 473. On Mi-
crosoft’s attitude toward a neutral HTML, see Bank, 74 (“If Microsoft doesn’t get to do
anything ‘proprietary’ with HTML in the browser,” Gates wrote, then “we have to stop
viewing HTML as central to our strategy and get onto another strategy.”). See also ibid.,
73 (regarding Office). See also Moody, 200. Moody believes that because the Apache
server was a dominant server for HTTP, and uncontrolled by any commercial venture, it
did not fracture in the same way. See ibid., 149.
58 For articles identifying scenarios in which this proposition fails, see Douglas Licht-

man et al., “Shared Information Goods,” Journal of Law & Economics 42 (1999): 117;
Hal R. Varian, “Buying, Sharing, and Renting Information Goods,” Journal of Industrial
Economics 48 (2000): 473; Douglas Lichtman, “Property Rights in Emerging Platform
Technologies,” Journal of Legal Studies 29 (2000): 615.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 285

N O T E S 2 8 5

59 James Boyle, Shamans, Software, and Spleens: Law and the Construction of the In-

formation Society (Cambridge, Mass.: Harvard University Press, 1996), ch. 4.
60 This is especially the case with embedded systems, which have exploded recently

because of the ability to incorporate essential operating system code freely. See “Em-
bedded Linux Basics,” http://lw.itworld.com/linuxworld/lw-2000–05/lw-05-embedded_p.
html; “Companies Bet on Embedded Linux,” http://www.forbes.com/2000/04/07/mu4.
html.

CHAPTER 5
1 See Susan J. Douglas, Inventing American Broadcasting 1899–1922, reprint ed. (Bal-
timore: Johns Hopkins University Press, 1997), 227–28; Eszter Hargittai, “Radio’s
Lessons for the Internet,” Communications ACM 43 (2000): 51, 54.
2 See National Broadcasting Co. v. U.S., 319 U.S. 190, 210 (1943). Radio “broadcast-

ing,” however, did not begin until KDKA began its service in Pittsburgh in 1920. See
Thomas W. Hazlett, “The Wireless Craze, the Unlimited Bandwidth Myth, the Spec-
trum Auction Faux Pas, and the Punchline to Ronald Coase’s ‘Big Joke’: An Essay on Air-
wave Allocation Policy” (Working Paper 01–01, AEI-Brookings Joint Center for
Regulatory Studies, January 2001), 95 (Harvard Journal of Law and Technology, spring
2001).
3 See Red Lion Broadcasting v. FCC, 395 U.S. 367, fn4 (1969). The statute was the

Radio Act of 1927, ch. 169, 44 Stat. 1162 (1927).
4 For an excellent account of the emergence of broadcasting as we would recognize it,

see Douglas, 292–322.
5 Robert W. McChesney, Rich Media, Poor Democracy: Communication Politics in

Dubious Times (Urbana: University of Illinois Press, 1999), 192.
6 Ibid., 194.
7 Hazlett, “The Wireless Craze,” 48.
8 Edward S. Herman and Robert W. McChesney, The Global Media: The New Mis-

sionaries of Corporate Capitalism (London and Washington, D.C.: Cassell, 1997), 138.
9 Erik Barnouw, A History of Broadcasting in the United States, vol. 1 (New York: Ox-

ford University Press, 1966–70), 96, 177–78, 243, 262–66.
10 See Peter W. Huber, Michael K. Kellogg, and John Thorne, Federal Telecommuni-

cations Law, 2nd ed. (Gaithersburg, Md.: Aspen Law & Business, 1999), 220–21,
865–66.
11 NBC v. U.S., 213.
12 Ibid., 228 (Murphy, J., dissenting).
13 R. H. Coase, “The Federal Communications Commission,” Journal of Law and

Economics 2 (1959): 1.
14 The idea originated in an article by a University of Chicago law student. See Leo

Herzel, “ ‘Public Interest’ and the Market in Color Television Regulation,” University of
Chicago Law Review 18 (1951): 802, 811–12 (“The FCC could lease channels for a
stated period to the highest bidder without making any other judgment of the economic
or engineering adequacy of the standards to be used by the applicant.”). See also Rich-
ard A. Posner, Economic Analysis of Law, 4th ed. (Boston: Little, Brown, 1992), 673 (say-
ing the Supreme Court’s rationale for different spectrum regulation rules “is economic
nonsense”).
Ronald Coase, however, made the idea famous. See Yochai Benkler, “Overcoming
Agoraphobia: Building the Commons of the Digitally Networked Environment,” Har-
vard Journal of Law & Technology 11 (1998): 287, 316–17.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 286

2 8 6 N O T E S

15 As Coase wrote: “But it is a commonplace of economics that almost all resources

used in the economic system (and not simply radio and television frequencies) are lim-
ited in amount and scarce, in that people would like to use more than exists. Land, labor
and capital are all scarce, but this, of itself, does not call for government regulation. It is
true that some mechanism has to be employed to decide who, out of the many
claimants, should be allowed to use the scarce resource. But the way this is usually done
in the American economic system is to employ the price mechanism, and this allocates
resources to users without the need for government regulation.” R. H. Coase, “The Fed-
eral Communications Commission,” Journal of Law & Economics 2 (1959): 1, 14.
16 Hazlett has been prolific in advancing this argument. See, e.g., Hazlett and Sosa,

“Chilling the Internet? Lessons from FCC Regulation of Radio Broadcasting,” Michigan
Telecommunications & Technology Law Review 4 (1997–98): 35; Hazlett, “Physical
Scarcity, Rent Seeking, and the First Amendment,” Columbia Law Review 97 (1997):
905; Hazlett, “Assigning Property Rights to Radio Spectrum Users: Why Did FCC Li-
cense Auctions Take 67 Years?,” Journal of Law & Economics 41 (1998): 529; Hazlett,
“Spectrum Flash Dance: Eli Noam’s Proposal for ‘Open Access’ to Radio Waves,” Jour-
nal of Law & Economics 41 (1998): 805; Hazlett and Sosa, “Was the Fairness Doctrine a
‘Chilling Effect’? Evidence from the Postderegulation Radio Market,” Journal of Legal
Studies 26 (1997): 279; Hazlett, “The Rationality of U.S. Regulation of the Broadcast
Spectrum,” Journal of Law & Economics 23 (1990): 133.
17 See, e.g., Hazlett, “Spectrum Flash Dance,” 816. (“Profit maximization will force

competitive band managers to devise better technical means to increase wireless com-
munications traffic.”)
18 The source of the skepticism is traced, as Yochai Benkler describes, to Claude E.

Shannon, “Communication in the Presence of Noise,” Proceedings of the IRE 37 (1949):
10, and Claude E. Shannon, “A Mathematical Theory of Communication,” Bell System
Technology Journal 27 (1948): 379 and 623 (two-part publication). “These articles lay
out the theoretical underpinnings of direct sequencing spread spectrum.” Benkler, Over-
coming Agoraphobia, 323, note 171.
19 In order to be called “spread spectrum,” two conditions must be met: (1) the trans-

mitted signal bandwidth is greater than the minimal information bandwidth needed to
successfully transmit the signal; and (2) some function other than the information itself
is being employed to determine the resultant transmitted bandwidth. Robert C. Dixon,
“Why Spread Spectrum?,” IEEE Communication Society Magazine (July 1975): 21–25.
See also Yochai Benkler, “From Consumers to Users: Shifting the Deeper Structures
of Regulation Toward Sustainable Commons and User Access,” Federal Communica-
tions Law Journal 52 (2000): 561, 576–77. For an excellent study that links techno-
logical and policy questions, see Stuart Buck et al., “Spread Spectrum: Regulation in
Light of Changing Technologies” (1998), http://cyber.law.harvard.edu/ltac98/student-
papers.html. See also “The Unwired World” (Special Report), Scientific American 278
(1998): 69.
20 E-mail from David P. Reed to Lawrence Lessig, March 25, 2001.
21 For an introduction, see http://www.freesoft.org/CIE/Topics/60.htm. (This scheme,

one of many that can regulate access to a hardware medium, is referred to as CSMA/
CD, an acronym for “carrier sense, multiple access/collision detect.”) See also Doug-
las E. Comer, Internetworking with TCP/IP, 4th ed., vol. 1 (Upper Saddle River, N.J.:
Prentice-Hall, 2000), 28 (explaining how the Ethernet handles access and collision de-
tection through CSMA/CD).
22 Hazlett, “Physical Scarcity,” 928–29; Hazlett, “The Wireless Craze,” 10.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 287

N O T E S 2 8 7

23 For discussions of the technology, see Benkler, From Consumers to Users, 576–77;

Benkler, Overcoming Agoraphobia, 290 (“Technological developments in digital infor-
mation processing and wireless communications have made possible an alternative regu-
latory approach. It is now possible to regulate wireless communications as we do the
Internet—with minimal standard protocols—or the highway system—with limited gov-
ernmentally imposed rules of the road.”); ibid., part IV, 322–30 (describing examples of
current business models utilizing spread spectrum technologies where unlicensed op-
erations are permitted: proprietary infrastructure cellular network—Metricom’s Rico-
chet wireless network, ad hoc network of equipment owned by users—rooftop networks,
and publicly owned infrastructure of unlicensed devices—the NSF field tests).
24 See Steve Stroh, “Hollywood Star Was a Wireless Pioneer,” CLEC Magazine

(March–April 2000): 20. See also http://www.ncafe.com/chris/pat2/index.html (describ-
ing Lamarr’s contribution).
25 Anna Couey, “The Birth of Spread Spectrum: How ‘The Bad Boy of Music’ and

‘The Most Beautiful Girl in the World’ Catalyzed a Wireless Revolution—in 1941,”
http://www.sirius.be/lamarr.htm. Lamarr’s work inspired what is known as the “frequency-
hopping” form of spread spectrum. “Direct sequence,” or CDMA, was developed later.
26 For more on the technology, see Roger L. Peterson, Rodger E. Ziemer, and David E.

Borth, Introduction to Spread-Spectrum Communications (Englewood Cliffs, N.J.: Prentice-
Hall, 1995); Amer A. Hassan, John E. Hershey, and Gary J. Saulnier, Perspectives in
Spread Spectrum (Boston and London: Kluwer Academic, 1998).
27 Telephone interview with Dave Hughes, November 13, 2000.
28 Ibid.
29 See http://www.techweek.com/articles/8-23-99/wireless.htm.
30 See, e.g., U.S. Congress, Office of Technology Assessment, Telecommunications

Technology and Native Americans: Opportunities and Challenges, OTA-ITC-621 (Wash-
ington, D.C.: U.S. Government Printing Office, August 1995). See also statement
“FCC Takes Steps to Promote Access to Telecommunications on Tribal Lands” (FCC
news release, June 8, 2000).
31 The FCC has set aside three data bands as “unlicensed,” meaning users can deploy

devices that rely upon these bands without permission from the FCC or a spectrum
owner. The regulations governing this use are found in part 15 of the FCC’s rules, and
the bands are 915 MHz (902–928 MHz), 2.4 GHz (2,400–2,483.5 MHz), and 5.7 GHz
(5,725–5,850 MHz). Part 15 devices may not cause any harmful interference to autho-
rized services and must accept any interference that may be received. C.F.R., 47, §15.5.
Operation under these rules was limited to frequency-hopping and direct sequence
spread spectrum systems. But the FCC announced its intention to broaden the tech-
nologies permitted. See “In the Matter of Amendment of Part 15 of the Commission’s
Rules Regarding Spread Spectrum Devices” (DA 00–2317) (May 11, 2001); “Operation
of Radio Frequency Devices Without an Individual License,” 54 Fed. Reg. 17,710
(1989), codified at C.F.R., 47; pt. 2 and pt. 15 (1996). In addition, the FCC has six unli-
censed bands for non–spread spectrum use. They are the Citizens Band, the Radio Con-
trol Service Band, the Low Power Radio Service Band, the Wireless Telemetry Service
Band, the Medical Implant Communications Band, and the Family Radio Service Band.
32 802.11b refers to a standard developed by the IEEE. See http://www.manta.ieee.

org/groups/802/11/. The standard enables wireless protocols in the unlicensed spectrum
in the 2.5 GHz range at 11 Mbps.
33 See also James Gleick, “The Wireless Age: Theories of Connectivity” New York

Times Magazine, April 22, 2001 (discussing other wireless technologies).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 288

2 8 8 N O T E S

34 No doubt there are many who are competing to develop technologies for these un-

licensed options. For a description of the range of companies, see Amara D. Angelica,
“Wireless Internet Access,” TechWeek, http://www.techweek.com/articles/8-23-99/wireless.
htm.
35 This does not mean the regulation would necessarily be harmless. Bad regulation at

the device level could well inhibit innovation just as bad regulation at the spectrum level
has inhibited innovation.
36 These “free spectrum” or “open spectrum” advocates are all very different, though

they share a belief that spectrum should be managed differently. The differences among
them boil down to their attitude about scarcity. At one extreme is a group we might call
utopians. These people, such as Dave Hughes, Paul Baran, David Reed, and George
Gilder, believe that it is more likely than not that spectrum, properly used, would in
essence be unlimited. See, e.g., Paul Baran, “Is the UHF Frequency Shortage a Self-
Made Problem?,” (paper given at the Marconi Centennial Symposium, Bologna, Italy,
June 23, 1995, on file with the Columbia Law Review), 3; Gilder, Telecosm, 158–64;
David Hughes and Kambiz Hooshmand, “ABR Stretches ATM Network Resources,”
Data Communications 24, no. 5 (April 1995): 123.
At the other extreme is Eli Noam, who believes that spectrum is scarce and therefore
needs a system for rationing. See Eli Noam, “The Future of Telecommunications Regu-
lation,” NRRI Quarterly Bulletin 20 (1999): 17; Eli Noam, “Spectrum Auctions: Yes-
terday’s Heresy, Today’s Orthodoxy, Tomorrow’s Anachronism: Taking the Next Step
to Open Spectrum Access,” Journal of Law & Economics 41 (1998): 765; Eli Noam,
“Beyond Auctions: Open Spectrum Access,” in Regulators’ Revenge: The Future of Tele-
communications Deregulation, Tom W. Bell and Solveig Singleton, eds. (Cato Insti-
tute, 1998), 1: Eli Noam, “Will Universal Service and Common Carriage Survive
the Telecommunications Act of 1996?,” Columbia Law Review 97 (1997): 955; Eli
Noam, “Spectrum and Universal Service,” Telecommunications Policy 21 (1997); Eli
Noam, “Taking the Next Step Beyond Spectrum Auctions: Open Spectrum Access,”
IEEE Communications Magazine 33 (1995): 66 ; Eli Noam, “The Federal-State Friction
Built into the 1934 Act and Options for Reform,” in American Regulatory Federalism &
Telecommunications Infrastructure, Paul Teske, ed. (Hillsdale, N.J.: Lawrence Erlbaum
Associates, 1995), 113–14; Eli Noam, “Beyond Liberalization II: The Impending Doom
of Common Carriage,” Telecommunications Policy 18 (1994): 435; Eli Noam, “A Public
and Private-Choice Model of Broadcasting,” Public Choice 55 (1987): 163. His system
has the flavor of a market, but it is more the kind of market that controls access to sub-
ways than the ownership of taxis: just as a subway rider purchases a token at the moment
he or she needs to ride the subway, the user of spectrum, in real time, would purchase a
token that would assure his or her right to use spectrum. Unlike subway tokens, however,
these tokens would fluctuate in price as the demand for spectrum changes. During times
of great congestion, the price would go up; during times of low usage, the price would
fall. This market would be competitive, so the prices would be neutral, and hence the
proposal still fits my definition of a commons. But it is different from Gilder’s and Reed’s
in that it requires money to get access.
In the middle is Yochai Benkler, who is agnostic about the technology but clear about
the constitutional norm. See Yochai Benkler, “Siren Songs and Amish Children: Au-
tonomy, Information, and Law,” New York University Law Review 76 (2000): 23, 81,
where he talks about constitutionality. If spectrum is effectively unlimited, or if it could
be organized to be effectively unlimited, then it should be structured to permit free ac-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 289

N O T E S 2 8 9

cess. But if congestion is a problem, then we should allocate access to the spectrum to
minimize that congestion and possibly, if necessary, adopt a structure like Noam’s. See
also Benkler, “From Consumers to Users,” 561; Benkler, “Viacom-CBS Merger: From
Consumers to Users: Shifting the Deeper Structures of Regulation Toward Sustainable
Commons and User Access,” Federal Communications Law Journal 52 (2000): 561; Ben-
kler, “Free as the Air to Common Use: First Amendment Constraints on Enclosure of
the Public Domain,” New York University Law Review 74 (1999): 354; Benkler, “Com-
munications Infrastructure Regulation and the Distribution of Control over Content,”
Telecommunications Policy 22 (1998): 183; Benkler, “Overcoming Agoraphobia,” 287.
These are important differences, though in the end they matter in only one sense. No
one believes it has been proven that spectrum as it is presently used is unlimited. David
Reed points to the research of Tim Shepard and others demonstrating that a wireless
network could be structured so that an increase in the number of users actually in-
creases total capacity. Tim Shepard, Decentralized Channel Management in Scalable
Multihop Spread-Spectrum Packet Radio Networks, MIT, EECS thesis, 1995. See also
Timothy J. Shepard, “A Channel Access Scheme for Large Dense Packet Radio Net-
works,” at http://www.acm.org/pubs/articles/proceedings/comm/248156/p219-shepard/
p219-shepard.pdf. Reed argues that the “ ‘capacity’ of a free space radio network is not
fixed, but instead is an increasing function of the density of user ‘terminals’ in that space.”
Telephone interview with David Reed, February 7, 2001. This means that as more peo-
ple enter the shared spectrum space (as the number of terminals, that is, increases), the
available spectrum increases, not decreases. The more nodes there are on the network,
the closer these nodes are; the closer they are, the weaker the signal connecting these
nodes must be; the weaker the signal, the more signals there can be. A network of wire-
less nodes could expand spectrum capacity as the number of nodes increases.
But even without this increasing capacity, Baran and Hughes both argue that given ex-
isting capacity, properly deployed, we could fulfill all the need we have for delivering
data across the ether without any constraint at all.
The differences between those who see bandwidth as essentially unlimited and those
who see it as scarce obscure a more fundamental agreement: All would agree that spec-
trum use could undergo a radical shift—a paradigm change, in Eli Noam’s terms. All
would agree that this change would fundamentally alter the nature of our use of spec-
trum and would lead to an explosion of innovation in the use of spectrum that would not
otherwise, under either the government or the market model, exist. Everyone now con-
cedes that state-licensed spectrum has stifled innovation—the most glaring example was
the government’s stalling, because of the pressure of industry, the development of FM
radio. See Edwin R. Armstrong: A Man and His Invention (Eli Noam, ed., forthcoming).
All agree that the alternative of allowing spectrum to be sold is like “having the old
AT&T auction off the right to compete against itself.” Noam, “Beyond Spectrum Auc-
tions,” 473.
Noam and Benkler have an even stronger (from a legal perspective) argument against
the sale of spectrum. In a world where the control of spectrum was not “necessary,” as the
Supreme Court said it was in the NBC case, why was control of spectrum constitution-
ally permitted? Spectrum is speech, and the regulation of spectrum is the regulation of
speech. The constitutional status of spectrum auctions is rendered problematic by the
emergence of this alternative technology. Control is not “necessary” anymore, any more
than control of newspapers is necessary. It would certainly be unconstitutional to force
newspapers to buy a license to print (the way taxi drivers have to buy a medallion to drive
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 290

2 9 0 N O T E S

a taxi), so why is it any less unconstitutional to force a newspaper to buy the right to
broadcast?
The key to this use of spectrum would be a robust software-defined radio (SDR) tech-
nology, meaning a radio whose protocols get set by software, enabling easy switching
among different frequencies, carriers, and networks. Just as a modem, for example, can
switch among a number of protocols for modulating data transmissions, so too could a
SDR switch among protocols for communicating between radios. The switching in
SDRs is much greater than in modems, however, as it would include not only protocols,
but also power levels and modes of transmitting. But in principle, the system could allow
radios to sniff the environment and determine which kind of communication for that
environment would work best.
The government is doing some of the most important work in this area. DARPA,
which was responsible in part for the birth of the Internet, is experimenting with SDRs
that could communicate in battlefield contexts. To make this succeed, however, the gov-
ernment needs “a common architecture so that you’re not tied to one particular tech-
nology as an implementation of your next-generation radio.” Telephone interview with
Bill Lane, FCC, November 15, 2000. It is also important, as Intel has argued in com-
ments to the FCC, that the rules unbundle radio control from user application software,
so as to facilitate the broadest range of innovation at the software level. As Intel argues,
“[A] programmable platform would lower entry costs for the independent third party de-
velopers who played a key role in delivering innovative solutions elsewhere.” Intel,
“Comments of Intel Corporation, In the Matter of Authorization and Use of Software
Defined Radios,” ET Docket No. 00–47 (Washington), 4.
FCC chairman Michael Powell has a similar intuition about the potential:
“[A]dvanced technologies such as spread spectrum have ushered in all sorts of innova-
tive and efficient services. Indeed, rather than being a uniquely scarce resource[], spec-
trum has the potential to be a bottomless resource, unlike coal, oil, or timber[,] which
are more susceptible to depletion. Perhaps it is uniquely abundant rather than uniquely
scarce.” Michael K. Powell, “Willful Denial and First Amendment Jurisprudence,” Re-
marks Before the Media Institute, April 22, 1998 (transcript available at http://www.
fcc.gov/Speeches/Powell/spmkp808.html), 5 (emphasis added).
37 George Gilder, Telecosm: How Infinite Bandwidth Will Revolutionize Our World

(New York: Free Press, 2000), 160.

CHAPTER 6
1 Carol Rose, “The Comedy of the Commons: Custom, Commerce, and Inherently
Public Property,” University of Chicago Law Review 53 (1986): 711, 713.
2 Ibid., 712.
3 Ibid., 744.
4 See ibid., 752 (describing how “antiholdout” rules historically protected public

usage of roads from private actors who might “siphon off its public value”). The devel-
opment of the highway system has had strong direct (employment) and indirect (pro-
ductivity) effects on the United States economy. See M. Ishaq Nadiri and Theofanis P.
Mamuneas, “Contribution of Highway Capital to Output and Productivity Growth in
the U.S. Economy and Industries,” Department of Transportation Federal Highway Ad-
ministration, August 1998, available at http://www.fhwa.dot.gov/policy/gro98cvr.htm;
T. A. Heppenheimer, “The Rise of the Interstates,” American Herald of Inventions &
Technology 7 (1991): 8; Mark H. Rose, Interstate 15–67 (Knoxville: University of Ten-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 291

N O T E S 2 9 1

nessee Press, 1990). See also The New America That’s Coming (1956) (special report by
editors of Automotive Industries) (describing effect of “superhighways program on indus-
try, commerce and vehicle design”).
To the extent transportation has been privately owned, the dominant regulatory rule
has been common carriage. Railroads, for example, were privately owned; but they were
also subject to common carrier obligations. The effect of these common carrier regula-
tions was to render the common carrier a commons as well. See, e.g., Andrew A. Nimel-
man, “Of Common Carriage and Cable Access: Deregulation of Cable Television by
the Supreme Court,” Federal Communications Law Journal 34 (1982): 167, 173; Robert
Means and Deborah Cohn, “Common Carriage of Natural Gas,” Tulane Law Review 59
(1985): 529. As Yochai Benkler writes, however, “after the internal combustion engine
was invented, it was not a better system for awarding railroad franchises that was needed,
but a well-regulated commons like our national highway system.” Yochai Benkler, “The
Commons as a Neglected Factor of Information Policy” (paper presented at Tele-
communications Policy Research Center conference, October 3–5, 1998), 68. And in-
deed, the revenue from rail transportation was $28,348,895,000 in 1992, compared
with $135,437,000,000 for local and long-haul trucking services. See section 21,
“Transportation-Land,” of 1997 U.S. Economic Census conducted by U.S. Census Bu-
reau, available at http://www.census.gov/prod/2/gen/96statab/transind.pdf.
5 Rose, “The Comedy of the Commons,” 759, citing President of Cincinnati v. Lessee

of White, 31 U.S. (6 Pet.) 429 (1832) (recognizing an implied dedication of a square for
traditional public use). See also Hanoch Dagan and Michael A. Heller, “The Liberal
Commons,” Yale Law Journal 110 (2001): 549; Michael A. Heller, “The Tragedy of the
Anticommons: Property in the Transition from Marx to Markets,” Harvard Law Review
111 (1998): 621, 622–26; Alison Rieser, “Prescriptions for the Commons: Environmen-
tal Scholarship and the Fishing Quotas Debate,” Harvard Environmental Law Review 23
(1999): 393; Elinor Ostrom, Governing the Commons: The Evolution of Institutions for
Collective Action (Cambridge, England, and New York: Cambridge University Press,
1990), 2–23.
Henry Smith has identified a similar strategic cost in semicommons contexts. See
Henry E. Smith, “Semicommon Property Rights and Scattering in the Open Fields,”
Journal of Legal Studies 29 (2000): 131, 161–62.
6 Rose, “The Comedy of the Commons,” 769.
7 Robert Merges offers a complementing argument, focusing on interoperability and

the value brought by individuals’ investment in, for example, learning the commands in
a program. Robert Merges, “Who Owns the Charles River Bridge? Intellectual Property
and Competition in the Software Industry” (working paper, 1999).
8 The classic text supporting a broad range of open or free resources, building on the

work of Joseph Schumpeter, is Richard R. Nelson and Sidney G. Winter, An Evolution-
ary Theory of Economic Change (Cambridge, Mass.: Belknap Press of Harvard University
Press, 1982). As they write: “[I]nnovation in the economic system—and indeed the cre-
ation of any sort of novelty in art, science, or practical life—consists to a substantial extent
of a recombination of conceptual and physical materials that were previously in exis-
tence. The vast momentum of scientific, technological, and economic progress in the
modern world derives largely from the fact that each new achievement is not merely the
answer to a particular problem, but also a new item in the vast storehouse of components
that are available for use, in ‘new combinations,’ in the solution of other problems in the
future.” Ibid., 130. For a recent work advancing an analytic framework for evaluating “in-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 292

2 9 2 N O T E S

novation policy,” see Brett Frischmann, “Innovation and Institutions: Rethinking the
Economics of U.S. Science and Technology Policy,” Vermont Law Review 24 (2000): 347.
9 See David P. Reed, Jerome H. Saltzer, and David D. Clark, “Comment on Active

Networking and End-to-End Arguments,” IEEE Network 12, no. 3 (May–June 1998):
69–71.
10 Ostrom too has found that commons problems are best solved as commons where

environments are “uncertain and complex.” Ostrom, 88–89.
11 The argument is related to a point made by a number of scholars about patents.

Where the use of an invention is unknown, the transaction costs of licensing the inven-
tion are high. This can mean that patents over such inventions inhibit, rather than in-
duce, innovation. See Arti Kaur Rai, “Regulating Scientific Research: Intellectual
Property Rights and the Norms of Science,” Northwestern University Law Review 94
(1999): 77, 136–37; James Bessen and Eric Maskin, “Sequential Innovation, Patents,
and Imitation” (Cambridge University, Department of Economics, Working Paper
00–01, January 2000).
12 As David Reed puts the same point:

In an application of technologies and structure you don’t understand, there’s a great
value in creating the opportunity and framework in which innovators can experiment
and build on a platform that allows them [to] do what they want to do.
Telephone interview with David Reed, February 7, 2001.
13 Mark Gaynor et al., “Theory of Service Architecture,” 42. Baldwin and Clark use

a similar notion to explain the value of modular design. See Carliss Y. Baldwin and
Kim B. Clark, Design Rules, vol. 1 (Cambridge, Mass.: MIT Press, 2000), 234–37
(“Modularity creates design options and in so doing can radically change the market
value of a given set of designs.”).
14 Telephone interview with David Reed.
15 Though, as he acknowledges, the idea was first suggested by Dean Kim Clark. See

Kim B. Clark, “The Interaction of Design Hierarchies and Market Concepts in Tech-
nological Evolution,” Research Policy 14 (1985): 235–51. For a more recent study adding
support to Christensen’s thesis, see Richard Foster and Sarah Kaplan, Creative Destruc-
tion: Why Companies That Are Built to Last Underperform the Market—and How to Suc-
cessfully Transform Them (New York: Currency/Doubleday, 2001). There is also a link to
the work of William Abernathy, who has argued that “as designers worked on the agenda
of problems established by the dominant design, the locus of competition between firms
would shift from product improvement to cost reduction. A ‘productivity dilemma’
would then emerge, as the search for lower cost drove out, first, large innovations, and
later, all but the most minor innovations.” See Baldwin and Clark, 57.
The intuition behind Christensen’s work—that an existing firm holds psychological
commitments that make it hard for the firm to follow new leads—has support in a wide
range of work in innovation economics. See, e.g., Paul Romer, “Thinking and Feeling,”
American Economic Review 90 (2000): 439.
16 E-mail from David S. Isenberg to Lawrence Lessig, January 29, 2001.
17 Clayton M. Christensen, The Innovator’s Dilemma: The Revolutionary National

Bestseller That Changed the Way We Do Business (Cambridge, Mass.: Harvard Business
School Press, 1997).
18 Telephone interview with David Reed.
19 See Jim Carlton, Apple: The Inside Story of Intrigue, Egomania, and Business Blun-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 293

N O T E S 2 9 3

ders (New York: Times Books/Random House, 1997), 38–61 (describing Apple’s internal
licensing debate).
20 Baldwin and Clark, 11. In Baldwin and Clark’s terms, modularity describes an

“interdependence within and independence across modules.” Carliss Y. Baldwin and
Kim B. Clark, Design Rules, vol. 1 (Cambridge, Mass.: MIT Press, 2000), 63. Modules
are units “whose structural elements are powerfully connected among themselves and
relatively weakly connected to elements in other units.” They thus depend upon “a
framework—and architecture—that allows for both independence of structure and inte-
gration of function.” Ibid.
“Architecture” is thus fundamental to all modular systems. As Baldwin and Clark de-
scribe, “[T]he word is both an evocative term with a rich set of associations, and a tech-
nical term, meant to have a precise meaning in a particular context.” Ibid., 215. A proper
architecture enables parts to be broken into smaller units, with clean and clear interfaces
among the smaller modules.
Baldwin and Clark thus advance four “principles of modularity”: (1) Create nested,
regular, hierarchical structures in a complex system; (2) define independent compo-
nents within an integrated architecture; (3) establish and maintain rigorous partitions of
design information into hidden and visible subsets; (4) invest in clear interfaces and
“good” module tests. Ibid., 413. These together “made it possible for human beings of
individually limited capacity to design and produce ever more complex machines and
programs. [T]he resulting modular designs could change in unpredicted, yet coordi-
nated ways. They could be improved via a decentralized, value-seeking process, which
we have called design evolution.” Ibid.
21 See Michael Walzer, Spheres of Justice: A Defense of Pluralism and Equality (New

York: Basic Books, 1984).
22 See, e.g., Margaret Jane Radin, Contested Commodities (Cambridge, Mass.: Har-

vard University Press, 1996), 132–36.
23 Letter from Thomas Jefferson to Isaac McPherson (August 13, 1813), in The Writ-

ings of Thomas Jefferson 6 (H. A. Washington, ed., 1861), 175, 180.
24 Rose, “The Comedy of the Commons,” 742.
25 Some researchers have suggested that there is a way of organizing access to

spectrum that would increase the spectrum as the number of users increases. See
Timothy J. Shepard, “A Channel Access Scheme for Large Dense Packet Radio Net-
works,” at http://www.acm.org/pubs/articles/proceedings/comm/248156/p219-shepard/
p219-shepard.pdf. See also Rick Boyd-Merritt, “Engineer’s Wireless Internet in a Box
Draws Interest,” EE Times.com, http://www.eetimes.com/story/OEG19981007S0014.
26 The term is of recent origin. Professor Fisher traces its origins to the late nineteenth

century; William W. Fisher III, “The Growth of Intellectual Property: A History of the
Ownership of Ideas in the United States” (1999) 2, 8, available at http://cyber.law.
harvard.edu/ipcoop/97fish.1.html. The term appears to have been used twice before
1900; Mitchell v. Tilghman, 86 U.S. 287 (1873) and Davoll v. Brown, 7 Fed. Cas. 197
(Cir. Ct., D. Mass. 1845). These early usages were essentially translations of European
documents, except for Davoll, where the court uses the term in exactly its modern sense:

Only thus can ingenuity and perseverance be encouraged to exert themselves in this way
usefully to the community; and only in this way can we protect intellectual property, the
labors of the mind, productions and interests as much a man’s own, and as much the fruit
of his honest industry, as the wheat he cultivates, or the flocks he rears.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 294

2 9 4 N O T E S

Its appearance in federal cases since the turn of the twentieth century has grown dra-
matically:

Number of Per 100k
Decade References cases
1900–1919 1
1920–1929 0
1930–1939 4 1.163
1940–1949 9 3.399
1950–1959 15 5.038
1960–1969 11 2.827
1970–1979 56 8.213
1980–1989 341 26.42
1990–1999 1721 86.37
2000–2001 466

I am grateful to Professor Hank Greely for these data.
Of course, in some sense intellectual property is certainly property; see Frank
Easterbrook, “Intellectual Property Is Still Property,” Harvard Journal of Law & Public
Policy 13 (1991): 108. But to the untrained, the connotations of the term property will not
suggest the limited nature this “property” is to have. As Vaidhyanathan argues:
It is essential to understand that copyright in the American tradition was not meant to be
a “property right” as the public generally understands property. It was originally a narrow
federal policy that granted a limited trade monopoly in exchange for universal use and
access.
Siva Vaidhyanathan, Copyrights and Copywrongs: The Rise of Intellectual Property
and How It Threatens Creativity (New York: New York University Press, 2001), 21.
27 Jessica Litman, Digital Copyright (Amherst, N.Y.: Prometheus Books, 2000), 15.
28 See, e.g., Pamela Samuelson et al., “A Manifesto Concerning the Legal Protection

of Computer Programs,” Columbia Law Review 94 (1994): 2308, 2427; Raymond Shih
Ray Ku, “Copyright & Cyberspace: Napster and the New Economics of Digital Tech-
nology” (draft on file with author, April 7, 2001), 9 (“[W]hile digital technology facili-
tates the copyright and distribution of digital information, it also permits greater control
over the use and distribution of information.”).

CHAPTER 7
1 As Judge Posner writes, distinguishing the rules for land from the rules for copy-
righted material:
One reason [the two are different] is that it is more inefficient to have unowned land
lying around (say, as the result of the expiration of a time-limited property right) than to
have unowned intellectual property. Ideally, all land should be owned by someone, to
prevent the congestion externalities that we discussed in connection with the natural pas-
ture from arising. But . . . there is no parallel problem concerning information and ex-
pression. A’s use of some piece of information will not make it more costly for B to use
the same information.
Richard A. Posner, Economic Analysis of Law, 4th ed. (Boston: Little, Brown, 1992), 41.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 295

N O T E S 2 9 5

2 “Works are not crafted out of thin air.” James Boyle, Shamans, Software, and Spleens:

Law and the Construction of the Information Society (Cambridge, Mass.: Harvard Uni-
versity Press, 1996), 57.
3 Feist Publications, Inc. v. Rural Telephone Service Co., Inc., 499 U.S. 340, 345–47

(1991).
4 Siva Vaidhyanathan, Copyrights and Copywrongs: The Rise of Intellectual Property

and How It Threatens Creativity (New York: New York University Press, 2001), 203.
5 The United States finally extended copyright protection to foreign publishers

through the International Copyright Act of 1891, ch. 565, §13, 26 Stat. 1110 (1891). Be-
fore 1891, “the United States was notorious for its singular and, in many regards, cava-
lier attitude toward the intellectual property of foreigners.” William P. Alford, “Making
the World Safe for What? Intellectual Property Rights, Human Rights and Foreign Eco-
nomic Policy in the Post–European Cold War World,” New York University Journal of
International Law & Politics 29 (1997): 135, 146. See also Jessica Litman, Digital Copy-
right (Amherst, N.Y.: Prometheus Books, 2000), 15.
6 Richard A. Posner, Law and Literature, rev. and enlarged ed. (Cambridge, Mass.:

Harvard University Press, 1998), 389. The first United States case to decide the question
comes in 1853, when a circuit court held that the copyright to Uncle Tom’s Cabin did
not reach a German translation of the same work. Vaidhyanathan, 92–93.
7 John Tebbel, A History of Book Publishing in the United States: The Creation of an

Industry 1630–1865 (New York: R. R. Bowker, 1972), 141.
8 These laws were Pub. L. No. 87-668, 76 Stat. 555 (1962); Pub. L. No. 89-142, 79

Stat. 581 (1965); Pub. L. No. 90-141, 81 Stat. 464 (1967); Pub. L. No. 90-416, 82 Stat.
397 (1968); Pub. L. No. 91-147, 83 Stat. 360 (1969); Pub. L. No. 91-555, 84 Stat. 1441
(1970); Pub. L. No. 92-170, 85 Stat. 490 (1971); Pub. L. No. 92-566, 86 Stat. 1181
(1972); Pub. L. No. 93-573, title I, §104, 88 Stat. 1873 (1974).
9 On books, see Stephen Breyer, “The Uneasy Case for Copyright: A Study of Copy-

right in Books, Photocopies, and Computer Programs,” Harvard Law Review 84 (1970):
281 (acknowledging the economic rationale for copyright protection of books and films,
but not software). The MPAA estimates the average cost of a feature film (including stu-
dio overhead and capitalized interest) was $51.5 million in 1999. See MPAA, “MPAA
Average Negative Costs,” slide 14 of 44 (visited June 21, 2001), http://www.mpaa.org/
useconomicreview/2000Economic/slide.asp?ref=14.
10 As Yochai Benkler writes, “[M]ainstream economics very clearly negates the super-

stition that if some property rights in information are good, then more rights in informa-
tion are even better.” Yochai Benkler, “A Political Economy of the Public Domain:
Markets in Information Goods Versus the Marketplace of Ideas,” in Expanding the
Boundaries of Intellectual Property: Innovation Policy for the Knowledge Society, Ro-
chelle Cooper Dreyfuss and Diane Leenheer Zimmerman, eds. (Oxford: Oxford Uni-
versity Press, 2001), 267, 271. As Vaidhyanathan argues:
Through a series of case studies in different media through the 20th Century, it argues
for ‘thin’ copyright protection: just strong enough to encourage and reward aspiring
artists, writers, musicians, and entrepreneurs, yet porous enough to allow full and rich
democratic speech and the free flow of information.
Vaidhyanathan, 8.
The skepticism among economists about perfect or extremely strong copyright pro-
tection is well known. For an expansive economic account, see Richard Watt, Copyright
and Economic Theory: Friends or Foes? (Cheltenham, England, and Northampton,
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 296

2 9 6 N O T E S

Mass.: E. Elgar, 2000). For a rich philosophical survey of justifications for copyright, see
Peter Drahos, A Philosophy of Intellectual Property (Aldershot, England, and Brookfield,
Vt.: Dartmouth Publishing Company, 1996).
11 Posner, Law and Literature, 391. A related point is made by Watt, who describes

conditions under which piracy of copyrighted work is in fact favorable to the copyright
owner. See Richard Watt, Copyright and Economic Theory: Friends or Foes? (Chel-
tenham, England, and Northampton, Mass.: E. Elgar, 2000), 58–67, 201 (“some copy-
right ‘piracy’ is highly likely to be socially efficient”). As Watt concludes, economic
theory then is guilty of pointing out that there exist cases in which legal copyright pro-
tection hampers rather than helps society in general. Perhaps more surprisingly, econo-
mists can show that legal copyright protection can also hamper copyright holders and
producers of originals themselves. Hence, “economic theory can perhaps best be
thought of as throwing out a warning to copyright advocates, that they should take care
not to lobby for policy that ends up damaging the interests of copyright holders, or those
of the society in general.” See ibid., 200.
12 Twentieth Century Music Corp. et al. v. Aiken, 422 U.S. 151, 154–55 (1975).
13 While Fourneaux is credited with inventing the first player piano, in 1902, Melville

Clark was the first to create one with the full eighty-eight-key range of the standard
piano. Clark was also one of the first to produce the player and the piano combined in a
self-contained unit. See Harvey Roehl, Player Piano Treasury: The Scrapbook History of
the Mechanical Piano in America as Told in Story, Pictures, Trade Journal Articles and
Advertising (Vestal, N.Y.: Vestal Press, 1961); Arthur W. J. G. Ord-Hume, Pianola: The
History of the Self-Playing Piano (London: George Allen & Unwin, 1984).
14 Edward Samuels, The Illustrated Story of Copyright (New York: St. Martin’s Press,

2000), 34.
15 White-Smith Music Publishing Co. v. Apollo Co., 209 U.S. 1, 21 (1908).
16 Congress’s initial statute was Act of March 4, 1909, ch. 320(e), 35 Stat. 1075 (1909),

superseded by 17 U.S.C. §115 (1988). See generally Fred H. Cate, “Cable Television
and the Compulsory Copyright License,” Federal Communications Law Journal 42
(1990): 191; C. H. Dobal, “A Proposal to Amend the Cable Compulsory License Provi-
sions of the 1976 Copyright Act,” Southern California Law Review 61 (1988): 699; Paul
Glist, “Cable Copyright: The Role of the Copyright Office,” Emory Law Journal 35
(1986): 621; Stanley M. Besen, Willard G. Manning Jr., and Bridger M. Mitchell,
“Copyright Liability for Cable Television: Compulsory Licensing and the Coase Theo-
rem,” Journal of Law & Economics 21 (1978): 67.
Robert Merges has argued that compulsory rights do create problems in contexts such
as this, and that some property rights will induce the creation of independent institutions
that could most cheaply negotiate the rights. See Robert P. Merges, “Contracting into
Liability Rules: Intellectual Property Rights and Collective Rights Organizations,” Cali-
fornia Law Review 84 (1996): 1293; Robert P. Merges, “Institutions for Intellectual Prop-
erty Transactions: The Case of Patent Pools,” in Expanding the Boundaries of Intellectual
Property, Rochelle Cooper Dreyfuss and Diane Leenheer Zimmerman, eds. (Oxford:
Oxford University Press, 2001), 131. Ian Ayres and Eric Talley reach a very different con-
clusion. See Ian Ayres and Eric Talley, “Solomonic Bargaining: Dividing a Legal Enti-
tlement to Facilitate Coasean Trade,” Yale Law Journal 104 (1995): 1027, 1092–94
(arguing that a liability rule will induce parties to reveal their true valuations and hence
is more likely to produce a Coasean trade).
17 See Copyright Act of 1909, ch. 320, §(e), 35 Stat. 1075 (1909), superseded by 17

U.S.C. §115 (1982).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 297

N O T E S 2 9 7

18 Fortnightly Corp. v. United Artists Television, Inc., 392 U.S. 390 (1968), and

Teleprompter Corp. v. Columbia Broadcasting System, Inc., 415 U.S. 394 (1974).
19 The compulsory right was incorporated in §111 of the 1976 act. As Paul Goldstein

has written, “[M]ore explicitly than any other aspect of the [1976 Act], it commits its
operation to assumptions about industry structure and regulation.” “Preempted State
Doctrines, Involuntary Transfers and Compulsory Licenses: Testing the Limits of Copy-
right,” U.C.L.A. Law Review 24 (1977): 1107, 1127–35. See also Melville B. Nimmer
and David Nimmer, “Nimmer on Copyright,” §8.18[E], 4.
20 See generally Samuels, 181–82. The satellite TV retransmission right was enacted in

1988. See Satellite Home Viewer Act, Act of November 16, 1988, Pub. L. No. 100-667,
102 Stat. 3935, codified at 17 U.S.C. §119 (supp. 1993). Jukeboxes were covered by the
1976 act, 17 U.S.C. §116. That provision was in tension with the Berne Convention,
which forbade compulsory licenses for public performances. In 1989, Congress added
§116A, which added negotiated agreements between performance rights associations. In
1993, Congress then repealed the original §116 and renamed §116A to §116. See Scott
M. Martin, “The Berne Convention and the U.S. Compulsory License for Jukeboxes:
Why the Song Could Not Remain the Same,” Journal of the Copyright Society U.S.A. 37
(1990): 262.
21 As Calabresi and Melamed describe, a resource is protected with a liability rule

when one using the resources must pay compensation for the use. The resource is pro-
tected by a property rule when one using the resource must negotiate for it before it can
be taken. Guido Calabresi and Douglas Melamed, “Property Rules, Liability Rules, and
Inalienability: One View of the Cathedral,” Harvard Law Review 85 (1972): 1089, 1092.
When a copyright is protected by a liability rule, those wishing to use the resource can
take the resource, as long as they pay the liability price. When it is protected by a prop-
erty rule, taking the resource without paying for it can make one criminally liable.
Robert P. Merges, “Institutions for Intellectual Property Transactions: The Case of
Patent Pools,” in Expanding the Boundaries of Intellectual Property, R. Dreyfuss, ed. (Ox-
ford: Oxford University Press, 2001), 131–32.
22 “Chicago school” analysts argued that a monopolist possesses a fixed amount of

market power and therefore can extract only a fixed amount of monopoly profit from
consumers, whether from one market or several. On this basis, they concluded that
leverage of monopoly power from one market into another is impossible. See, e.g.,
Robert H. Bork, The Antitrust Paradox: A Policy at War with Itself (New York: Basic
Books, 1978); Richard A. Posner, Antitrust Law: An Economic Perspective (Chicago: Uni-
versity of Chicago Press, 1976). More recent economic analyses have demonstrated sev-
eral mechanisms by which market power in one market can be used to harm
competition in another market. As Steven Salop and Craig Romaine put it:

Post-Chicago economic analysis has suggested that there are a number of limiting
assumptions required for this single monopoly profit theory to apply. When these
assumptions are relaxed, the theory’s strong result and the public policy implications
no longer hold. There are a number of common market situations in which integration
into a second market may raise anticompetitive concerns. These include markets
in which the first monopoly is regulated, markets that are characterized by economies
of scale and scope and in which the inputs are not used in fixed proportions, and mar-
kets with multiple types of buyers. In such markets, it is possible for a monopolist to
profitably extend its power into a second market and harm consumers. (footnote
omitted)
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 298

2 9 8 N O T E S

Steven C. Salop and R. Craig Romaine, “Preserving Monopoly: Economic Analysis,
Legal Standards, and Microsoft,” George Mason Law Review 7 (1999): 617, 625. See also
Michael D. Whinston, “Tying, Foreclosure, and Exclusion,” American Economic Re-
view 80 (1990): 837 (demonstrating multiple situations in which foreclosure of the tied
market can occur, including a case where the monopolist can precommit to the tie
through product design or production processes); Janusz A. Ordover, Garth Saloner, and
Steven C. Salop, “Equilibrium Vertical Foreclosure,” American Economic Review 80
(1990): 127 (integration across multiple products permits competitor to exclude uninte-
grated rival). See also Louis Kaplow, “Extension of Monopoly Power Through Lever-
age,” Columbia Law Review 85 (1985): 515 (expressing early skepticism about the
Chicago school analysis). See also Dennis W. Carlton and Michael Waldman, “The
Strategic Use of Tying to Preserve and Create Market Power in Evolving Industries”
(September 1998 working paper) (arguing that tying deters entry in primary tying
markets in addition to providing leverage into tied markets).
23 See Douglas Abell, “Pay-for-Play,” Vanderbilt Journal of Entertainment Law & Prac-

tice 2 (2000): 52.
24 See 17 U.S.C. §111 (2000). The statute initially set up a Copyright Royalty Tri-

bunal, but that was abolished in favor of private negotiation in 1993. See Robert P.
Merges, Peter S. Menell, and Mark A. Lemley, Intellectual Property in the New Techno-
logical Age, 2nd ed. (Gaithersburg: Aspen Law & Business, 2000), 481.
Congress initially tried to balance this effective control by establishing rules of access
such as the “Fairness Doctrine.” See Jerry Kang, Communications Law and Policy: Cases
and Materials (Gaithersburg, Md.: Aspen Law & Business, 2001), 85–86. These rules
have been drawn into constitutional doubt—see, for example, Miami Herald Publishing
Co. v. Tornillo, 418 U.S. 241 (1974), Huddy v. FCC, 236 F.3d 720, 723 (D.C. Cir.
2001)—and are generally viewed as a failure. See L. A. Scot Powe, American Broad-
casting and the First Amendment (Berkeley: University of California Press, 1987),
197–209.
25 As FCC chairman Powell has described it:

In 1969, broadcasting consisted of a handful of radio stations in any given market plus
two or three television stations affiliated with one of the three major networks. Occa-
sionally, larger markets had an independent station too. Three major networks held
more than 90% of the market for video programming. Not so anymore. Not only has the
market share of the three largest networks been eroded by cable programming, the last
time I looked there were about seven “declared” national television networks. . . . Obvi-
ously, things have changed a lot. . . . In our current technological environment, it can
reasonably be argued that there is a bounty, not a scarcity of outlets for expressing one’s
viewpoint. In the traditional broadcasting arena, the numbers are impressive: There are
1,207 commercial TV stations and 367 noncommercial stations. There are also some
5,000 TV translators and 2,000 low-power TV stations. In addition, there are almost
12,500 radio stations.
Michael K. Powell, “Willful Denial and First Amendment Jurisprudence, Remarks
Before the Media Institute,” April 22, 1998 (transcript available at http://www.fcc.gov/
Speeches/Powell/spmkp808.html).
26 See Robert M. Fano, “On the Social Role of Computer Communications,” Pro-

ceedings of the IEEE 60 (September 1972): 1249.
27 Simson L. Garfinkel, Architects of the Information Society: Thirty-Five Years of the

Laboratory for Computer Science at MIT (Cambridge, Mass.: MIT Press, 1999), 8–9.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 299

N O T E S 2 9 9

28 Ronald Coase, “Looking for Results,” interviewed by Thomas W. Hazlett, Reason

(January 1997).
29 See Linda Shrieves, “When It’s Your Turn, Here’s Why You’re Served a Chorus

of . . . ; The Birthday Song Is Still Copyrighted and Nets Nearly $1 Million a Year in
Royalties,” Orlando Sentinel, February 27, 2001, E1.
30 Or at least not yet. George Gilder has repeatedly argued that a future infrastructure

based on fiber optics would provide “infinite bandwidth.” See George Gilder, Telecosm:
How Infinite Bandwidth Will Revolutionize Our World (New York: Free Press, 2000);
George Gilder, “Rulers of the Rainbow: The New Emperors of the Telecosm Will Use
the Infinite Spectrum of Light—Visible and Invisible—to Beef up Bandwidth,” Forbes
ASAP (October 1998): 104; and George Gilder, “Into the Fibersphere (Fiber Optics),”
Forbes (December 1992): 111.
31 Ben H. Bagdikian, The Media Monopoly, 6th ed. (Boston: Beacon Press, 2000), 4.
32 Robert W. McChesney, Rich Media, Poor Democracy: Communication Politics in

Dubious Times (Urbana: University of Illinois Press, 1999), 18.
33 The reasons for this increased concentration are hard to track precisely. There are a

number of changes that have certainly occurred. The relaxation of rules on ownership of
radio stations, for example, has exploded concentration in radio station ownership. This,
in turn, has led to an increase in the modern equivalent of “payola.” See Douglas Abell,
“Pay-for-Play,” Vanderbilt Journal of Entertainment Law & Practice 2 (2000): 52. As
Boehlert describes it:
There are 10,000 commercial radio stations in the United States; record companies rely
on approximately 1,000 of the largest to create hits and sell records. Each of those 1,000
stations adds roughly three new songs to its playlist each week. The [independents] get
paid for every one; $1,000 on average for an “add” at a Top 40 or rock station, but as high
as $6,000 or $8,000 under certain circumstances.
Eric Boehlert, “Pay for Play,” Salon, March 14, 2001, http://www.salon.com/ent/
feature/2001/03/14/payola/print.html, 2.
34 Allyson Lieberman, “Sagging Warner Music out of Tune with AOL TW,” New York

Post, April 19, 2001, 34. See also Charles Mann, “The Heavenly Jukebox,” Atlantic
Monthly (September 2000), 39, 53.
35 Boehlert, 2.
36 McChesney, 18.
37 Ibid., 17.
38 Ibid., 33.
39 Ibid., 18.
40 According to the National Cable Television Association’s figures, the top seven

“multiple system operators,” or MSOs, controlled 90 percent of the national cable tele-
vision market at the end of 2000. http://www.ncta.com/industry_overview/top50mso.cfm.
See also Richard Waters, “Appeals Court Overrules Curbs on Cable TV Ownership in
U.S. Federal Rules,” Financial Times, March 3, 2001, 7. As of March 2001, AOL–Time
Warner’s cable market share was about 20 percent, while AT&T’s share stood at 42 per-
cent (including AT&T’s purchase of MediaOne Group and its 25.5 percent stake in
Time Warner Entertainment). AT&T’s 42 percent market share well exceeds the FCC’s
cap of 30 percent, leading AT&T to challenge the cap in court as being “arbitrary.” See
Edmund Sanders and Sallie Hofmeister, “Court Rejects Limits on Cable Ownership;
Television: Controversial 30% Cap Is Deemed Unconstitutional, but Consumer Groups
Call the Decision ‘Devastating,’ ” Los Angeles Times, March 3, 2001, C1.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 300

3 0 0 N O T E S

41 Bagdikian, 4.
42 Ibid., x.
43 See Mike Hoyt, “With ‘Strategic Alliances,’ the Map Gets Messy,” Columbia Jour-

nalism Review (January–February 2000), at http://www.cjr.org/year/00/1/hoyt.asp;
Global Media Economics: Commercialization, Concentration and Integration of World
Media Markets, Allan B. Albarran and Sylvia M. Chan-Olmsted, eds. (Ames: Iowa State
University Press, 1998), 19–31; Dennis W. Mazzocco, Networks of Power: Corporate T.V.’s
Threat to Democracy (Boston, Mass.: South End Press, 1994), 1–8. Cf. Benjamin M.
Compaine, “Distinguishing Between Concentration and Competition,” in Who Owns
the Media, 3rd ed., Benjamin M. Compaine and Douglas Gomery, eds. (Mahwah, N.J.:
L. Erlbaum Associates, 2000), 537; Douglas Gomery, “Interpreting Media Ownership,”
in Who Owns the Media?, 3rd ed., Benjamin M. Compaine and Douglas Gomery, eds.
(Mahwah, N.J.: L. Erlbaum Associates, 2000), 507.
44 Bagdikian, 7. There are, of course, many who believe there is no necessary link be-

tween the mergers and these features of modern media. See, e.g., Steven Rattner, “A
Golden Age of Competition,” in Media Mergers, Nancy J. Woodhull and Robert W.
Snyder, eds. (New Brunswick, N.J.: Transaction Publishers, 1998), 9. See also Bruce M.
Owen, Economics and Freedom of Expression: Media Structure and the First Amendment
(Cambridge, Mass.: Ballinger Publishing Company, 1975).
45 Andrew Kreig, Spiked: How Chain Management Corrupted America’s Oldest News-

paper (Old Saybrook, Conn.: Peregrine Press, 1987).
46 Bagdikian, 30.
47 McChesney, 245.
48 Compare Judge Posner’s comment: “[T]he management of a large publicly held

corporation will have difficulty finding issues on which a partisan stand would not alien-
ate large numbers of shareholders.” Posner, Economic Analysis of Law, 674.
49 Bagdikian, 129.
50 Ibid., 35.
51 McChesney, 80, 179.
52 Ibid., 250.
53 Ibid., 148.
54 Ibid., 168.

CHAPTER 8
1 See Lawrence Lessig, Code and Other Laws of Cyberspace (New York: Basic Books,
1999).
2 For examples of on-line mapping services, see MapQuest.com at http://www.

mapquest.com; Maps On Us at http://www.mapsonus.com; and MapBlast! at http://www.
mapblast.com.
3 For examples of on-line translation Web sites, see AltaVista World/Translate at

http://world.altavista.com; FreeTranslation.com at http://www.freetranslation.com; and
From Language to Language at http://www.langtolang.com.
4 A short list of many examples of on-line dictionaries includes Merriam-Webster

OnLine at http://www.m-w.com; Cambridge Dictionaries Online at http://dictionary.
cambridge.org; and AllWords.com at http://www.allwords.com. There are also sites that
perform aggregate searches through multiple multilingual dictionaries, such as
yourDictionary.com at http://www.yourdictionary.com.
5 As we’ll see in chapter 11, this is not a slight constraint. Because his site was non-

commercial, Eldred could include only work that had fallen into the public domain.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 301

N O T E S 3 0 1

When Eldred began, the content constraint meant that works published before 1923
were free, works published after 1923 were only possibly free. But in 1998, Congress
changed that by passing the Sonny Bono Copyright Term Extension Act. The Bono Act
extended the term of existing copyrights by twenty years, meaning work that was to fall
into the public domain in 1999 would now not fall into the public domain until 2019.
As we’ll see, this turned Eldred into an activist.
6 See http://www.apple.com/hotnews/articles/2001/03/imacdirector/.
7 There are skeptics, however, about whether diversity will increase. Le Duc, for ex-

ample, argues that the real constraint on diversity in films is not the channels of distri-
bution, but rather the limited attention viewers have for stars. There are only so many
stars we can like; they are the true constraint on this mode of production; and as long as
that limit remains, the range of film will be restricted as well. Don R. Le Duc, Beyond
Broadcasting: Patterns in Policy and Law (New York: Longman, 1987), 128.
8 On the risk of liability, see JaNet Kornblum, “Lyrics Site Takes Steps to Avoid Nap-

ster Woes,” USA Today, December 12, 2000, available at http://www.usatoday.com/life/
cyber/tech/jk121200.htm.
9 Tom Parsons, “World Wide Web Gives Poets, Poetry Room to Grow,” Dallas Morn-

ing News, July 30, 2000, 8J.
10 For example, a Web site devoted to Chaucer uses multiple frames to navigate

quickly through The Canterbury Tales and define the medieval English terms. See Li-
brarius at http://www.librarius.com/cantales.htm. Another site provides a high-tech mul-
timedia companion to the printed Anthology of Modern American Poetry. See Modern
American Poetry at http://www.english.uiuc.edu/maps.
11 See Favorite Poem Project at http://www.favoritepoem.org (featuring readings of

famous poems read by individual Americans); Internet Poetry Archive at http://www.
ibiblio.org/ipa (offering modern poetry readings by the poets); and e-poets.net at
http://www.e-poets.net (featuring contemporary audio poetry).
12 Free in the sense that I have defined the term. The technology is offered under a

nondiscriminatory license. The underlying technology is patented. See Brad King,
“MP3.com Open to Friends,” Wired News, January 19, 2001, at http://www.wired.com/
news/mp3/0,1285,41195,00.html. “Vorbis” is an alternative to MP3 that is royalty free
and compresses more than the MP3 format does. Vito Pilieci, “MP3 May Go Way of
Eight-Track: Vorbis Audio File Players Would Be Free of Royalty, Patent Fees,” National
Post, June 29, 2000, C8. The licensing does, however, create problems for open code de-
velopers. See Wendy C. Freedman, “Open Source Movement Vies with Classic IP
Model, Free Software Is Bound to Have a Significant Effect on Patent, Copyright, Trade
Secret Suits,” National Law Journal 22 (March 13, 2000): B14.
13 Telephone interview with Michael Robertson, November 16, 2000.
14 Courtney Love, “Courtney Love Does the Math,” Salon (June 12, 2000): 5. Love

has offered a slightly exaggerated but illustrative description of how the market for music
now works:

This story is about a bidding-war band that gets a huge deal with a 20 percent royalty rate
and a million-dollar advance. (No bidding-war band ever got a 20 percent royalty, but
whatever.) This is my “funny” math based on some reality and I just want to qualify it by
saying I’m positive it’s better math than what Edgar Bronfman Jr. [the president and
CEO of Seagram, which owns Polygram] would provide. What happens to that million
dollars? They spend half a million to record their album. That leaves the band with
$500,000. They pay $100,000 to their manager for 20 percent commission. They pay
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 302

3 0 2 N O T E S

$25,000 each to their lawyer and business manager. That leaves $350,000 for the four
band members to split. After $170,000 in taxes, there’s $180,000 left. That comes out to
$45,000 per person. That’s $45,000 to live on for a year until the record gets released.
The record is a big hit and sells a million copies. (How a bidding-war band sells a million
copies of its debut record is another rant entirely, but it’s based on any basic civics-class
knowledge that any of us have about cartels. Put simply, the antitrust laws in this country
are basically a joke, protecting us just enough to not have to rename our park service the
Phillip Morris National Park Service.) So, this band releases two singles and makes two
videos. The two videos cost a million dollars to make and 50 percent of the video pro-
duction costs are recouped out of the band’s royalties. The band gets $200,000 in tour
support, which is 100 percent recoupable. The record company spends $300,000 on in-
dependent radio promotion. You have to pay independent promotion to get your song on
the radio; independent promotion is a system where the record companies use middle-
men so they can pretend not to know that radio stations—the unified broadcast system—
are getting paid to play their records. All of those independent promotion costs are
charged to the band. Since the original million-dollar advance is also recoupable, the
band owes $2 million to the record company. If all of the million records are sold at full
price with no discounts or record clubs, the band earns $2 million in royalties, since their
20 percent royalty works out to $2 a record. Two million dollars in royalties minus $2 mil-
lion in recoupable expenses equals . . . zero! How much does the record company make?
They grossed $11 million. It costs $500,000 to manufacture the CDs and they advanced
the band $1 million. Plus there were $1 million in video costs, $300,000 in radio pro-
motion and $200,000 in tour support. The company also paid $750,000 in music pub-
lishing royalties. They spent $2.2 million on marketing. That’s mostly retail advertising,
but marketing also pays for those huge posters of Marilyn Manson in Times Square and
the street scouts who drive around in vans handing out black Korn T-shirts and back-
wards baseball caps. Not to mention trips to Scores and cash for tips for all and sundry.
Add it up and the record company has spent about $4.4 million. So their profit is $6.6
million; the band may as well be working at a 7-Eleven.

Ibid. Compare Senator Hatch’s very different account of music production outside
the control of the labels:

I will quote him at length, because his experience is instructive. He said: “As a result of
doing it on my own, I get about $7 for every CD that sells in a store. And about $10 per
CD sold at concerts. In contrast, I’ve got a friend who is also a performer/songwriter who
opted to sign with a . . . label. He recorded a CD that cost about $18,000 to make, which
the label paid for. Now, when one of his CDs sells at a store or at a concert, he makes
about $1. The rest of the $7–$10 which I make on my CD sales goes to his label. On top
of that, he has to pay back the $18,000 it cost to make the CD out of his $1-per-CD cut.
In other words, he won’t make a dime until he has sold 18,000 CDs. And then, he still
won’t own the CD, the label will. They maintain the copyright. It’s kind of like paying
off your mortgage and then having the bank still own your house.

Orrin G. Hatch, “Address of Senator Orrin G. Hatch Before the Future of Music
Coalition,” Future of Music Coalition, January 10, 2001, 2. Currently, the average album
release is under twenty-five thousand per CD. Jon Healey, “Industry Seeks to Justify Huge
Overhead on the Price of Compact Disks,” Knight Ridder Tribune Business News, Sep-
tember 3, 2000.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 303

N O T E S 3 0 3

15 Telephone interview with Michael Robertson.
16 Ibid.
17 Ibid.
18 See Testimony of the Future of Music Coalition on “Online Music and Copyright

Law,” submitted to Senate Judiciary Committee, April 3, 2001, 13 (“The fastest-growing
demographic segment using Napster are adults over the age of 24. Research reports have
confirmed that one of the major reasons that they are doing so is to access commercial
recordings that are no longer commercially available.”).
19 See Paul Goldstein, Copyright’s Highway: From Gutenberg to the Celestial Jukebox

(New York: Hill and Wang, 1994). There are two parts to this conception, of course. One
is the “celestial” part—emphasizing universal access. The other is the “jukebox” part—
emphasizing payment. Napster emphasized the first.
20 See Scott Kirsner, “Firefly: From the Media Lab to Microsoft,” Wired News, April 9,

1998, at http://www.wired.com/news/business/0,1367,11585,00.html; Daniel Lyons,
“The Buzz About Firefly,” New York Times, June 29, 1997, section 6, 37. See also An-
drew L. Shapiro, The Control Revolution: How the Internet Is Putting Individuals in
Charge and Changing the World We Know (New York: PublicAffairs, 1999), 84–101.
21 See Laura J. Gurak, Persuasion and Privacy in Cyberspace: The Online Protests Over

Lotus MarketPlace and the Clipper Chip (New Haven, Conn., and London: Yale Uni-
versity Press, 1997); Seth Safier, “Between Big Brother and the Bottom Line: Privacy in
Cyberspace,” Virginia Journal of Law & Technology 5 (2000): 6; Andrew Shapiro, “Pri-
vacy for Sale: Peddling Data on the Internet,” Nation (June 23, 1997). The FTC has
taken an active role in monitoring the monitors. See “United States Federal Trade
Commission, Privacy Online: Fair Information Practices in the Electronic Marketplace:
A Federal Trade Commission Report to Congress” (Washington, D.C.: Federal Trade
Commission, May 2000), available at http://www.ftc.gov/reports/privacy2000/privacy2000.
pdf. For a list of current legislation proposed, see current information on the status of
pending privacy bills, available at http://www.epic.org/privacy/bill_track.html.
22 This is the argument made by Cass Sunstein, in Republic.com (Princeton, N.J.:

Princeton University Press, 2001). As Sunstein argues, how groups are structured—what
their composition is, how they deliberate—affects the results that deliberation produces.
Cass Sunstein, Republic.com, 65–71.
23 The success rate of advertising is highly controversial. The general consensus is that

direct snail mail advertising response rates are generally in the 1–3 percent range. Di-
rected e-mail advertising campaigns may have response rates in the 10–15 percent range,
though some estimates run as high as 25 percent. The click-through rate for banner ads
on the Web is much lower, estimated at 0.5 percent. Mark Brownlow for Internet Busi-
ness Forum, available at www.ibizbasics.com/online040301.htm.
24 A “lower cost,” of course, does not mean no cost. There is still the cost of publishing

a book and at least some cost in an initial promotion.
25 For a summary of peer-to-peer standards in progress, see http://peer-to-peerwg.org.

See also http://p2ptracker.com (summarizing current technology); “Business, Band-
width May Dash Hopes of a Peer-to-Peer Utopia,” http://news.cnet.com/news/0-1005-
201-3248711-0.html.
26 Clay Shirky, “Clay Shirky’s Writings About the Internet: Economics and Culture,

Media and Community, Open Source,” November 16, 2000, www.openp2p.com/pub/
a/p2p/2000/11/24/shirky1-whatisp2p.html.
27 See, e.g., Nelson Minar and Marc Hedlund, “A Network of Peers: Peer-to-Peer
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 304

3 0 4 N O T E S

Models Throughout the History of the Internet,” in Peer-to-Peer: Harnessing the Benefits
of a Disruptive Technology, Andy Oram, ed. (Beijing and Cambridge, Mass: O’Reilly,
2001), 3–15 (describing how the original Internet was “fundamentally designed as a
peer-to-peer system” but became increasingly client/server oriented over time owing to
Web browser applications, firewalls, and other factors).
28 For background on SETI, see “History of SETI,” at http://www.seti-inst.edu/

general/history.html; Eric Korpela et al., “SETI@home: Massively Distributed Comput-
ing for SETI,” at http://www.computer.org/cise/articles/seti.htm.
29 Howard Rheingold, “You Got the Power,” Wired (August 2001), at http://www.wired.

com/wired/archive/8.08/comcomp.html?pg=1&topic=&topic_set=.
30 For a useful survey of issues related to P2P, see Peer-to-Peer: Harnessing the Benefits

of a Disruptive Technology, Andrew Oram, ed. (Beijing and Cambridge, Mass.: O’Reilly,
2001). See also http://www.oreillynet.com/p2p/ (collecting articles). Xerox PARC has
conducted an interesting study of the free-riding problem with p2p technologies. See
Eytan Adar and Bernardo A. Huberman, “Free Riding on Gnutella,” First Monday, Oc-
tober 2000, 5 at http://www.firstmonday.dk/issues/issue5_10/adar/index.html. For a pes-
simistic view of the potential for P2P, see Lee Gomes, “P-to-P, B-to-B, RIP?,” Wall Street
Journal, April 4, 2001, B1.
31 Rheingold, You Got the Power.
32 Andy Oram, “The Value of Gnutella and Freenet,” Webreview (May 2000), at

http://www.webreview.com/pi/2000/05_12_00.shtml.
33 Again, this is just relative. The claim is not that there is equal access to equally valu-

able data. Amazon.com is in a better position to market than tinybookseller.com. Like-
wise, it may well be that scale here makes all the difference. There have been suggestions
that these architectures may reduce competition. See Yannis Bakos and Erik Brynjolfs-
son, “Bundling and Competition on the Internet,” Marketing Science, January 2000,
available at http://ecommerce.mit.edu/erik/bci-final.pdf (finding that “economies of
aggregation” for information goods could adversely affect competition). At this stage,
however, I don’t think we know enough to say.
For a careful analysis mapping the path to concentration in distribution from the na-
ture of IP rights, see Martin Kretschmer, George Michael Klimis, and Roger Wallis,
“The Changing Location of Intellectual Property Rights in Music: A Study of Music
Publishers, Collecting Societies and Media Conglomerates,” Prometheus 17 (1999):
163.
34 James Boyle, “A Politics of Intellectual Property: Environmentalism for the Net?,”

Duke Law Journal 47 (1997): 87. See also Carol Rose, “The Several Futures of Property:
Of Cyberspace and Folk Tales, Emission Trades and Ecosystems,” Minnesota Law Re-
view 83 (1998): 129.
35 See Carl Shapiro, “Will E-Commerce Erode Liberty?,” Harvard Business Review

(May–June 2000): 189 (book review of Lessig, Code and Other Laws of Cyberspace [New
York: Basic Books, 1999]).

CHAPTER 10
1 Peter W. Huber, Michael K. Kellogg, and John Thorne, Federal Telecommunications
Law, 2nd ed. (Gaithersburg, Md.: Aspen Law & Business, 1999), 164–65.
2 The regulations that effected this neutrality were many and are described compre-

hensively in Huber, Kellogg, and Thorne.
3 In Carterfone, the FCC required the Bell system to allow the connection of
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 305

N O T E S 3 0 5

customer-provided equipment as long as it did not harm the network. The device at issue
in the decision permitted communication between a mobile radio and the landline net-
work. The Bell system declined to permit the connection but could not demonstrate the
harm interconnection would create. The FCC ordered the carrier to permit the inter-
connection, and this requirement, in turn, spurred a great deal of innovation for the tele-
phone network. Ibid., 409.
4 My emphasis on neutrality does not deny the effect of other regulatory measures. For

example, the fact that ISPs enjoyed business rates for their usage—flat rate pricing,
rather than metered pricing—effected an important subsidy for Internet service. See
Peter W. Huber et al., Federal Telecommunication Law (Gaithersburg, Md: Aspen Busi-
ness & Law, 1999), §11.5, 1030.
5 On ATM circuits, see Douglas E. Comer, Computer Networks and Internets, 2nd ed.

(Upper Saddle River, N.J.: Prentice-Hall, 1999), 88, 184. On the link to quality of
service, see Paul A. David and Raymond Werle, The Evolution of Global Networks: Tech-
nical, Organizational and Cultural Dimensions (May 2000), 10.
6 Likewise, if there are vices in networks that are technically not end-to-end compli-

ant, then we might imagine rules that would balance those vices to better achieve the
value of end-to-end. For example, other networks that are not end-to-end compliant—
such as ATM networks or multiprotocol label switching (MPLS) networks—may have
an aim that respects the values that end-to-end promotes. The TCP/IP network, while
not “optimized” for any particular use, in effect is optimized for some uses rather than
others. It does well, for example, with applications that do not suffer from network la-
tency (e-mail), but not with applications that do suffer from network latency (voice-over
IP). These other networks would better enable these other applications, thus rendering
the network as a whole more neutral among applications. If the value of end-to-end in-
heres in the consequences of this neutrality, then a properly implemented mix might
achieve end-to-end values without every part of the network being end-to-end. I am
grateful to Tim Wu for making this point to me.
7 The Telecommunications Act of 1996 does not define broadband. It refers to broad-

band as a characteristic of “advanced telecommunications capability,” which is defined
as “high-speed, switched, broadband telecommunications capability that enables users
to originate and receive high-quality voice, data, graphics, and video telecommunica-
tions using any technology.” Telecommunications Act of 1996, Pub. L. No. 104–104,
§706 (c)(1), 110 Stat. 56 (1996). See also 47 U.S.C. §157 note (2001). The FCC filed its
Section 706 Report to Congress in 1999 and defined broadband as “the capability of sup-
porting, in both the provider-to-consumer (downstream) and the consumer-to-provider
(upstream) directions, a speed (in technical terms, “bandwidth”) in excess of 200 kilobits
per second (kbps) in the last mile.” 14 FCC Rcd. 2398 at 2406 ¶20 (1999). See also
http://www.fcc.gov/Bureaus/Cable/Reports/broadbandtoday.pdf.
8 For an excellent and comprehensive history of and background on broadband issues

related to cable, see Kim Maxwell, Residential Broadband: An Insider’s Guide to the Bat-
tle for the Last Mile (New York: John Wiley & Sons, 1999). See also Patrick R. Parsons
and Robert M. Frieden, The Cable and Satellite Television Industries (Boston: Allyn and
Bacon, 1998).
9 See Mark S. Nadel, “Cablespeech for Whom?,” Cardozo Arts & Entertainment Law

Journal 4 (1985): 51, 70 and n. 104.
10 See Patrick R. Parsons and Robert M. Frieden, The Cable and Satellite Television In-

dustries (Boston: Allyn and Bacon, 1998), chs. 2, 5; Robert W. Crandall and Harold
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 306

3 0 6 N O T E S

Furchtgott-Roth, Cable TV: Regulation or Competition? (Washington, D.C.: Brookings
Institution, 1996), 1–23; Porter Bibb, It Ain’t as Easy as It Looks: Ted Turner’s Amazing
Story (New York: Crown Publishers, 1993) (Ted Turner biography).
11 By 1987, of the 87.5 million homes with TVs, 44.1 million (50.4 percent) were

cable subscribers. Currently about 66 percent of TV households in the country use
cable. Parsons and Frieden, 3, 121–22.
12 For a summary, see Jerry Kang, Communications Law and Policy: Cases and Mate-

rials (Gaithersburg, Md.: Aspen Law & Business, 2001), 154–61.
13 Mark E. Laubach, David J. Farber, and Stephen D. Dukes, Delivering Internet Con-

nections over Cable: Breaking the Access Barrier (New York: John Wiley, 2001), 11–12.
For a primer on DOCSIS and its history, see http://www.cablemodem.com/
DOCSIS.pdf.
14 See, e.g., Federal Communications Commission, “In the Matter of Applications for

Consent to the Transfer of Control of Licenses and Section 214 Authorizations from
Tele-Communications, Inc., Transferor to AT&T Corp., Transferee, CS Docket No.
98–178,” February 18, 1999, ¶89, available at http://www.fcc.gov/Bureaus/Cable/
Orders/1999/fcc99024.txt (“According to AT&T-TCI, any equal access conditions such
as those advocated by opponents to the requested transfers will impose substantial in-
vestment costs and expenses on @Home, which will only delay and diminish its deploy-
ment of broadband services to residential customers.”).
15 See Thomas Starr, John M. Cioffi, and Peter Silverman, Understanding Digital

Subscriber Line Technology (Upper Saddle River, N.J.: Prentice-Hall, 1999) (citing
1976).
16 Comer, Computer Networks and Internets, 159.
17 Throughout this section, by “telephone companies” I am referring to the regional

Bell operating companies (the RBOCs). Non-RBOCs are not subject to the same obliga-
tions of open access under the statute. See AT&T Corp. v. City of Portland, 216 F.3d 871,
879 (9th Cir., 2000).
18 There are more traditional concerns as well. “Over the long term, the cable

providers’ tying strategy will thus undermine competitive investment in both the broad-
band transport and portal markets, insulating cable providers from conduit and content
competition, and ensuring that the delivery of Internet-based video by competing con-
duits does not erode cable providers’ monopoly power in the market for traditional video
programming.” Daniel L. Rubinfeld and Hal J. Singer, “Vertical Foreclosure in High
Technology Industries: A Case Study of the AOL Time Warner Merger” (Rubinfeld-
Singer White Paper), 10. Rubinfeld believes the vertical integration of cable will create
an incentive to pursue two foreclosure strategies: (1) conduit discrimination; and (2)
content discrimination. Ibid., 29–30.
19 See, for example, Cisco White Paper, “Controlling Your Network—A Must for

Cable Operators” (1999), 5, available at http://www.cptech.org/ecom/openaccess/cisco1.
html, describing tools to effect discrimination. As one research report summarizes the
problem:

The situation is analogous to a customer trying to drive to the bookstore of their choice
only to find that roadblocks have been established to channel customers to another book-
seller, to the exclusion of all other booksellers. Unlike the roadway analogy, in the online
world the end user may not be aware that the roadblocks have been placed and may have
their behavior influenced without even knowing that the ISP has limited their ability to
choose.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 307

N O T E S 3 0 7

AARP, “Tangled Web: The Internet and Broadband Open Access Policy,” January
2001, research.aarp.org, 13.
20 See, e.g., http://web.mit.edu/Saltzer/www/publications/openaccess.html. For Saltzer’s

model licenses for cable access, see http://web.mit.edu/Saltzer/www/publications/
clauses2.html.
21 See Brock Meeks, “Excite@Home Keeps a Video Collar,” ZDNet News, Novem-

ber 1, 1999, at http://www.zdnet.com/zdnn/stories/news/0,4586,2385059,00.html. See
also Harold Feld, “Whose Line Is It Anyway?: The First Amendment and Cable Open
Access,” CommLaw Conspectus 8 (2000): 23, 34; Mark Cooper, “Transforming the In-
formation Superhighway into a Private Toll Road,” Colorado Law Review 71 (2000):
1011, 1055. An industry trade journal notes that both Excite@Home and Road Runner
limit consumers to ten-minute streaming segments. “PC-TV Convergence Driving
Streaming Industry Growth,” Warren’s Cable Reg. Monitor, March 1, 1999, 1999 WL
6825624.
22 David Lieberman, “Media Giants’ Net Change: Major Companies Establish Strong

Foothold Online,” USA Today, December 14, 1999, B2. See also telephone interview
with David Isenberg, February 14, 2001 (“They couldn’t possibly open up their system
so that it would have the capability to do TV over IP. That would be killing themselves.”).
23 Bar et al., 32.
24 Ibid., 34.
25 Ibid., 35.
26 Ibid., 29.
27 François Bar has a similar assessment. See François Bar, “The Construction of

Marketplace Architecture,” Brookings & Internet Policy Institute 15 (forthcoming 2001),
(describing the decline of end-to-end).
28 This, of course, is assured only if there is no actual or effective tying between cable

products and other services that might effectively protect cable from meaningful com-
petition with DSL.
29 See Joanna Glasner, “DSL Rhymes with Hell,” Wired News, January 2001, available

at http://www.wired.com/news/business/0,1367,41433,00.html (DSL numbers); Roy
Mark, “U.S. Scores First Decline in Internet Subscribers,” dc.internet.com, May 2,
2001, available at: http://dc.internet.com/news/article/0,1934,2101_756771,00.html
(cable numbers).
30 See Morgan Stanley Dean Witter, The Broadband Report (May 2000). See also

David Lake, “Strike Up the Broadband,” Industry Standard, February 2, 2001, at
http://www.thestandard.com/article/0,1902,21892,00.html. (“In the fourth quarter of
2000, cable-modem providers watched their subscriber base increase 19 percent to 4.2
million users. Still, the adoption of DSL is happening at a faster pace. At the beginning of
last year, the number of cable-modem subscribers outnumbered DSL subscribers 5 to 1.
Now that ratio hovers at almost 2 to 1, in favor of cable-modem access. Nevertheless,
cable modems are expected to retain an edge. In 2003, 51 percent of broadband sub-
scribers will use cable modems, while DSL is expected to account for only 37 percent of
the high-speed market, according to Jupiter Research.”)
31 Laubach, Farber, and Dukes, 238.
32 The campaign was conceived of by Jan Brandt. “Since Brandt’s arrival at AOL

in 1993, membership has grown from 250,000 to 8 million [in 1997].” Upside maga-
zine said that “Brandt’s carpet-bombing techniques have redefined the use of direct
mail in the high-tech industry and pioneered the get-something-for-nothing marketing
coups copied by Netscape and other Internet underdogs to achieve brand-name recog-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 308

3 0 8 N O T E S

nition in no time flat.” http://media.aoltimewarner.com/media/cb_press_view.cfm?
release_num=147.
33 On the character of these virtual worlds, see Julian Dibbell, My Tiny Life: Crime

and Passion in a Virtual World (London: Fourth Estate, 1998). See also Lynn Cherny,
Conversation and Community: Chat in a Virtual World (Stanford, Calif.: CSLI Publica-
tions, 1999).
34 America Online Inc., “Open Access Comments of America Online, Inc.” before

the Department of Telecommunications and Information Services, San Francisco, Oc-
tober 27, 1999.
35 Comments of America Online, Inc., “In the Matter of Transfer of Control of FCC

Licenses of MediaOne Group, Inc. to AT&T Corporation,” Federal Communications
Commission, CS Docket No. 99–251, August 23, 1999. As it argued:
What this merger does offer, however, is the means for a newly “RBOC-icized” cable in-
dustry reinforced by interlocking ownership relationships to (1) prevent Internet-based
challenge to cable’s core video offerings; (2) leverage its control over essential video fa-
cilities into broadband Internet access services; (3) extend its control over cable Internet
access services into broadband cable Internet content; (4) seek to establish itself as the
“electronic national gateway” for the full and growing range of cable communications
services.

To avoid such detrimental results for consumers, the Commission can act to ensure that
broadband develops into a communications path that is as accessible and diverse as nar-
rowband. Just as the Commission has often acted to maintain the openness of other last-
mile infrastructure, here too it should adopt open cable Internet access as a competitive
safeguard—a check against cable’s extension of market power over facilities that were
first secured through government protection and now, in their broadband form, are
being leveraged into cable Internet markets. Affording high-speed Internet subscribers
with an effective means to obtain the full range of data, voice and video services available
in the marketplace, regardless of the transmission facility used, is a sound and vital
policy—both because of the immediate benefit for consumers and because of its longer-
range spur to broadband investment and deployment. Here, the Commission need do
no more than establish an obligation on the merged entity to provide non-affiliated ISPs
connectivity to the cable platform on rates, terms and conditions equal to those accorded
to affiliated service providers. (AOL, FCC, 4)
36 Comments of AT&T Wireless Services, Inc., “In the Matter of Inquiry Regarding
Software Defined Radios,” Federal Communications Commission, ET Docket No.
00–47, July 14, 2000, 15.
37 AT&T Canada Long Distance Services, “Comments of AT&T Canada Long Dis-

tance Services Company,” before the Canadian Radio-television and Telecommunica-
tions Commission, Telecom Public Notice CRTC 96–36: “Regulation of Certain
Telecommunications Service Offered by Broadcast Carriers,” February 4, 1997. For the
best analysis of this change in position, see the submission by Mark Cooper of the Con-
sumer Federation of America, Petition to Deny, in re “Application of America Online
and Time Warner for Transfers of Control,” before the FCC, CS 00–30 (April 26, 2000):
The dominant and vertically integrated position of cable broadcast carriers requires a
number of safeguards to protect against anti-competitive behaviour. These carriers have
considerable advantages in the market, particularly with respect to their ability to make
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 309

N O T E S 3 0 9

use of their underlying network facilities for the delivery of new services. To grant these
carriers unconditional forbearance would provide them with the opportunity to leverage
their existing networks to the detriment of other potential service providers. In particular,
unconditional forbearance of the broadband access services provided by cable broadcast
carriers would create both the incentive and opportunity for these carriers to lessen com-
petition and choice in the provision of broadband service that could be made available
to the end customer. Safeguards such as rate regulation for broadband access services
will be necessary to prevent instances of below cost and/or excessive pricing, at least in
the near term.
Telephone companies also have sources of market power that warrant maintain-
ing safeguards against anti-competitive behaviour. For example, telephone companies
are still overwhelmingly dominant in the local telephony market, and until this domi-
nance is diminished, it would not be appropriate to forbear unconditionally from rate
regulation of broadband access services. (AT&T, p. 15)
In the opinion of AT&T Canada LDS, both the cable companies and the tele-
phone companies have the incentive and opportunity to engage in these types of anti-
competitive activities as a result of their vertically integrated structures. For example,
cable companies, as the dominant provider of broadband distribution services, would be
in a position to engage in above cost pricing in uncontested markets, unless effective
constraints are put in place. On the other hand, the telephone company will likely be the
new entrant in broadband access services in most areas, and therefore expected to price
at or below the level of cable companies. While this provides some assurances that tele-
phone companies are unlikely to engage in excessive pricing, it does not address the in-
centive and opportunity to price below cost. Accordingly, floor-pricing tests would be
appropriate for services of both cable and telephone companies. (AT&T, 16–17)
Furthermore, in the case of both cable and telephone broadcast carriers, safeguards
would also need to be established to prevent other forms of discriminatory behaviour and
to ensure that broadband access services are unbundled. (AT&T, 17)
38 See, e.g., Lisa Bowman, “Will Merger Shut Lid on Open Access?,” ZDNet
News, January 11, 2000, available at http://www.zdnet.com/zdnn/stories/news/
0,4586,2420130,00.html.
39 My point is not that this is the only threat. For example, as Denise Caruso has ar-

gued, the merger of Internet backbone providers might lead to a situation where peering
on the Internet (exchanging data between peers neutrally) will cease. In “Mergers
Threaten Internet’s Informal System of Data Exchange,” New York Times, February 14,
2000, available at http://www.nytimes.com/library/tech/00/02/biztech/articles/14digi.
html, she writes, “In the early days of the Internet, self-interest forced backbone
providers into peering [the free sharing of data between service providers]. . . . But it is
scarcely true today. . . . [U]pon completion of the Worldcom-Sprint merger, a single
company would control nearly half of the Internet’s backbone—making it, literally and
figuratively, without peer. Given the furious pace and high stakes of the telecommuni-
cations industry today, some fear that it is only a matter of time before one big backbone
provider or another refuses to exchange data traffic with one of its peers. What happens
then? ‘Well, they would have a legitimate excuse,’ says Hal Varian, dean of the school of
information management at the University of California at Berkeley. ‘An ISP could com-
plain, and rightly so, that another ISP was sending them huge amounts of traffic and
putting a load on their system. . . . That’s an excuse to say, ‘We can’t handle this guy’s
packets; we aren’t going to connect with him.’ ” Ibid.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 310

3 1 0 N O T E S

40 Timothy F. Bresnahan and Garth Saloner, “Large Firms’ Demand for Computer

Products and Services: Competing Market Models, Inertia, and Enabling Strategic
Change,” October 1994 (Research Paper No. 1318 in the Stanford Graduate School of
Business Jackson Library).
41 National Research Council, The Internet’s Coming of Age (Washington, D.C.: Na-

tional Academy Press, 2000), chapter 3, 24.
42 Tim Berners-Lee, Weaving the Web: The Original Design and Ultimate Destiny of

the World Wide Web by Its Inventor (San Francisco: HarperSanFrancisco, 1999), 130.
43 See Daniel L. Rubinfeld and Hal J. Singer, “Open Access to Broadband Networks:

A Case Study of the AOL/Time Warner Merger,” Berkeley Technology Law Journal 16
(2001): 631, 672 (“Our analysis has shown that a policy of . . . conduit discrimination
may be profitable post acquisition [and] that content discrimination is likely to be prof-
itable post-acquisition.”).
44 See Seth Schiesel, “Cable Giants Block Rival Ads in Battle for Internet Customers,”

New York Times, June 8, 2001, A1.
Denise Caruso raised concerns about the free speech aspects of this change eighteen
months earlier. As she argued, there is an increasing possibility that most of the Internet’s
content will be controlled by a few—or perhaps even one—large corporations, raising
some very troubling issues. “The reasons for urgency are twofold. First is the issue of how
to open privately owned broadband Internet access to all comers. In addition, the free-
speech issue arises when any single entity, of any size, controls both a transmission me-
dium and the information that flows over it. Open access is a particular concern. . . .
Powerful corporations like AT&T and the proposed AOL Time Warner would have the
power to balkanize the broadband Internet for their own purposes, with no legal reason
to open their networks to competitors.” “Convergence Raises Concerns About Access,”
New York Times, January 31, 2000, available at http://www.nytimes.com/library/tech/00/
01/biztech/articles/31digi.html.
45 See Federal Trade Commission, “FTC Approves AOL/Time Warner Merger with

Conditions,” December 14, 2000, http://www.ftc.gov/opa/2000/12/aol.htm; Federal
Trade Commission, Docket No. C-3989. See also John R. Wilke, “AOL and Time-
Warner Pledge Cable Access to Ease FTC Fears,” Wall Street Journal, December 14,
2000.
46 Or alternatively, a tragedy of an anticommons. The opportunity for any number of

players to interfere with open access to the network could be viewed as an anticommons.
See Michael A. Heller, “The Tragedy of the Anticommons,” Harvard Law Review 111
(1998): 621; James M. Buchanan and Yong J. Yoon, “Symmetric Tragedies: Commons
and Anticommons,” Journal of Law & Economics 43 (2000): 1.
47 This is not quite the tragedy that Hardin is describing. There is no common physi-

cal resource that is being overused. But there is a common virtual resource—the oppor-
tunity to innovate freely on the network—that is being misused by restricting the scope
of that innovation with respect to one part of the network. That produces an externality,
even if it doesn’t “overuse” a resource.
48 See Brief of Amici Curiae, Reed Elsevier Inc. et al., 7–9, eBay Inc. v. Bidder’s Edge

Inc., 100 F. Supp. 2d 1058 (N.D. Cal. 2000) (No. C-99–21200RMW).
49 See Chip Bayers, “The Bot.com Future,” Wired (March 2000): 210.
50 The total number of addresses available under Ipv6 is 1038. That is a huge number:

“If the address space of IPv4 is compared to 1 millimeter, the address space of IPv6
would be 80 times the diameter of the galactic system.” See http://www.wide.ad.jp/wg/
iPv6/index.html.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 311

N O T E S 3 1 1

51 By forcing more traffic over a single port, this arms race interferes with the oppor-

tunity to optimize traffic based on ports.
52 The functioning of ports is explained in Craig Hunt, TCP/IP Network Administra-

tion, 2nd ed. (Sebastopol, Calif.: O’Reilly, 1998), 42–47. IPSec is described in a series of
RFCs, which are available at http://www.ietf.org/html.charters/ipsec-charter.html. I am
not claiming IPSec would necessarily be consistent with end-to-end, but simply that it
provides a consistent protocol that could be implemented consistent with end-to-end.
53 Garrett Hardin, “The Tragedy of the Commons,” Science 162 (1968): 1243, 1244

(emphasis added).
54 National Research Council, 24–25.
55 As Faulhaber put it, “[T]he third view is that, as Adam Smith pointed out, produc-

ers are always conspiring about how to fleece the unsuspecting consuming public; it is
only the competitive market that keeps their avarice in check. But producers will always
search for ways to escape competition, through marketing, customer lock-in, predatory
pricing, network effects, etc. AND . . . technological innovation is just a part of this
strategic quest for greater profits, perhaps at the consumers’ expense.” “Faulhaber Com-
ments at E-2-E Workshop,” http://www.law.stanford.edu/e2e/papers.html. François Bar
has made a similar point: “If the Internet could reduce friction, the same technology can
also be deployed to create more of it.” François Bar, “The Construction of Marketplace
Architecture,” 5.
56 Charles Platt, “The Future Will Be Fast but Not Free,” Wired, May 2001, available

at http://www.wired.com/wired/archive/9.05/broadband_pr.html (emphasis added).

CHAPTER 11
1 See 47 U.S.C.§223 (Supp. 1996); ACLU v. Reno, 521 U.S. 844 (1997).
2 ACLU v. Reno, 217 F. 3d 162 (3d Cir., 2000).
3 Lawrence Lessig, Code and Other Laws of Cyberspace (New York: Basic Books,

1999), 225.
4 See 17 USCA §106 (2001): “[O]wner of copyright under this title has the exclu-

sive rights . . . (5) in the case of . . . pictorial, graphic, or sculptural works . . . to display
the copyrighted work publicly.” But the claim would be weak. Except in the most ex-
treme circumstances, the public display of a copyrighted work would be fair use. William
Carleton, “Copyright Royalties for Visual Artists: A Display-Based Alternative to the
Droit de Suite,” Cornell Law Review 76 (1991): 510, 525 (“The general principle that sec-
tion 109(c) observes, however, according to the House Report, is that ‘the lawful owner
of a copy of a work should be able to put his copy on public display without the consent
of the copyright owner’ ” [quoting 1976 U.S. Code Cong. & Admin. News 5659, 5693]).
See also Ringgold v. Black Entertainment Television, Inc., 126 F. 3d 70 (2nd Cir., 1997).
5 See Kevin V. Johnson, “Show’s Fan Sites Fight Off ‘Demon;’ Fox: Production Com-

pany Cites Its Copyrights,” USA Today, December 23, 1999, 4D, available at 1999 WL
6862067; and Aaron Barnhart and Kevin V. Johnson, “Twentieth, the Web Slayer: Stu-
dio Shifts Its Crusade to ‘Buffy’ Fans’ Web sites,” Electronic Media, December 6, 1999,
9, available at 1999 WL 8767348. Fox does not limit itself to the Web. See Twentieth
Century Fox Film Corp. v. 316 W. 49th Street Pub. Corp., No. 90 Civ. 6083 (MJL), 1990
WL 165680 (S.D.N.Y., October 23, 1990) (holding that a nightclub could not display
images of The Simpsons on its wall without Fox’s permission). Fox says that it “appreci-
ates” fan sites, but not enough to allow them to exist freely. Johnson, 4D. As Warner
Bros. Online president Jim Moloshok says, “We decided that we were going to create a
better experience for the fans.” Ibid.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 312

3 1 2 N O T E S

6 Ibid.
7 Fara Warner et al., “Holes in the Net . . . ,” Wall Street Journal, August 30, 1999, A1.
8 See http://www.olga.net/about.
9 See, e.g., Siva Vaidhyanathan, Copyrights and Copywrongs: The Rise of Intellectual

Property and How It Threatens Creativity (New York: New York University Press, 2001),
355 (“In July 1999, journalist Michael Colton posted an Internet parody of Talk maga-
zine, which is a partnership between Hearst Magazines and Walt Disney–owned Mira-
max Films. Miramax lawyers sent a cease-and-desist letter to Earthlink, the Internet
company that owned the server on which the parody sat. Earthlink immediately shut
down the parody. It only restored the site after Talk editor Tina Brown appealed to the
Miramax legal department to let the parody stand. Because of widespread misunder-
standing of copyright law, cease-and-desist letters carry inordinate cultural power and
can chill if not directly censor expression.”).
10 See Digital Millennium Copyright Act, 105 P. L. 304, Sec. 202(c)(1)(iii) (1998).
11 Lessig, 173–75.
12 See Bennett Haselton, “Amnesty Intercepted,” December 12, 2000, http://www.

peacefire.org/amnesty-intercepted.
13 See Microsystems Software, Inc. v. Scandinavia Online AB, 98 F. Supp. 2d 74

(D. Mass., 2000), aff’d, 226 F. 3d 35 (1st Cir., 2000) (enjoining “all persons in active con-
cert” with Jansson and Skala from “publishing the software source code and binaries
known as [CPHack]”).
14 See http://www.aclu.org/court/cyberpatrol_motion.html.
15 Sony Computer Entertainment, Inc. v. Connectix Corp, 203 F. 3d 596 (9th Cir.,

2000).
16 ProCD, Inc. v. Zeidenberg, 86 F. 3d 1447, 1454 (7th Cir., 1996).
17 See Paul Goldstein, “Copyright and the First Amendment,” Columbia Law Review

70 (1970): 983 (describing the ongoing potential of copyright’s grant of monopoly over
expression as censorship); Pamela Samuelson, “Reviving Zacchini: Analyzing First
Amendment Defenses in Right of Publicity and Copyright Cases,” Tulane Law Review
57 (1983): 836 (same).
18 For a detailed analysis of the DMCA, see David Nimmer, “A Riff on Fair Use in the

Digital Millennium Copyright Act,” University of Pennsylvania Law Review 148 (2000):
673 (discussing the formulation, adoption, and practical effect of the DMCA).
19 See Carolyn Andrepont, “Legislative Updates: Digital Millennium Copyright Act:

Copyright Protections for the Digital Age,” DePaul-LCA Journal of Art & Entertainment
Law 9 (1999): 397 (characterizing the DMCA as a necessary tool for protecting on-line
copyrighted material); and Michelle A. Ravn, Note, “Navigating Terra Incognita: Why
the Digital Millennium Copyright Act Was Needed to Chart the Course of Online Ser-
vice Provider Liability for Copyright Infringement,” Ohio State Law Journal 60 (1999):
755 (arguing that the DMCA was particularly needed to clarify liability for on-line copy-
right infringements).
20 Or at least not significantly. It is true that if you could play DVD movies on any ma-

chine, there would be more machines that might demand DVD content.
21 The MPAA filed suit against four Web site operators in the Southern District of

New York and in the District of Connecticut. The DVD Copy Control Association filed
suit in California State court against about twenty named defendants and five hundred
unnamed ones. For a history of these suits, see Openlaw/DVD: Resources, at
http://eon.law.harvard.edu/openlaw/DVD/resources.html (visited April 19, 2001).
22 Universal City Studios, Inc. v. Reimerdes, 82 F. Supp. 2d 211 (S.D.N.Y., 2000) and
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 313

N O T E S 3 1 3

DVD Copy Control Ass’n, Inc. v. McLaughlin, No. CV 786804, 2000 WL 48512 (Cal.
Superior, January 21, 2000).
23 See Michael A. Geist, “iCraveTV and the New Rules of Internet Broadcasting,”

University of Arkansas at Little Rock Law Review 23 (2000): 223; and John Borland, “On-
line TV Service May Spark New Net Battle,” CNET.com, at http://news.cnet.com/
news/0-1004-200-1477491.html (last visited April 4, 2001) (describing the launch of the
iCraveTV.com Web site). For a scholarly analysis of the case, see Howard P. Knopf,
“Copyright and the Internet in Canada and Beyond: Convergence, Vision and Divi-
sion,” European Intellectual Property Review (2000): 262.
24 China Online (visited April 17, 2001), http://www.chinaonline.com (a U.S.-based

English-language news site on China). Or alternatively, Human Rights in China (visited
April 17, 2001), http://www.hrichina.org (a New York–based Chinese/English-language
Web site chronicling Chinese human rights abuses).
25 See Association “Union des Étudiants Juifs de France”, la “Ligue contre le Racisme

et l’Antisémitisme” v. Yahoo! Inc. et Yahoo France, T.G.I. Paris, Ordonnance de réfèré
du 22 mai 2000, available at: http://www.legalis.net/jnet/decisions/responsabilite/
ord_tgi-paris_220500.htm.
26 See Jack L. Goldsmith and Alan O. Sykes, “The Internet and the Dormant Com-

merce Clause,” Yale Law Journal 110 (2001): 785. See also Lawrence Lessig and Paul
Resnick, “Zoning Speech on the Internet: A Legal and Technical Model,” Michigan
Law Review 98 (1999): 395
27 A very similar point is made by Denise Caruso. See “Case Illustrates Entertainment

Industry’s Copyright Power,” New York Times, March 13, 2000, available at http://www.
nytimes.com/library/tech/00/03/biztech/articles/13digi.html. As Caruso writes, “The
most chilling aspect of the case . . . was [iCraveTV’s] response. That is, [it] did not argue
the legality of the action against [it], but instead responded by inventing a technology
that could stop the discussion dead in its tracks.” Caruso continues: “Many people are
likely to object strongly to [the site’s] balkanized Internet . . . such a system would de-
volve the Internet into a model very much like the restricted, centralized control of cable
television. That is a business model with which the $65 billion media and entertainment
industries—the very ones that nearly sued the pants off [of the site]—are quite familiar.”
Ibid.
28 See Press Release, Recording Industry Association of America, RIAA Statement

Concerning MP3.Com’s New Services, January 21, 2000, at http://www.riaa.com/
PR_Story.cfm?id=47.
29 The court’s ruling in the case determines that the damages are $25,000 per CD.

This leads to possible damages of $118 million if the total is determined to be at least
4,700 CDs. UMG Recordings, Inc. v. MP3.Com, Inc., No. 00 CIV. 472(JSR), 2000 WL
1262568, at *6 (S.D.N.Y., 2000).
30 See, e.g., Amy Harmon, “Powerful Music Software Has Industry Worried,” New York

Times, March 7, 2000, available at http://www.nytimes.com/library/tech/00/03/biztech/
articles/07net.html; Karl Taro Greenfeld, “The Free Juke Box,” Time (March 27, 2000),
available at http://www.time.com/time/everyone/magazine/sidebar_napster.html; Andy
Oram, “Gnutella and Freenet Represent True Technological Innovation,” May 12, 2000,
at http://www.oreillynet.com/pub/a/network/2000/05/12/magazine/gnutella.html; also
Peer-to-Peer: Harnessing the Benefits of a Disruptive Technology, Andy Oram, ed. (Beijing
and Cambridge, Mass.: O’Reilly, 2000).
31 Home Recording of Copyrighted Works: Hearing on H.R. 4783, H.R. 4794, H.R.

4808, H.R. 5250, H.R. 5488, and H.R. 5750 Before the Subcomm. on Courts, Civil Lib-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 314

3 1 4 N O T E S

erties, and the Admin. of Justice of the Comm. on the Judiciary, 97th Cong. (2nd ses-
sion), 8 (1983) (testimony of Jack Valenti, president, Motion Picture Association of
America, Inc.). See also Sam Costello, “How VCRs May Help Napster’s Legal Fight,”
Industry Standard, July 25, 2000, available at http://www.thestandard.com/article/
0,1902,17095,00.html.
32 Universal City Studios, Inc. v. Sony Corp. of Am., 480 F. Supp. 429, 432 (C.D. Cal.,

1979) (district court opinion by Judge Ferguson).
33 Ibid., 432.
34 A&M Records, Inc. v. Napster, Inc. 239 F. 3d 1004 (9th Cir., 2001).
35 Harper & Row Publishers, Inc. v. Nation Enterprises, 471 U.S. 539, 588 (1985).
36 Joseph Story, “Commentaries on the Constitution of the United States,” R. Ro-

tunda and J. Nowak, eds. (1987), §502, 402.
37 See Testimony of Professor Peter Jaszi, “The Copyright Term Extension Act of

1995: Hearings on S.483 Before the Senate Judiciary Comm.,” 104th Cong. (1995),
available at 1995 WL 10524355, at *6.
38 Eldred v. Reno, 239 F. 3d 372, 375 (D.C. Cir., 2001).
39 Recording Industry Association of America, 2000 Year-end Statistics (2001), at

http://www.riaa.com/pdf/Year_End_2000.pdf.
40 Jeff Leeds, “Album Sales Test the Napster Effect,” Los Angeles Times, June 20,

2001, C1.
41 “This case has always been about sending a message to the technology and venture

capital communities that consumers, creators and innovators will best flourish when
copyright interests are respected.” Jim Hu and Evan Hansen, “Record Label Signs Deal
with Napster,” October 31, 2000, http://news.cnet.com/news/0–1005–200–3345604.
html. See also “Online Entertainment: Coming Soon to a Digital Device Near You:
Hearing Before the Senate Comm. on the Judiciary,” 107th Cong. (2001) (statement of
Hilary Rosen, president and CEO, Recording Industry Association of America), avail-
able at http://judiciary.senate.gov/te040301hr.htm; Press Release, Record Industry Asso-
ciation of America, Hilary Rosen Press Conference Statement, February 12, 2001,
available at http://www.riaa.com/News_Story.cfm?id=371.
42 As Yochai Benkler writes, “[I]ncreases in intellectual property rights are likely to

lead, over time, to concentration of a greater portion of the information production func-
tion in the hands of large commercial organizations that vertically integrate new pro-
duction with inventory management.” Yochai Benkler, “The Commons as a Neglected
Factor of Information Policy,” October 3–5, 1998, 74.
Likely, and we might add, have. Compare, as David Isenberg points out, the connec-
tion to the history of AT&T: “You can see now in the record industry, for example, that
the record companies are unwilling to give up this idea of the control of the physical
medium even though they could perhaps do a very good job of artist development.” Tele-
phone interview with David Isenberg, February 14, 2001.
43 Telephone interview with Michael Robertson, November 16, 2000.
44 Ibid.
45 See also Richard Watt, Copyright and Economic Theory: Friends or Foes? (Chel-

tenham, England, and Northampton, Mass.: E. Elgar, 2000), 161–200 (describing role
of cooperatives and collectives). For a more comprehensive analysis of compulsory
rights, see generally Marshall Leaffer, Understanding Copyright Law, 3rd ed. (New York:
M. Bender, 1999), 69–71 (detailing both support for and criticism of compulsory li-
censes), cited in Michael Freno, note, “Database Protection: Resolving the U.S. Data-
base Dilemma with an Eye Toward International Protection” 34 (2000), Cornell
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 315

N O T E S 3 1 5

International Law Journal 34: 165, 209 (promoting compulsory licenses in the context of
databases); Christopher Scott Harrison, comment, “Protection of Pharmaceuticals as
Foreign Policy: The Canada-U.S. Trade Agreement and Bill C-22 Versus the North
American Free Trade Agreement and Bill C-91” (2001), North Carolina Journal of Inter-
national Law and Communications Regulation 26: 457, 525 (advocating generally the
free distribution of compulsory licenses); Anthony Reese, “Copyright and Internet
Music Transmissions: Existing Law, Major Controversies, Possible Solutions” (2001),
University of Miami Law Review 55: 237, 270 (arguing for the extension of compulsory
licenses in the area of music on the Internet); Bess-Carolina Dolmo, note, “Examining
Global Access to Essential Pharmaceuticals in the Face of Patent Protection Rights: The
South African Experience” (2001), Buffalo Human Rights Law Review 7: 137 (explain-
ing the benefits of compulsory licenses for developing countries); Sheldon W. Halpern,
“The Digital Threat to the Normative Role of Copyright Law” (2001), Ohio State Law
Journal 62: 569, 593 (supporting compulsory licensing for digital images); Laura N. Ga-
saway, “Impasse: Distance Learning and Copyright” (2001), Ohio State Law Journal 62:
783 (questioning the practicality of compulsory licenses); Robert P. Merges, “One Hun-
dred Years of Solicitude: Intellectual Property Law, 1900–2000,” (2000), California Law
Review 88: 2187, 2194, n. 15 (criticizing compulsory licenses in one context); Robert P.
Merges, “Contracting into Liability Rules: Intellectual Property Rights and Collective
Rights Organizations,” California Law Review 84: 1293, (arguing against compulsory li-
censes for digital content); Ralph Oman, “The Compulsory License Redux: Will It Sur-
vive in a Changing Marketplace?” (1986), Cardozo Arts & Entertainment Law Journal 5:
37, 48 (noting that many actors in the area of intellectual property prefer private solu-
tions over compulsory licenses); Scott L. Bach, note, “Music Recording, Publishing, and
Compulsory Licenses: Toward a Consistent Copyright Law,” Hofstra Law Review 14:
379, 398–401 (arguing that compulsory licenses are unfair to many artists, discourage in-
novation, and may be unconstitutional under the copyright clause).
46 For a similar argument, see Raymond Shih Ray Ku, “Copyright & Cyberspace:

Napster and the New Economics of Digital Technology” (draft on file with author,
April 7, 2001) (“[C]yberspace and the economics of digital technology require the un-
bundling of the public’s interests in the creation and distribution of digital works.”).
47 Christopher Stern, “Napster Copyright Fight Goes to Hill,” Washington Post,

April 4, 2001, E03.
48 Vaidhyanathan, 14.
49 Richard A. Posner, Law and Literature, rev. and enlarged ed. (Cambridge, Mass.: Har-

vard University Press, 1998), 392.
50 Ibid., 396. See also William M. Landes and Richard A. Posner, “An Economic

Analysis of Copyright Law,” Journal of Legal Studies 18 (1989): 325.
51 For the history of Judge Kozinski, see Susan Rice’s “Profile,” in Los Angeles Daily

Journal, September 29, 1988, 1. See also History of the Federal Judiciary (Washington,
D.C.: Federal Judicial Center, 2001) (bio of Alex Kozinski from Federal Judges Bio-
graphical Database), at http://air.fjc.gov/history/judges_frm.html.
52 Vanna White v. Samsung Elecs. Am., Inc.; David Deutsch Assocs., 989 F. 2d 1512,

1514 (1993).
53 Ibid., 34.
54 Ibid., 27.
55 Ibid.
56 Ibid.
57 Ibid., 31.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 316

3 1 6 N O T E S

58 Ibid., 27.
59 Ibid., 30.
60 I do not address the reach of trademark law in this area, but it would only strengthen

the argument I am making. Unlike traditional intellectual property, trademarks are per-
petual, and their effective power has expanded dramatically. This has become especially
significant as the domain name system has had to deal with the conflict between trade-
marks and domain names. This has tempted the World Intellectual Property Association
to build control for trademark interests into the very architecture of the network. See
¶¶23–28, “Executive Summary of the Interim Report of the Second WIPO Internet Do-
main Name Process,” available at http://wipo2.wipo.int.
61 The origin of modern economic work here is Kenneth Arrow’s “Economic Welfare

and the Allocation of Resources for Invention,” in National Bureau Committee for Eco-
nomic Research, The Rate and Direction of Inventive Activity, Economic and Social Fac-
tors, Richard Nelson, ed. (Princeton, N.J.: Princeton University Press, 1962), 609.
Harold Demsetz responded to this by arguing in favor of a stronger property-based
regime. See Harold Demsetz, “Information and Efficiency: Another Viewpoint,” Journal
of Law & Economics 12 (1969): 1. On the question of optimal protection, see Richard
Gilbert and Carl Shapiro, “Optimal Patent Length and Breadth,” RAND Journal of Eco-
nomics 21 (1990): 106.
62 For a careful account of the Framers’ view of the patent power, see Edward C. Wal-

terscheid, “Patents and the Jeffersonian Mythology,” John Marshall Law Review 29
(1995): 269. Professor Pollack makes a persuasive argument that the conception is lim-
ited by Lockean conceptions of the property right. See Malla Pollack, “The Owned Pub-
lic Domain: The Constitutional Right Not to Be Excluded—or the Supreme Court
Chose the Right Breakfast Cereal in Kellogg v. National Biscuit Co.,” Hastings Commu-
nications & Entertainment Law Journal 22 (2000): 265. For an introduction to the ratio-
nale, see Richard A. Posner, Economic Analysis of Law, 4th ed. (Boston: Little, Brown,
1992), 38–41.
63 Steven Shavell and Tanguy van Ypersele, “Rewards Versus Intellectual Property

Rights” (NBER Working Paper No. 6956, February 1999), 27. Shavell and Ypersele sug-
gest a reward system to replace a patent system. A similar proposal has been made by Mi-
chael Kremer, “Patent Buy-Outs: A Mechanism for Encouraging Innovation” (NBER
Working Paper No. 6304, December 1997). Chicago professor Douglas Lichtman has a
related proposal to subsidize access to patented drugs. See Douglas Lichtman, “Pricing
Prozac: Why the Government Should Subsidize the Purchase of Patented Pharmaceu-
ticals,” Harvard Journal of Law & Technology 11 (1997): 123. For criticism of the reward
alternative, see F. Scott Kieff, “Property Rights and Property Rules for Commercializing
Inventions,” Minnesota Law Review 85 (2001): 697, 709–21.
64 Though the economic argument about the effect of patents on innovation remain

ambiguous at best. See, e.g., Roberto Mazzoleni and Richard R. Nelson, “Economic
Theories About the Benefits and Costs of Patents,” Journal of Economic Issues 32 (De-
cember 1998): 1031.
65 Adam B. Jaffe, “The U.S. Patent System in Transition: Policy Innovation and the

Innovation Process” (NBER Working Paper Series, August 1999), 24–25.
66 Ibid., 26.
67 Ibid.
68 A similar skepticism has been raised about strong property rights where network ex-

ternalities are strong. See Pamela Samuelson et al., “Manifesto Concerning the Legal
Protection of Computer Programs,” Columbia Law Review 94 (1994): 2308, 2375, citing
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 317

N O T E S 3 1 7

Joseph Farrell, “Standardization and Intellectual Property,” Jurimetrics Journal 30
(1989): 35, 36–38, 45–46 (discussing network effects).
69 See, e.g., Josh Lerner, “150 Years of Patent Protection” (NBER Working Paper

No. 7477, January 2000) (examines 177 policy changes in the strength of protection
across sixty countries, over a 150-year period, and concludes that strengthening patent
protection had few positive effects on patent applications in the country making
the policy change); Mariko Sakakibara and Lee Branstetter, “Do Stronger Patents In-
duce More Innovation? Evidence from the 1988 Japanese Patent Law Reforms,” Rand
Journal of Economics 32 (2001): 77 (authors find no evidence of an increase in R&D
spending or innovative output). For a long-standing source of skepticism about the effect
of strong patents on innovation, see Robert P. Merges and Richard R. Nelson, “On the
Complex Economics of Patent Scope,” Columbia Law Review 90 (1990): 839 (strong
patent assertions in electrical lighting, automobiles, airplanes, radio slowed down inno-
vation).
70 See, e.g., Jean O. Lanjouw, “The Introduction of Pharmaceutical Product Patents

in India: ‘Heartless Exploitation of the Poor and Suffering’?” (NBER Working Paper No.
6366, 1998); F. Scott Kieff, “Property Rights and Property Rules for Commercializing
Inventions,” Minnesota Law Review 67 (2001): 727–28. For an excellent and compre-
hensive account of the actual patenting practice, see John R. Allison and Mark A. Lem-
ley, “Who’s Patenting What? An Empirical Exploration of Patent Prosecution,”
Vanderbilt Law Review 53 (2000): 2099, 2146; John R. Allison and Mark A. Lemley,
“How Federal Circuit Judges Vote in Patent Validity Cases,” Florida State University Law
Review 27 (2000): 745, 765 (concluding no systematic bias in judges’ votes).
71 Jaffe, 46.
72 Ibid., 47. Jaffe’s argument here is narrower than the point I am making in this sec-

tion. His concern is the social costs from too much effort being devoted to the pursuit of
patented innovation. My concern is the cost of patents on the innovation process gener-
ally.
73 “Patently Absurd?” The Economist (June 21, 2001). As the article goes on to report,

interviewees from patenting firms indicated that “rather than patenting to win exclusive
rights to a valuable new technology, patents were filed more for strategic purposes.”
74 Benjamin Franklin, “having no desire of profiting by patents myself,” cited “a prin-

ciple which has ever weighed with me . . . [t]hat, as we enjoy great advantages from the
inventions of others, we should be glad of an opportunity to serve others by any inven-
tion of ours; and this we should do freely and generously.” Benjamin Franklin, The Au-
tobiography of Benjamin Franklin, Frank Woodworth Pine, ed. (Garden City, N.Y.:
Garden City Publishing Co., 1916 ed.; originally published 1793), 215–16.
75 George Washington Carver, for example, observed about his inventions that “God

gave them to me, how can I sell them to someone else?” “Inventors,” Atlanta Journal &
Constitution, February 12, 1999, A17.
76 Robert K. Merton, “A Note on Science and Democracy,” Journal of Law & Political

Sociology 1 (1942): 115, 123. (“Patents proclaim exclusive rights of use and, often,
nonuse. The suppression of invention denies the rationale of scientific production and
diffusion.”) As Professor Arti Rai describes it, “[T]raditional scientific norms promote a
public domain of freely available scientific information, independent choice in the se-
lection of research topics, and (perhaps above all) respect for uninhibited scientific in-
vention.” Arti Kaur Rai, “Regulating Scientific Research: Intellectual Property Rights
and the Norms of Science,” Northwestern University Law Review 94 (1999): 77, 89–90.
More famously, Merton is known for characterizing “science” as “communistic”:
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 318

3 1 8 N O T E S

“Communism,” in the nontechnical and extended sense of common ownership of
goods, is a second integral element of the scientific ethos. The substantive findings of sci-
ence are a product of social collaboration and are assigned to the community. They con-
stitute a common heritage in which the equity of the individual producer is severely
limited. An eponymous law or theory does not enter into the exclusive possession of the
discoverer and his heirs, nor do the mores bestow upon them special rights of use and
disposition. Property rights in science are whittled down to a bare minimum by the ra-
tionale of the scientific ethic. The scientist’s claim to “his” intellectual “property” is lim-
ited to that of recognition and esteem which, if the institution functions with a modicum
of efficiency, is roughly commensurate with the significance of the increments brought
to the common fund of knowledge. Eponymy—for example, the Copernican system,
Boyle’s law—is thus at once a mnemonic and a commemorative device.
Merton, 121.
77 Fred Warshofsky, The Patent Wars (New York: Wiley, 1994), 170. Of course, I don’t

mean to suggest that Microsoft is against software patents. Indeed, in the same memo,
Gates goes on to recommend the Microsoft strategy to respond to this new world of
patents:
The solution . . . is patent exchanges . . . and patenting as much as we can. . . . A future
start-up with no patents of its own will be forced to pay whatever price the giants choose
to impose. That price might be high: Established companies have an interest in excluding
future competitors.
Ibid., at 170–71 (emphasis added).
78 Thomas Jefferson, Letter to James Madison, in Julian P. Boyd, ed., The Papers of

Thomas Jefferson 13 (Princeton, N.J.: Princeton University Press, 1956), 440, 442–43.
79 To qualify for patent protection, an invention must be novel, 35 U.S.C. §§101,

102 (1982); it must provide utility, ibid.; and it must be nonobvious, ibid., §103. The
invention must also be within the list of patentable subject matter, 35 U.S.C. §101
(1982).
80 Jaffe, 9. The changes have also coincided with an increase in the number of patents.

On one account, that increase may be because patents are indeed a spur to inno-
vation. Researchers, however, have concluded differently. Samuel Kortum and Josh
Lerner, “Stronger Protection or Technological Revolution: What Is Behind the Recent
Surge in Patenting?,” Carnegie-Rochester Conference Series on Public Policy 48 (1998):
247 (attributing the increase to improvements in the management of research); Bron-
wyn H. Hall and Rosemarie Ham Ziedonis, “The Patent Paradox Revisited: An Empiri-
cal Study of Patenting in the U.S. Semiconductor Industry, 1979–1995,” Rand Journal of
Economics 32 (2001): 101 (attributing the increase to portfolio races).
81 E-mail from Greg Aharonian, author, Internet Patent News Service, May 28, 2001,

on file with author.
Greg Aharonian is perhaps the leading expert on the practice of software and Internet-
related patents. While he is a supporter of patents in principle, he has been a strong critic
of the U.S. Patent Office. Aharonian estimates the number of software patents in a num-
ber of ways. He provided the following data to me:
“TOTAL is the total number of patents issued that year, GREG is my count or esti-
mated count of software patents (using the Greg Aharonian scheme) issued in that year.
SOFTWARE is the number of patents in that year that include the word software some-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 319

N O T E S 3 1 9

where in the specification, claims, or abstract. 364&395 is the total number of patents is-
sued in these two main computing classes.”

YEAR TOTAL GREG SOFTWARE (364&395)
1999 154,534 21,000 (est.) 17,603 6,410
1998 150,961 17,500 (est.) 16,100 10,571
1997 124,181 13,000 (est.) 10,017 8,190
1996 121,799 9,000 9,104 7,922
1995 113,941 6,142 6,951 6,114
1994 113,706 4,569 6,009 5,745
1993 109,876 2,400 4,929 4,862
1992 107,489 1,624 4,068 4,073
1991 106,831 1,500 3,543 3,817
1990 99,210 1,300 3,046 3,606
1989 102,686 1,600 3,090 3,980
1988 84,433 800 2,053 2,708
1987 89,578 800 2,038 2,766
1986 77,039 600 1,483 2,202
1985 77,268 500 1,324 1,978
1984 72,668 400 1,003 1,857
1983 62,005 350 635 1,517
1982 63,291 300 603 1,446
1981 71,105 300 544 1,257
1980 66,206 250 465 1,239
1979 52,480 200 269 986
1978 70,564 150 299 1,272
1977 69,797 100 300 1,320
1976 70,924 100 208 1,113
1975 72,156 100 188 817
1974 62,342 100 188 838
1973 61,019 150 122 871
1972 58,603 200 110 906
1971 50,904 100 68 896
2,537,596 33,635 96,360 91,279

82 Seth Shulman, Owning the Future (Boston: Houghton Mifflin, 1999), 69.
83 Douglas Brotz made this statement at the Public Hearing on Use of the Patent Sys-
tem to Protect Software Related Inventions, January 26, 1994, at the San Jose Conven-
tion Center, transcript available at http://lpf.ai.mit.edu/Patents/testimony/statements/
adobe.testimony.html.
84 “Oracle Corporation opposes the patentability of software.” Statement available at

http://www.base.com/software-patents/statements/oracle.statement.html (visited June 8,
2001).
85 Patent law has long protected process patents. Mark A. Lemley, “Patent Scope and

Innovation in the Software Industry,” California Law Review 89 (2001): 1, 8.
86 Rochelle Cooper Dreyfuss, “State Street or Easy Street: Is Patenting Business Meth-

ods Good for Business?,” U.S. Intellectual Property: Law & Policy 6 (2000): 27.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 320

3 2 0 N O T E S

87 State St. Bank & Trust Co. v. Signature Fin. Group, Inc., 149 F. 3d 1368 (Fed. Cir.,

1998).
88 Carl Shapiro, “Navigating the Patent Thicket: Cross Licenses, Patent Pools, and

Standard-Setting” (working draft paper presented at National Bureau of Economic Re-
search annual conference, May 2000): 2.
89 See Laura Rohde, “BT Flexes Muscles Over U.S. Hyperlink Patent,” June 21, 2000,

available at http://www.idg.net/idgns/2000/06/21/BTFlexesMusclesOverUSHyperlink.
shtml. The Amazon patent has received increasingly skeptical review in the courts. Af-
ter a district court enjoined Barnes & Noble from using a similar technology, Amazon.
com v. Barnesandnoble.com, 73 F. Supp. 2d 1228 (W.D. Wash., 1999), the Court of
Appeals for the Federal Circuit reversed the injunction, finding that Barnesandnoble.
com had mounted a substantial challenge to the validity of the patent. Amazon.com v.
Barnesandnoble.com, 239 F. 3d 1343 (CAFC, 2001).
90 James Bessen and Eric Maskin, “Sequential Innovation, Patents, and Imitation,”

January 2000, available at http://researchoninnovation.org/patent.pdf (visited on June 10,
2001). See also Patently Absurd: “[T]he rush to acquire patent portfolios could slow
down the generation of new ideas.”
91 See Edmund W. Kitch, “The Nature and Function of the Patent System,” Journal

of Law & Economics 20 (1977): 265 (advocating prospect theory). See also Merges and
Nelson, 839 (propounding a race-to-invent theory; lethargy in patent development will
lead to a problem of “under-fishing”; thus, where innovation is cumulative, narrow
patents are better). See also Mark F. Grady and Jay I. Alexander, “Patent Law and Rent
Dissipation,” Virginia Law Review 78 (1992): 305 (arguing that patents reduce rent dis-
sipation from races to initial invention, races to improvements, and wasteful efforts to
keep secrets). For a critique of Grady and Alexander, see Robert P. Merges, “Rent Con-
trol in the Patent District: Observations on the Grady-Alexander Thesis,” Virginia Law
Review 78 (1992): 359.
92 As Professor Julie Cohen describes it:

Kitch based his “prospect” theory on an analogy to nineteenth-century mining claims,
which reserved for first-comers all rights to explore the described terrain. Under the
prospect theory of patent scope, issued patents would operate as broad reservations of
rights in the technical landscape. As a result, patentees could credibly seek to exact roy-
alties for nearly all improvements, whether literally infringing or not. Improvers, mean-
while, would need to think twice before refusing such demands. To a greater degree than
ever before, second-comers would need permission to develop and market their innova-
tions.

Julie E. Cohen and Mark A. Lemley, “Patent Scope and Innovation in the Software
Industry,” California Law Review 89 (2001): 1, 14–15.
93 Ibid., 15. (“It assumes that owners can readily identify, and would readily license,

successful improvers; that the gains from coordination would outweigh the costs of any
strategic behavior by owners and improvers; and that the initial allocation of stronger
property rights to the prospect owner would not adversely affect improvers’ incentives [or
that an overall increase in productivity would outweigh any such adverse effect].”)
94 Cf. Rai, 121 (criticizing patents because of the “losses in creativity and high trans-

action costs that such grants generate”).
95 This point is not specific to the Internet or to software technologies. Professor Arti

Rai has made a similar point in the context of basic scientific research:
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 321

N O T E S 3 2 1

Bargaining between the patent holder and improver would face a variety of obstacles.
First, there would be very substantial difficulties in valuation. It is by no means clear how
one goes about valuing an EST of unknown function, let alone uncertain but potentially
highly lucrative research using that EST. Disagreement about the value of the patented
invention relative to the value of the research project might make it very difficult for the
parties to agree on the terms for a license. This disagreement might well be exacerbated
by cognitive biases that could lead the EST patent owner to overvalue its asset. In par-
ticular, the owner might overestimate the possibility that the EST would lead to the find-
ing of a valuable drug therapy and then negotiate based on this irrational estimation.
Negotiation would also be hampered by Arrow’s information paradox: it might be im-
possible for the subsequent researcher to get a license ex ante without disclosing to the
patent holder valuable information about its own research plans. But because this re-
search plan would not be protectable as intellectual property, the competitor might fear
that the patent holder would appropriate the information for its own use, with no com-
pensating benefit to the competitor. Even if these difficulties did not lead to bargaining
breakdown, they would create transaction costs that reduced the cooperative surplus to
be gained from a license and would thus deter at least some inventors and improvers
from negotiating in the first instance. Transaction costs would be compounded by the
likelihood that the would-be follow-on improver would likely have to negotiate licenses
not simply with one owner of basic research but with many such owners. For example,
in order to develop a commercial treatment for a genetic disease (particularly a polygenic
disease), it may be necessary to have access to a large number of ESTs and SNPs, each
conceivably patented by a different entity. Similarly, in order to screen potential phar-
maceutical products for therapeutic effects and side-effects at the pre-clinical testing
stage, it may be very useful to have access to a large number of different receptors, each
potentially controlled by a different owner. Some of these difficulties might be addressed
if the bargaining occurred ex post, after the follow-on improver had already developed
the improvement. In that case, both the initial inventor and the follow-on improver
would have greater knowledge of the value of the improvement relative to that of the
original invention. Moreover, in the ex post situation, the possibility of “blocking
patents” might provide a way around Arrow’s information paradox. Blocking patents arise
when a follow-on improver takes a patented product and improves it in a nonobvious
way. Although the follow-on improver can then secure a patent on that improvement,
the improvement may nonetheless infringe the original patent. In the blocking patent
situation, neither the initial patent holder nor the follow-on improver can sell the im-
provement without cross-licensing. But, if the improvement is truly significant, such that
the initial inventor would want to sell the improvement itself, it would presumably have
a financial incentive to cross-license the improver.
Ibid., 125–26. Rebecca Eisenberg makes a similar point: “Michael Heller and I argue
that too many patent rights on ‘upstream’ discoveries can stifle ‘downstream’ research
and product development by increasing transaction costs and magnifying the risk of bar-
gaining failures.” Rebecca S. Eisenberg, “Bargaining over the Transfer of Proprietary Re-
search Tools,” 224. See also Robert P. Merges, “Institutions for Intellectual Property
Transactions: The Case of Patent Pools,” in Expanding the Boundaries of Intellectual
Property, Rochelle Cooper Dreyfuss and Diane Leenheer Zimmerman, eds. (Oxford:
Oxford University Press, 2001), 127–28.
96 Bessen and Maskin, “Sequential Innovation, Patents, and Imitation.”
97 Steven Levy, “The Great Amazon Patent Debate,” Newsweek (March 13, 2000): 74
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 322

3 2 2 N O T E S

(“I asked Bezos if Amazon would have developed 1-Click even if there were no patent
system to protect it and anyone could legally rip it off. ‘Yes,’ he responded without hesi-
tation. ‘Very definitely.’ ”). This point suggests a related reason to be skeptical about these
patents. Patents can induce overinvestment in patent protection; the low additional cost
to get the protection may induce too much patent duplication. See Posner, Economic
Analysis of Law, 39.
98 For skepticism about software patents, see, e.g., Pamela Samuelson, “Benson Revis-

ited: The Case Against Patent Protection for Algorithms and Other Computer Program–
Related Inventions,” Emory Law Journal 39 (1990): 1025; A. Samuel Oddi, “Beyond
Obviousness: Invention Protection in the Twenty-first Century,” American University
Law Review 38 (1989): 1097 (arguing that patents should be issued only for major inno-
vations); Pamela Samuelson et al., 2308. See also Cohen and Lemley (arguing for nar-
rowing to permit reverse engineering). On business method patents, see, e.g., “One-click
Monster,” American Lawyer (May 2000): 51; Rochelle Cooper Dreyfuss, “Are Business
Method Patents Bad for Business?,” Computer & High Technology Law Journal 16
(2000): 263 ; William Krause, “Sweeping the E-Commerce Patent Minefield: The Need
for a Workable Business Method Exception,” Seattle University Law Review 24 (2000):
79; Jared Earl Grusd, “Internet Business Methods: What Role Does and Should the Law
Play?,” Virginia Journal of Law & Technology 4 (1999): 9.
99 “In the United States, despite the long-standing controversy around software pa-

tents, there has been virtually no government effort to study the economic effects of
expanded patent protection. The one government-commissioned study of which I am
personally aware was suspended at the request of a multinational company with a unique
position in software patents.” Brian Kahin, comments in response to “The Patent-
ability of Computer-Implemented Inventions,” available at http://europa.eu.int/comm/
internal_market/en/intprop/indprop/maryland.pdf. The National Academies has launched
a study, “Intellectual Property Rights in the Knowledge-Based Economy,” http://www.
nationalacademies.org/ipr.
100 See “Internet Society Panel on Business Method Patents,” http://www/oreillynet.

com/lpt/a/434 (“Property, the first of the three debates I argued, I would argue is beyond
reproach and the burden of proof is not on those who would need to say property should
be but on those who say property should not be because historically societies that did not
respect property rights, all rights, ended up in the dust bin of history.” [Jay Walker]).
101 Friedrich A. von Hayek, “Free Enterprise and Competitive Order,” in Individual-

ism and Economic Order (Chicago: University of Chicago Press), 107, 114. Nobel
Prize–winning economist Milton Friedman expressed similar skepticism. “[T]rivial
patents . . . are often used as a device for maintaining private collusive arrangements that
would otherwise be more difficult or impossible to maintain.” Milton Friedman, Capi-
talism and Freedom (Chicago: University of Chicago Press, 1962), 127.
102 Lest too much of my own bias become apparent, I should note that there are some

who argue strongly (and in some ways convincingly) that between a well-functioning
copyright and well-functioning patent system, a patent system for software may well be
better. As Mark Haynes argues, “[U]nlike copyright, the patent system encourages im-
provement patents, through which competitors are able to neutralize the patent portfo-
lios of others.” Mark A. Haynes, commentary, “Black Holes of Innovation in the
Software Arts,” Berkeley Technology Law Journal 14 (1999): 567, 574. Likewise, as Mark
Lemley and David O’Brien have argued, patents would encourage the deployment of
reusable software “components.” Mark A. Lemley and David W. O’Brien, “Encouraging
Software Reuse,” Stanford Law Review 49 (1997): 255. Thus, conceivably, a patent sys-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 323

N O T E S 3 2 3

tem would encourage innovation. And Professor Scott Kieff has argued quite convinc-
ingly that patents spur not just innovation, but also commercialization. See F. Scott
Kieff, “Property Rights and Property Rules for Commercializing Inventions,” Minnesota
Law Review 85 (2001): 697. Commercialization, Kieff argues, depends upon property
rules protecting invention, not liability rules. Compare Ian Ayres and Paul Klemperer,
“Limiting Patentees’ Market Power Without Reducing Innovation Incentives: The Per-
verse Benefits of Uncertainty and Non-Injunctive Remedies,” Michigan Law Review 97
(1999): 985 (arguing for liability, rather than property protection, for patent rights).
I’m skeptical, but not a committed skeptic. These accounts don’t (and don’t purport
to) account for the full costs of a patent system, or the particular costs that would affect
software developers. Nor do they deny the burdens imposed by the current patent sys-
tem. Thus, while such a benefit is possible, that it is possible in the current regime has
not, in my view, been shown.
103 Telephone interview with Bob Young, November 14, 2000. See also Kahin, 3–4

(“At the level of individual patents, patents may benefit small firms more than large firms
because small-firm options for appropriating returns from innovation are fewer. For ex-
ample, small firms may be less able to exploit first-mover advantages. At the portfolio
level, large firms with large portfolios benefit disproportionately from network effects
and economies of scale and scope. This includes their ability to manage transaction
costs, which is the subject of the third perspective that I will explore at greater length.”);
Kahin, 4 (“Small firms [with some exceptions] are generally disadvantaged because they
lack in-house patent counsel and their business focus can be easily distracted by litiga-
tion or even claims of infringement.”).
104 “In the short run, individual patents work to [the benefit of small firms], while in

the long run and in the aggregate, patents favor large firms. The more pervasive patent-
ing becomes, the more the long-term, portfolio-level effects dominate.” Kahin, 4.
Patents have also been said to be a useful tool for old companies to protect themselves
against the new. As Gary Reback, Silicon Valley entrepreneur and attorney, describes the
history in the Valley, “[W]e went through a long period like that where I saw companies
like IBM going around Silicon Valley . . . leaning on the new generation of companies
like Sun . . . to develop a revenue stream out of their patent portfolio.” Telephone inter-
view with Gary Reback, November 21, 2000.
105 Tim Berners-Lee, Weaving the Web: The Original Design and Ultimate Destiny of

the World Wide Web by Its Inventor (San Francisco: HarperSanFrancisco, 1999), 196.
106 Richard Stallman, “The GNU Operating System and the Free Software Move-

ment,” in Open Sources: Voices from the Open Source Revolution, Chris DiBona, Sam
Ockman, and Mark Stone, eds. (Beijing and Sebastopol: O’Reilly, 1999), 53, 67.
107 Robert Young and Wendy Goldman Rohm, Under the Radar: How Red Hat

Changed the Software Business—and Took Microsoft by Surprise (Scottsdale, Ariz.: Cori-
olis Group Books, 1999), 135–36.
108 Peter Wayner, Free for All: How Linux and the Free Software Movement Undercut

the High-Tech Titans (New York: HarperBusiness, 2000), 223–24.
109 Kahin, 5.
110 Telephone interview with Bob Young.
111 Ian Mount, “Would You Buy a Patent License from This Man?,” eCompany, April

2001, available at http://www.ecompany.com/articles/mag/0,1640,9575,00.html.
112 Carl Shapiro, “Navigating the Patent Thicket: Cross Licenses, Patent Pools, and

Standard-Setting,” in Innovation Policy and the Economy, vol. 1, Adam Jaffe, Joshua
Lerner, and Scott Stern, eds. (Cambridge, Mass.: MIT Press, 2001), 8. Shapiro recom-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 324

3 2 4 N O T E S

mends that antitrust authorities permit packaging licensing for complementary, but not
substitute, patents, as a way to reduce the transaction costs associated with the hold-up
problem created by patents.
113 C. Shapiro, 7. A related argument is summarized by Denise Caruso. As she argues

in “Patent Absurdity,” New York Times, February 1, 1999, available at http://www.nytimes.
com/library/tech/99/02/biztech/articles/01digi.html, “Ideas are given their literal cur-
rency through patent and copyright laws, originally intended to stimulate innovation by
protecting inventors from idea snatchers and allowing them to profit more easily from
their talents. But some experts worry that an increasing number of individuals and com-
panies are perverting that original purpose with increasingly specious claims to owner-
ship, as well as by stockpiling patents into competitive arsenals.” Ibid. Caruso identifies
such things as “[a] patent for using a book as a teaching tool” and “a patent for down-
loading files over the Internet for a fee” as some of the more ridiculous recent develop-
ments in intellectual property law and adds that “various technology companies,
including IBM, Intel and Hewlett-Packard, use their vast holdings of patents as competi-
tive weaponry in seeking to disable each other with infringement charges. . . . Today’s re-
laxed rules for granting patents, and the greater ease with which arsenals can thus be
amassed, gives a decided battle advantage to industry heavyweights. . . . In today’s busi-
ness environment, in which a company’s market value is measured with increasing fre-
quency by the intellectual property it owns, arsenals of patents—specious or not—make
an unfortunate kind of sense.” Ibid.
114 Warshofsky, The Patent Wars, 170–71 (emphasis added).
115 Michael A. Heller, “The Tragedy of the Anticommons: Property in the Transition

from Marx to Markets,” Harvard Law Review 111 (1998): 621. The general issue of patents
in sequential, or cumulative, innovation is addressed by Suzanne Scotchmer, “Standing on
the Shoulders of Giants: Cumulative Research and the Patent Law,” Journal of Economic
Perspectives 5 (Winter 1991). James Bessen and Eric Maskin argue that this is precisely the
kind of innovation that is harmed the most by strong patent protection. See Bessen and
Maskin. See also Merges, Institutions for Intellectual Property Transactions, 125.
116 James M. Buchanan and Yong J. Yoon, “Symmetric Tragedies: Commons and

Anticommons,” Journal of Law & Economics 43 (2000): 1.
117 I’ve not discussed two other critical aspects of intellectual property law—either

trade secret law or trademark law. Both are relevant to the issues of control that I have de-
scribed, and trademark law in particular has expanded significantly. See, e.g., Alanna C.
Rutherford, “Sporty’s Farm v. Sportsman’s Market: A Case Study In Internet Regulation
Gone Awry,” Brooklyn Law Review 66 (2000): 421, 437 (describing the proliferation of
trademarks on the Net and its undermining of the original justification for trademark
law); Matthew A. Kaminer, “The Limitations of Trademark Law in Addressing Trade-
mark Keyword Banners,” Santa Clara Computer & High Technology Law Journal 16
(1999): 35, 61–62 (“We should not contain the growth of the Internet, but instead sup-
port it”) (quote on 62); Glenn A. Gunderson, “Expansion of Trademark Law Yields
Trickier Search: Development of Unusual Marks, Dilution Law and the Internet Com-
plicate Clearance Process,” National Law Journal, May 31, 1999, C9 (describing the
near doubling of trademark applications from 1990 to 1998), cited in Kathleen Dono-
hue, “Trademark Vigilance in the Twenty-first Century: A Pragmatic Approach,” Ford-
ham Intellectual Property Media & Entertainment Law Journal 9 (1999): 823, 828, n.18;
Claire Ann Koegler, “Here Come the Cybercops 2: Who Should Police Cybermarks?,”
Nova Law Review 22 (1998): 531, 532–33 (explaining how trademark law has greatly ex-
panded in the context of the Internet).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 325

N O T E S 3 2 5

CHAPTER 12
1 Thomas W. Hazlett, “Spectrum Flash Dance: Eli Noam’s Proposal for ‘Open Ac-
cess’ to Radio Waves,” Journal of Law & Economics 41 (1998): 805.
2 See Hazlett, “The Wireless Craze,” 17 (“Allocation and ‘technical’ rules protected

broadcasters from competition as well as from fees or competitive bidding”). The exam-
ples include AM radio’s protection from FM, ibid., 50–53; the artificial scarcity in
VHF-TV licenses, which produced just three national networks, ibid., 53–55; VHF-TV
blocking CATV, ibid., 55–62; AM and FM radio’s blocking of digital radio, ibid., 62–65;
NAB and NPR’s blocking of low-power FM radio, ibid., 68–82.
3 Eli Noam, “Beyond Spectrum Auctions: Taking the Next Step to Open Spectrum

Access,” Telecommunications Policy 21 (1997): 461, 462, n. 5.
4 Ibid. How this market system might work is not yet clear. Presumably, users would

bid in real time to get access to available spectrum. Bidding in turn would require that
users have some form of identification. The ID need not be a real identity. But every
transaction has a name. You generate less overhead validating every request if laying un-
known bids elsewhere is allowed. How those requests are handled would determine how
efficiently the system could work.
5 See, e.g., Yochai Benkler, “From Consumers to Users: Shifting the Deeper Struc-

tures of Regulation Toward Sustainable Commons and User Access,” Federal Communi-
cations Law Journal 52 (2000): 561; Eli Noam, “Spectrum Auctions: Yesterday’s Heresy,
Today’s Orthodoxy, Tomorrow’s Anachronism,” Journal of Law & Economics 41 (1998):
765; Yochai Benkler, “Overcoming Agoraphobia: Building the Commons of the Digi-
tally Networked Environment,” Harvard Journal of Law & Technology 11 (1997): 287.
6 See Hazlett, “Spectrum Flash Dance”; Gregory L. Rosston and Jeffrey S. Steinberg,

“Using Market-Based Spectrum Policy to Promote the Public Interest,” Federal Com-
munications Law Journal 50 (1996): 87.
7 Turner Broadcasting System, Inc. v. F.C.C., 520 U.S. 180, 189 (1997). I am assuming

(and I would argue) that ordinary First Amendment analysis should apply to the rules al-
locating spectrum. That assumption, however, is not obvious. You might take the view
that spectrum is just like paper (both are used to communicate), and there’s not a First
Amendment problem with the government’s nationalizing paper production. Cf. Ars-
berry v. Illinois, 244 F. 3d 558 (7th Cir., 2001) (examining tax as applied to the telephone
company).
But in my view, the appropriate analysis begins with the speaker, who would speak
using spectrum but for regulations by Congress of that spectrum. That is the same pos-
ture the Court has adopted when explaining why regulation of cable television is subject
to First Amendment analysis. In both cases, the issue is properly framed: What justifies
the state-imposed interference with the speaker’s ability to communicate? That question,
I expect, will be resolved through the same level of analysis as applied to cable. This is
because, as Benkler writes, “enclosure of information . . . affects different organizations
engaged in information production differently. This is so because information is not only
an output of information production, but also one of its most important inputs.” Yochai
Benkler, “The Commons as a Neglected Factor of Information Policy,” (paper presented
at the Telecommunications Policy Research Conference (October 5, 1998): 70. Thus,
“the availability of a commons creates incentives that make possible decentralization of
content-production.” Benkler, “The Commons,” 68.
8 “These units are so small as to make the transaction costs involved in negotiating

allocation of exclusive property rights to them prohibitive.” Benkler, “Overcoming Ago-
raphobia,” 174.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 326

3 2 6 N O T E S

9 See, e.g., Andy Kessler, velcap.com, “Steal This Bandwidth!,” e-mail on file with

author, June 19, 2001 (“The FCC should set aside some not-for-profit spectrum specifi-
cally for wireless access, probably at the same time they auction 3G licenses, and keep
encouraging the grassroots to run with new technology.”).
10 Telephone interview with Alex Lightman, January 31, 2001.
11 Ibid.
12 Telephone interview with Dave Hughes, November 13, 2000.
13 Federal Communications Commission, “Creation of Low Power Radio Service,”

January 27, 2000, at http://www.fcc.gov/Bureaus/Mass_Media/Orders/2000/fcc00019.doc.
14 Harry First, “Property, Commons, and the First Amendment: Towards a Core Com-

mon Infrastructure” (White Paper of the First Amendment Program at the Brennan
Center for Justice at New York University School of Law, 2001), 42–44.
15 Departments of Commerce, Justice, and State, the Judiciary and Related Agencies

Appropriations Act, 2001, H.R. 4942, enacting into law H.R. 5548, 106th Congress,
Title VI §632 (2000) (enacted).
16 See Bob Brewin, “Airports Ground Use of Wireless: Safety, Loss of Income from

Pay Phones Cited,” Computerworld (February 19, 2001): 1, http://computerworld.com/
cwi/story/0,1199,NAV47_STO57817_NLTpm,00.html.
17 The real aim of the airports, organized under “wirelessairport.org,” appears to be to

develop a proprietary, exclusive architecture for wireless technologies within airports, to
permit them to extract rents from the technology’s use.
18 See, e.g., Hazlett, “The Wireless Craze,” 42 (“The burden of proof is on the poten-

tial entrant. No incumbent must show that less competition serves the public in order to
preserve the status quo, it must only rebut proponents of competition. The default posi-
tion is that entry does not occur.”); Net@EDU, Position Paper on WLAN Radio Fre-
quency Interference Issues (“FCC Part 15 devices like wireless access points must accept
all interference that may be caused by the operation of an authorized radio station,
by another intentional or unintentional sources [sic].”), 1, available at http://www.
educause.edu/asp/doclib/abstract.asp?ID=NET0014 (visited June 11, 2001); ibid. (dis-
cussing ham radio). For a site collecting the FCC’s Technical Advisory Committee’s
work on spectrum, see http://www.jacksons.net/tac/.
19 Hazlett, “The Wireless Craze,” 45.
20 Thus, in setting aside 300 MHz within the U-NII band, the FCC stated:

We note that it may also be appropriate to reassess the technical parameters governing
U-NII devices in light of second generation MSS systems. For example, second genera-
tion MSS systems may be more sensitive and therefore more susceptible to interference
from U-NII devices.
Amendment of the Commission’s Rules to Provide for Operation of Unlicensed NII
Devices in the 5 GHz Frequency Range, 12 F.C.C.R. 1576, ¶96 (1997). But why should
U-NII devices respond to second-generation MSS devices, rather than the other way
around? See Yochai Benkler, “Overcoming Agoraphobia,” 338, n. 225.
21 FCC Regulations, part 97, subpart A, section 97.1, available at http://www.arrl.org/

FandES/field/regulations/news/part97/.
22 As Hazlett argues, the current regime does not actually sell the right to spectrum.

Instead, it sells the right to a certain kind of business, since the license received is a li-
cense to use certain equipment at a certain time for certain business purposes. Hazlett,
“The Wireless Craze,” 102. Thus the criticism of spectrum auctions that I offer here is
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 327

N O T E S 3 2 7

not a general criticism of auctions; it instead is critical, as Hazlett is, of the particular
mode of auctions currently pursued.
23 Ibid., 105.
24 S.J. Res. 125, 69th Cong., 1st Sess. (1926) (Clarence C. Dill).
25 Telephone interview with Dave Hughes.
26 A distinct but related concern is that the relative demand for different uses of spec-

trum can’t be known in advance. As Reed puts it, “[W]hether HDTV or more cellular
telephone service would be more useful to society” is not something we can know in ad-
vance. Telephone interview with David Reed, February 7, 2001. Yet the existing system
of allocation presumes as much.
27 Noam, “Beyond Liberalization II,” 473.
28 Telephone interview with Alex Lightman.
29 John Naughton, A Brief History of the Future: The Origins of the Internet (London:

Weidenfeld & Nicolson, 1999), 107.
30 Telephone interview with David Reed.
31 As well as other great books. See, for example, Orwell’s Revenge: The 1984

Palimpsest (New York: Free Press, 1994).
32 Huber, Law and Disorder in Cyberspace, 75.
33 As Carol Rose argues, however, commons traditionally are not regulated by the state

or through law. They instead are regulated through norms. Indeed, their nonstate char-
acter is an important element of their status a commons. Rose, “The Comedy of the
Commons,” 720–21. But today, the most important commons are built by the state and
regulated under general law. The highways and parks, for example, are not found by the
state anymore. For more on the role of highways in engendering growth because of their
commonslike character, see “Toll Roads and Free Roads,” 76th Congress, 1st Sess.,
House Document No. 272, 98 (1939) (Serial Set at 10339) (“The business-generating
potentiality of a heavy traffic stream is so great that there is an immediate development
of a great variety of roadside establishments all along every new heavily traveled route
that is created. Every new highway also, especially in the vicinity of cities, immediately
encourages residential development and attracts commercial establishments more in-
terested in the new facility provided by it than in catering to its traffic.”); Clay Commit-
tee Report, “A 10-Year National Highway Program,” 84th Congress, 1st Sess., House
Document No. 93, 7 (1955) (Serial Set at 11840) (“An adequate highway system is vital
to the continued expansion of the economy. The projected figures for gross national
product will not be realized if our highway plant continues to deteriorate. The relation-
ship is, of course, reciprocal; an adequate highway network will facilitate the expansion
of the economy which, in turn, will facilitate the raising of revenues to finance the con-
struction of highways.”).
34 The origin of this idea is a 1959 paper by J. P. Costas, “Poisson, Shannon, and the

Radio Amateur,” 47 Proc. IRE 2058 (December 1959). See also Shepard, “A Channel
Access Scheme for Large Dense Packet Radio Networks.”
35 Hazlett, “The Wireless Craze,” 136–37. Thomas W. Hazlett, “An Essay on Airwave

Allocation Policy,” AEI-Brookings Joint Center for Regulatory Studies (2001).
36 See, e.g., International Bandwidth 2001 (Washington, D.C.: TeleGeography, 2001).
37 As David Reed describes it:

We don’t understand how the capacity of communications relates to the number of users.
[The historical model assumes] capacity is constant. So if you have ten users, you’ll be
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 328

3 2 8 N O T E S

able to divide it up ten ways and each user will get one-tenth at most. . . . [But] there are
two . . . plausible arguments from physics that describe constructible systems where, as
the number of users of that same spectrum in the same volume of space increases, the
amount of available spectrum [increases as well] . . . which means that if you have one
hundred times as many stations in that same volume, you can get ten times (or maybe
even a hundred times) as much communication capacity.
Telephone interview with David Reed.
38 Noam, “Beyond Spectrum Auctions.”
39 Ibid., 465.
40 Ibid., 466.
41 George Gilder, Telecosm: How Infinite Bandwidth Will Revolutionize Our World

(New York: Free Press, 2000), 159.
42 Ibid., 159–60. Gilder’s point is correct whether or not there is a true “winner’s

curse.” A “winner’s curse” exists only when the bid was irrationally high. Richard H.
Thaler, “Anomalies: The Winner’s Curse,” Journal of Economic Perspectives 2 (1988):
191, 192. Whether or not the bid was irrational, it can still create this pressure on the sys-
tem.
43 The story is told in Lawrence Lessing, Man of High Fidelity: Edwin Howard Arm-

strong (New York: J. B. Lippincott, 1956).
44 The best example is the slow deployment of ultrawideband (UWB) technologies. A

kind of spread spectrum technology, UWB uses extremely low power transmissions that
do not rise above the noise floor. The application process at the commission has slowed
the approval process of UWB, as incumbents who would be affected by this new com-
petition can slow the process merely by raising questions about the potential interference
from this new technology. This shifts the cost of demonstrating the value of UWB to the
entrant, which significantly slows (and can kill) the entry process. Thus, “despite the
enormous spectrum efficiencies of UWB . . . UWB is going nowhere fast.” Hazlett,
“The Wireless Craze,” 87–88.

CHAPTER 13
1 Telephone interview with Pat Feely (July 18, 2001).
2 Interview with John Seely Brown in Stanford, Calif. (May 2, 2001).

CHAPTER 14
1 George Gilder, Telecosm: How Infinite Bandwidth Will Revolutionize Our World
(New York: Free Press, 2000), 163 (“[T]he FCC should not be in the business of licens-
ing spectrum. It should instead issue driver’s licenses for radios”); telephone interview
with Bill Lane, November 15, 2000 (describing regulatory role); telephone inter-
view with Dave Hughes, November 13, 2000 (describing regulatory role).
2 Software-defined radios were first demonstrated in 1995. Federal Communications

Commission, “In the Matter of Inquiry Regarding Software Defined Radios,” ET Docket
No. 00-47, FCC 00-103 (March 21, 2000), par. 4.
3 See, e.g., 42 U.S.C. 7601 (2001), Title III, Sect. 5(A) (granting a six-year compliance

extension only to air polluters who appreciably reduced their emissions before the EPA
deadline); Title IV, Sect. 3(b) (allowing factory owners’ only “substitution” of SO2 fa-
cility reduction requirements to a separate facility under the same ownership).
4 47 U.S.C. §301 (2001).
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 329

N O T E S 3 2 9

5 FCC Regulations, part 97, available at http://www.arrl.org/FandES/field/regulations/

news/part97/. Yochai Benkler proposes that the FCC reopen U-NII proceedings to fur-
ther free spectrum about the 6 GHz layer. Benkler, “Overcoming Agoraphobia,” 297.
Similarly, the FCC could allow part 15 unlicensed operations in the other ISM bands—
such as that allocated under part 18 to 24.05–24.25 GHz. FCC Regulations, part 18,
available at http://www.access.gpo.gov/nara/cfr/waisidx_98/47cfr18_98.html.
6 See, e.g., First, “Property, Commons and the First Amendment,” 9 (“Just as munici-

palities provide sidewalks, roads, and sewers, so too it is important to revive an interest in
designing the best possible approach towards public investment in the core infrastruc-
ture. This would require careful selection of the right level of public investment. At pres-
ent, some municipalities are acting in ways that suggests that such public investment
could entail deploying conduits and dark fiber [fiber without the electronics attached]
cables in municipal streets or sewage systems.”).
7 See ibid., at 54–55.
8 Canada has been experimenting extensively with this mode of providing connectiv-

ity. Through the development of “customer-owned IP networks,” the government is sup-
porting low-cost provision of IP services. For a description, see Timothy Denton,
“Customer-Owned IP Networks,” March 2001, available at http://www.tmdenton.com.
9 In the first half of 2001, Microsoft launched a campaign against government support

for what it called “open source” projects. These projects were “intellectual property-
destroyers,” since the license under which the code is distributed requires that derivative
work also be distributed subject to the same license. This destroys intellectual property,
Microsoft argued, because it means that no company can use “open source” products
without losing control over its own code. See, e.g., Ben Charny, “Microsoft Raps Open-
Source Approach,” CNET News, May 3, 2001, available at http://news.cnet.
com/news/0-1003-200-5813446.html (Mundie); interview with Bill Gates, Comdex, Fall
2000, available at http://www.key3media.com/comdex/fall2000/daily/keynotes/gates_
interview.html (Gates); Paula Rooney, Balmer: “Linux Is Top Threat to Windows,”
TechWeb, January 10, 2001, available at http://content.techweb.com/wire/story/
TWB20010110S0006 (Balmer).
If you’re this far into this book, then the mistakes in this argument should be patent.
The only open source or free software license that has a “viral” component is the GNU
GPL. “Open source” software is not software governed by the GPL. Thus Microsoft’s ar-
gument has nothing to do with “open source” software such as the Apache server or BSD
Unix. The only possible target of Microsoft’s attack is software licensed under the
GPL—namely, GNU/Linux.
But even here, Microsoft’s claims are just false. It is not the case that any software that
runs with Linux thereby becomes subject to the GPL, any more than any software
that runs on the Windows platform becomes subject to the license of Microsoft. The
GPL requires that only derivative work be licensed under the GPL—if you take the
Linux operating system and modify it in a particular way and then publish it to others,
then the product you publish must be under the GPL. That is, no doubt, a restriction on
the freedom of developers. They can’t simply “take” Linux and do with it as they wish.
But neither can anyone simply “take” the Windows operating system and do with it as
they wish.
The essence of Microsoft’s arguments is simply that the government should not sup-
port GPL research. Government funds should not promote a coding project that is not
wholly free for anyone to take and do with as they wish. Thus, Microsoft should have
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 330

3 3 0 N O T E S

no complaint if the government supports coding projects that are dedicated to the pub-
lic domain (like TCP/IP or the protocols of the Web). Its only complaint should be
when the government supports projects that are restrictive of citizens’ freedom—in
any way.
This is not an argument against open source, it is an argument against GPL. And if it
is a strong argument against GPL, then it is also an argument against the government
supporting proprietary projects as well. If it is wrong for the government to support Linux
because I am not free to do with Linux as I wish, then it is wrong for the government to
support Windows because here too I am not free to do with Windows as I wish.
If that is the principle Microsoft is advancing, it is an interesting and valuable point to
consider. But only if it is a principle, as opposed to FUD designed to scare the market
from a competitor’s products. We’ll have to see more before we can tell whether this is
principle or something less.
10 There may be a constitutional limitation on this form of regulation. At least

one court has concluded that requiring open access on cable lines violates the First
Amendment. See Comcast Cablevision of Broward County, Inc. v. Broward County, Fla.,
124 F. Supp. 2d 685 (S.D. Fla., 2000). For a powerful, if depressing, analysis, see Bran-
dan I. Koerner, “AT&T’s First Amendment Problem, and Ours,” The New Republic
(May 14, 2001): 18.
11 This is related to the National Research Council’s suggestion that the government

push “open IP services”—meaning a policy to preserve the essential features of end-to-
end IP. See National Research Council, 138–39.
12 Report of the National Broadband Task Force, The New National Dream: Network-

ing the Nation for Broadband Access (2001), 96. The OECD report is described in
“Broadband Blues,” The Economist (June 21, 2001).
13 Jessica Litman, Digital Copyright (Amherst, N.Y.: Prometheus Books, 2000), 78.
14 Some of the changes I propose here would require changes to or the abrogation of

some treaties. Four treaties are especially relevant. The first is the Berne Convention
for the Protection of Literary and Artistic Works, 828 U.N.T.S. 221, S. Treaty Doc. No.
99-27, which the United States joined in 1989. The Berne Convention provides for na-
tional treatment of foreign authors, meaning that each signatory nation is required to
protect foreign authors to the same degree that it protects domestic authors. Additionally,
the Berne Convention sets a limit on the minimum level of protection that a signatory
nation can offer and requires that materials be automatically protected from the moment
of creation, rather than being protected only once they are registered.
The second treaty is the Agreement on Trade Related Aspects of Intellectual Property
Rights, Including Trade in Counterfeit Goods, of the General Agreement on Tariffs and
Trade, 33 I.L.M. 83 (1994), better known as the TRIPs agreement. The TRIPs agree-
ment was adopted as part of the Uruguay Round of GATT. It sets out minimum stan-
dards of protection for intellectual property and is unusual in providing sanctions for
copyright violations. Member nations must enact laws giving foreign authors legal reme-
dies for copyright infringement. The agreement not only sets standards for those reme-
dies, but also subjects member nations to trade sanctions if they do not meet the
standards. TRIPs also narrows the scope of fair use by requiring that exceptions to copy-
right protection must occur only in special cases, must not conflict with the normal ex-
ploitation of the work, and must not unreasonably prejudice the interests of the holder
of the copyright.
The third and fourth treaties were adopted at the UN’s World Intellectual Property
Organizations (WIPO) Diplomatic Conference in 1996. They are, respectively, the
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 331

N O T E S 3 3 1

WIPO Copyright Treaty, December 20, 1996, WIPO Doc. CRNR/DC/94 (Decem-
ber 23, 1996), and the WIPO Performances and Phonograms Treaty, December 20,
1996, WIPO Doc. CRNR/DC/95 (December 23, 1996). The WIPO treaties expand the
protection afforded to on-line works by requiring countries to extend copyright laws to
the Internet. In the United States, this extension took the form of the Digital Millenium
Copyright Act of 1998 (DMCA), which prohibits both acts circumventing copy protec-
tion and the importation, manufacture, or sale of technologies developed primarily for
such circumvention. Willful violation of these provisions is subject to criminal penalties,
and both criminal and civil sanctions may be applied to violators even if the underlying
use is privileged (even if, for example, the use were to fall within traditional fair use).
15 There is an obvious, and important, problem of security raised by such a system,

though there are also obvious and feasible ways to minimize any security risk. For an ex-
ample of a secure system that was used to effect a settlement between IBM and Fujitsu,
see Robert H. Mnookin and Jonathan D. Greenberg, “Lessons of the IBM-Fujitsu Arbi-
tration: How Disputants Can Work Together to Solve Deeper Conflicts,” Dispute Reso-
lution Magazine (Spring 1998): 16.
16 Under existing law, “if the intended use is for commercial gain,” the likelihood of

market harm can be presumed. See Sony Corp. of America v. Universal City Studios,
Inc., 464 U.S. 417, 451 (1984). This presumption could be modified in the context of the
Internet so that innovators could defend a new use by demonstrating that no substantial
likelihood of harm to existing markets exists.
17 Of course, it is not as if artists are really being paid under the existing system—or at

least, not most of them. In 1999, eighty-eight recordings accounted for 25 percent of all
record sales. Charles C. Mann, “The Heavenly Jukebox,” Atlantic Monthly (September
2000).
18 The proposals for this are many. As Jon Potter describes it:

Once the art is disseminated to a single reseller, then other resellers can also have that art
for resale for the same terms and conditions, so maybe you can have a standardized
nondiscriminatory license provision rather than a compulsory statutory scheme of royal-
ties. So, if Sony needs to sell something through a Sony store, and tries to really control
all distribution, then Barnes & Noble gets to do the same deal—whatever the Sony store
does with Sony music, or whatever Barnes & Noble gets, Tower gets, but something so
that compensation is ensured. And competition is ensured.
Telephone interview with Jon Potter, December 7, 2000.
19 See, for example, “Artists’ Contribution to American Heritage Act of 2001,” 107th

Cong., 1st sess., H.R.1598; Artist-Museum Partnership Act, 107th Cong., 1st sess., S.694;
Arts and Collectibles Capital Gains Tax Treatment Parity Act, 107th Cong., 1st sess.,
S.638; Artists’ Contribution to American Heritage Act of 1999, 106th Cong., 1st
sess., H.R.3249; 106th Cong., 1st sess., S.217.
20 Stephen Fishman, The Public Domain (Berkeley, Calif.: Nolo, 2000), 2/9 (“Claim-

ing copyright in public domain works is a federal crime, but the maximum penalty for
engaging in this sort of criminal conduct is a fine of $2,500 [17 U.S.C. § 506(c)]. More-
over, no one has ever been prosecuted for violations [of this provision]”).
21 David Lange, “Recognizing the Public Domain,” Law & Contemporary Problems

44 (1981): 147, 166.
22 See Gail E. Evans, “Opportunity Costs of Globalizing Information Licenses: Em-

bedding Consumer Rights Within the Legislative Framework for Information Con-
tracts,” Fordham Intellectual Property, Media & Entertainment Law Journal 10 (1999):
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 332

3 3 2 N O T E S

267. For competing views on UCITA (most of them negative), see “Symposium: Intel-
lectual Property and Contract Law in the Information Age: The Impact of Article 2B of
the Uniform Commercial Code on the Future of Transactions in Information and Elec-
tronic Commerce,” Berkeley Technology Law Review 13 (1998): 809.
23 L. Ray Patterson, “Understanding the Copyright Clause,” Journal of the Copyright

Society U.S.A. 47 (2000): 365; L. Ray Patterson and Stanley W. Lindberg, The Nature
of Copyright: A Law of Users’ Rights (Athens, Ga.: University of Georgia Press, 1991);
L. Ray Patterson and Judge Stanley F. Birch, Jr., “Copyright and Free Speech Rights,”
Journal of Intellectual Property Law 4 (1996): 1; L. Ray Patterson, “Copyright and ‘The
Exclusive Right’ of Authors,” Journal of Intellectual Property Law 1 (1993): 1, 137; L. Ray
Patterson, “Free Speech, Copyright, and Fair Use,” Vanderbilt Law Review 40 (1987): 1.
24 Litman, 12–14.
25 The law does not ordinarily have an interest in forcing an owner of a property right

to “use it or lose it.” For example, I should not have to drive my car all the time to ensure
no one else gets to take it. But where property is not exhaustible, and is nonrivalrous,
then the balance in favor of the “use it or lose it” rule can shift. Litman’s rule is related
to a French requirement that if a publisher fails to exploit an assigned right, then, under
certain circumstances, the original author can reclaim the right. See Neil Netanel,
“Copyright Alienability Restrictions and the Enhancement of Author Autonomy: A Nor-
mative Evaluation,” Rutgers Law Journal 24 (1993): 347, 390–91.
26 On the role of damage remedies in patent infringement cases, see Mark Schanker-

man and Suzanne Scotchmer, “Damages and Injunctions in Protecting Intellectual
Property,” Rand Journal of Economics 32 (2001): 199 (in the context of research tools,
damages underdeter infringement, but that nondeterrence may be a good thing for the
patent holder, since it prevents a hold-up problem). Carl Shapiro, “Navigating the
Patent Thicket: Cross Licenses, Patent Pools, and Standard-Setting,” in Innovation
Policy and the Economy, vol. 1, Adam Jaffe, Joshua Lerner, and Scott Stern, eds. (Cam-
bridge, Mass.: MIT Press, 2001), 8.
27 Professor John Barton, for example, has proposed that Congress (1) raise the stan-

dards for patentability (such that an invention is nonobvious only “when the approach
seemed quite unlikely to work and still proved successful”); (2) decrease the use of
patents to bar research; and (3) improve the U.S. Patent Office to reduce invalid patents.
John H. Barton, “Reforming the Patent System,” Science 287 (2001): 1933.
28 See Brian Kahin, comments in response to “The Patentability of Computer-

Implemented Inventions,” 2000, 5 (noting that patent attorneys discourage software pro-
fessionals from reading patents to avoid “willful infringement”), available at http://
europa.eu.int/comm/internal_market/en/intprop/indprop/maryland.pdf.
29 See, e.g., Daniel R. Harris and Janice N. Chan, “Case Note: Wang Laboratories,

Inc. v. America Online, Inc. and Netscape Communications Corp.,” Computer & High
Technology Law Journal 16 (2000): 449, 457.
30 See, for example, Robert P. Merges, “As Many as Six Impossible Patents Before

Breakfast: Property Rights for Business Concepts and Patent System Reform,” Berkeley
Technology Law Journal 14 (1999): 577 (poor-quality patents, especially business patents,
reveal need for PTO reform; PTO jobs and incentives should be restructured; third par-
ties, especially the applicant’s competitors, should be consulted early and thoroughly
through a European-style opposition system).
31 See A Bill to Amend Title 35, U.S. Code, to Provide for Improvements in the

Quality of Patents on Certain Inventions, H.R. 107th Cong., 1st Sess., 1932, §4 (nonob-
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 333

N O T E S 3 3 3

viousness). Congressman Berman’s bill also proposes useful changes to facilitate early
challenge of business method patents before a patent on these methods is granted. See
A Bill to Amend Title 35, U.S. Code, to Provide for Improvements in the Quality of
Patents on Certain Inventions, H.R. 107th Cong., 1st Sess., 1332, §3.
32 Pamela Samuelson et al., “Manifesto Concerning the Legal Protection of Com-

puter Programs,” Columbia Law Review 94 (1994): 2308, 2331.
33 In March 2000, the U.S. Patent Office launched a Business Method Patent Initia-

tive designed “to ensure high quality patents in this fast-emerging technology.” Available
at http://www.uspto.gov/web/offices/com/sol/actionplan.html. Fifteen months later, the
U.S. Patent Office asked for comment on Prior Art Sources for Business Method Patents.
See http://www.uspto.gov/web/offices/com/sol/notices/ab26.html.

CHAPTER 15
1 Orrin G. Hatch, Address of Senator Orrin G. Hatch Before the Future of Music
Coalition, Future of Music Coalition, January 10, 2001, available at http://www.
senate.gov/~hatch/speech020.htm.
2 Eldred v. Reno, 239 F. 3d 372, 375 (D.C. Cir., 2001).
3 Testimony of Professor Peter Jaszi, The Copyright Term Extension Act of 1995: Hear-

ings on S.483 Before the Senate Judiciary Committee, 104th Cong. (1995), available at
1995 WL 10524355, *6.
4 Oral arguments, United States v. Microsoft, February 26, 2001, available at http://

www.microsoft.com/presspass/trial/transcripts/feb01/02–26.asp.
5 E-mail from John Gilmore, January 19, 2001, to EFF list, on file with author, 6.
6 Telephone interview with Marc Andreessen, December 15, 2000.
7 As Timothy Wu commented to me, “[T]he real successes on the Internet have not

been killer apps, but killer platforms.” E-mail from Tim Wu, June 28, 2001. Not, in
other words, amazing but proprietary applications that do extraordinary things, but
amazing and open platforms upon which others have been free to build. E-mail and the
Web are examples, Wu suggests, of killer platforms. Napster and Instant Messaging,
while popular, have none of the equivalent open platform characteristics.
8 David Bank’s recent book, Breaking Windows, argues that there is more reason to be

hopeful about Microsoft. According to Bank, “[I]nteroperability, not lock-in, has be-
come the winning strategy” for Microsoft. David Bank, Breaking Windows: How Bill
Gates Fumbled the Future of Microsoft (New York: Free Press, 2001), 237. In June 2000,
Microsoft “embraced XML and made it the centerpiece of its new Internet platform.”
The platform was designed, at least initially, to be a “complot for interoperability.” Ibid.,
198. Following Christensen’s recommendations in Innovator’s Dilemma, the strategy was
born from a spin-off that Microsoft created with executive Adam Bosworth. Ibid., 196. To
“his credit,” as Bank writes, “Gates never shut the XML effort down.” Ibid., 198.
9 Gordon Cook, “The Meaning of Current Events,” The COOK Report, June 20,

2001, available at http://cookreport.com
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 334
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 335

/ / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / / /

I n d e x

access: Time Warner merger with, 7, 164–66,
competition in, 173 267
granting of, 13–14, 20 wires of, 163–64
access fees, 20, 138 Apache server, 55–56, 60, 69–70, 71
adaptations, creative, 8 Apple Computer:
Adobe Corporation, 208 AirPort technology of, 81, 172, 224–225
advertising, 7 and DVD movies, 189
“ah-ha” technologies, 130 GUI interface of, 63
AIM, in instant messaging, 27 iMac computer of, 124
AirPort technology, 81, 172, 224–25 Macintosh computers of, 27–28, 42,
ALGOL, 51 63, 66
algorithms: Mac OS, 60–61, 63, 90–91
patents on, 207–8 and PC design, 159
peer-to-peer querying, 137 “rip, mix, burn” ad of, 9, 10, 11
Allman, Eric, 56 and unknown applications, 90–91
altruism, 13 AppleScript language, 58
Amateur Radio Service (ARS), 225, architecture:
244 changes in, 140
Amazon.com,132–33, 167, 169, 211 as choice, 37
Amnesty International, 184 code of, 35, 113
Andreessen, Marc, 265–66 of communication systems, 112–14,
Antheil, George, 79 232–33
anticircumvention program, 187–90 control via, 34, 151, 176
anticommons, 214–15 in cyberspace, 120–21, 140
antitrust laws, 67, 110 for distribution, 119
AOL (America Online), 162–68 freedom of, 35
access vs. control in, 163–66 importance of, 35
and competition, 167, 267 and innovation, 35–36, 92, 139, 140,
and copyright law, 183 162
early days of, 147–48, 162–63 of Internet, 7, 15, 16, 23, 35–37, 40,
and Internet, 7, 163, 165 44, 119, 135, 140, 210, 238, 264
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 336

3 3 6 I n d e x

architecture (cont’d): Bagdikian, Ben, 117
as physical constraint, 120 Baldwin, Carliss, 92
politics of, 35 band managers, 221
of the Web, 169, 170 bandwidth:
Armstrong, C. Michael, 154 broadening of, 229
ARS (Amateur Radio Service), 225, 244 infinite, 47
artwork: overuse of, 138
code layer in, 111–12 pricing system for allocation of, 47
content layer in, 105–07 Bank, David, 66, 67
control in, 107–10 Bar, François 158–59
copyright law applied to, 3 Baran, Paul, 26–27, 29–32, 33–34, 267
creativity in, 104–12, 200 Batman Forever (film), 4
digital, 8–9 Beam-it, 129, 192
physical layer in, 110–11 Behlendorf, Brian, 56
asynchronous transfer mode (ATM) Bell companies:
systems, 149 as Baby Bells, 149
AT&T: DSL used by, 155
as Bell system, 27 lines leased by, 45
breakup of, 45, 53, 149 patents held by, 27
cable of, 153–54, 164–65, 241 regulated to be open, 155, 248
in Canada, 164 regulation model for, 149
compensation system of, 30 services unbundled by, 149, 248
and competition, 28–29, 32–33, 37–38, see also AT&T
62, 158, 226 Bell Labs:
in computing business, 53 innovations channeled through, 30
consent decree with government innovative spirit of, 176
(1956), 51–52 inventions of, 29, 38
and FCC, 27, 30, 81, 221 and Unix operating system, 51–52
and foreign attachments rule, 30 Bender, James, 182–83
and Internet, 154 Benkler, Yochai, layer concept of, 23–25,
Kingsbury Commitment of, 28–29 103, 104, 113, 167, 238, 240
and monopoly power, 28, 29, 30, 32, Berkeley Internet Name Domain
148 (BIND), 56
in open access movement, 164, 165 Berlin, Irving, 107
permissible businesses of, 45 Berners-Lee, Tim, 37, 40, 41–44,
shareholder obligations of, 32 134,166, 213
smart network of, 38 BestBookBuys, 169
and telephone system, 24, 26–34, 44, 62 Bezos, Jeff, 211
and Unix operating system, 51–52, 53 Bidder’s Edge, 169, 170–71
and unknown applications, 90 BIND (Berkeley Internet Name
@Home, 154, 158 Domain), 56
Atkinson, Bill, 42 Bingaman, Anne, 64
ATM (asynchronous transfer mode) “Bluetooth” protocols, 81
systems, 149 Boies, David, 196
auction sites, 169 books:
audio changes, MP3, 123–24 banning of, 184
Audio Home Recording Act, 181 copyright protection of, 215
HTML, 122–23, 196–99
Baby Bells, 149 sales of, 211
backbone providers, 157 Boston Common, 19, 21
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 337

I n d e x 3 3 7

bots: creative rights in, 109
in copyright, 180–83 development of, 151
spiders and, 169–71 distribution of, 119
Bradner, Scott, 89 as end-to-end system, 151
Brief History of the Future, A (Naughton), incentives for, 151–52
32 layers of, 24
broadband: licensing of, 201
AOL and, 163–64 monopoly control of, 152, 153
cable, see cable broadband pricing of, 153
in Canada, 249 satellite TV and, 153
competition in, 173, 248 caches, 136–37
control of ISPs in, 246 Canada:
emerging technology of, 159, 161, 175 free TV in, 190–91
faster downstream than upstream, 159 open access in, 164, 192, 249
FCC and, 159, 164 Carlton, Jim, 90–91
markets in, 159 Carterfone decision, 148
open access for, 164, 165, 168 CATV, 151
broadcasting, see radio spectrum; television CBS, and spectrum allocation, 74
Brotz, Douglas, 208 CDDB (culture databases), 124–26
Brown, John Seely, 235–36 censorware, 184, 186–87
browsers, 41, 43, 134 CERN labs, 41, 42–44
BSD Unix, 53, 67 Charmed Technologies, 81–82
Buchanan, James, 214, 215 children, protection of, 178–79, 184, 200
bugs, in code, 61 China Online, 191
bundling behavior, 66 chip fabricators, 70
bureaucratization, 140–41 Christensen, Clay, 89, 91, 92, 139–40,
Burk, Dan, 170 146, 210
Bush, Vannevar, 42 Cisco technologies, 156, 158
business methods: Clark, David, 34, 37, 38
patents on, 207, 208–9, 210 Clark, Kim, 92
software-implemented, 208–9 Coase, Ronald, 12, 75, 114, 126, 221
COBOL (programming language), 51
cable broadband, 151–54, 166–68 code:
building control into, 156–58, 168 anticircumvention program and, 187–90
and Communications Act, 155 bugs in, 61
and competition, 161–62, 167 changes in, 140–41, 179
and DOCSIS standard, 153–54 commons of, 49–50, 53, 55, 57, 138
and FCC, 151–52 compiled, 50, 253
incentives to build, 174 as content, 50
and Internet, 153–54, 155–58, 167 content production in, 11, 53, 58–60, 67
and streaming video, 156–57, 158 derived from open code, 55, 59, 60
and TCP/IP, 248 encryption cracked by, 190
two-way communication on, 153–54 fair use of, 188, 190
Cable Labs, 153 forking of, 52, 67–68
cable television: hard-wired, 50–51
advertising on, 7 hidden, 58
broadcasts on, 111 incentives to build, 68–69, 71
CATV, 151 as law of cyberspace, 35
concentrated ownership of, 117, 159, licensing of, 59–60
173 limits on, 256–57
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 338

3 3 8 I n d e x

code (cont’d): inverse, 68
as logical layer, see code layer of knowledge, 49, 55, 57–58
nature of, 35–36, 50 market linked to, 138, 228
object, 50 norms for control of, 96
openness of, 52–54, 55–57, 59–60, 61, peer-to-peer innovations in, 137–38
67–69, 71–72, 210, 247, 253 rebuilding of, 255–56
performative, 57–58 as resource for decentralized
source, 50, 60, 67, 72 innovation, 85
code layer, 23–25 state intervention in, 228
in the arts, 111–12 tragedy of, 22, 23, 48, 168
changes to, 246–49 undermining technology added to,
closing of, 217, 238 48
in commerce, 112–14 value of, 87–88, 93, 173
end-to-end at, 58 Communications Act (1934), 74
innovation commons at, 85, 103, 138, on “cable services” (Title VI), 155
167, 175 on telecommunications services
of Internet, 48, 49–50, 56, 138, 161 (Title II), 155
neutral platforms of, 72, 246–49 Communications Decency Act (1996),
of the Web, 57, 85 178–79
Code and Other Laws of Cyberspace communications systems:
(Lessig), 35, 140, 179, 238 architecture of, 112–14, 232–33
Cohen, Julie, 257 on computers, 112–14
Cohen Theorem, 257 layers of, 23–25
Cold War, 12, 13 organizational models for, 167
command line interface, 62, 63 security of, 26–34
commerce: spectrum, see spectrum
censorship in, 186–87 two-way, 153–54
changes in policy of, 165 wireless, 74–75
code layer in, 112–14 wires of, 88
competition in, 114–15 community:
creativity in, 112–19 as built by noise, 133
of the different, 112 CATV, 151
economics and, 115 connectivity within, 80
innovation in, 6, 10 monopoly power over, 87, 216
markets in, 114–15 networks built by, 162
vertical integration in, 165–66 on-line, 162–63
commercial radio, 74 overconsumption regulated by, 22
commons, 19–23 relationship of resource to, 21
access fees for use of, 20, 138 spectrum of, 223–24
anticommons vs., 214–15 town square in, 87
built on control layer, 176 compensation without control, 201–2
of code, 49–50, 53, 55, 57, 138 compression technology, 123
coexistence of open and closed, 93–94 compulsory licensing right, 109, 201,
contributing back to, 97 255
as default, 228–29 CompuServe, 147–48
determinants of, 21 computer-communication systems,
freedom of, 15 architecture of, 112–14
as free resource, 19–21, 44, 69, 86–99 computers:
idea of, 15 ad hoc supercomputers, 136
innovation of, see innovation commons applications run on, 36–37, 88
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 339

I n d e x 3 3 9

competition in field of, 27–28 control of, 7, 107–8, 110, 156–58, 176,
costs of, 113 179–80, 183, 248
denial of service attack on, 169 CSS, 189
dominant companies in, 27 distribution of, 238
early coding of, 50–51 dynamic, 137
at edge vs. within networks, 34, 36–37 encryption of, 188–90
IBM-compatible, 13, 62, 65, 159, 189 exclusivity of, 128
icons of, 60–61 firewalls for blocking of, 172–73
incompatibility between, 41–42, 51 free, 20, 50, 106, 108, 110, 123, 177,
Intel chip architecture in, 62, 63, 66 196, 256
Internet and, see Internet indecent, 178–79, 184, 199
Mac design, see Apple Computer legal protection of, 10–11, 59, 183,
object code of, 50 185, 250–59
as physical layer, 23 policy-based routing of, 156
as property, 241 range of, 131
software of, see software as separate from medium, 166
wearable, 81–82 space-shifting of, 194
Web and, see World Wide Web content layer, 23–25
Computer Telephony, 38 in the arts, 105–7
Congress, U.S.: changes in, 249–61
and cable television, 151–53, 201 closing of, 217, 238
and copyright law, 98, 196–99, 200, copyrights in, 250–59
201, 254 digital technology in, 23, 120–21
First Amendment restrictions on, 187 free code at, 58, 72, 85
and indecent content, 178–79, 184 freedom at, 50
influence of lobbyists on, 267 innovation at, 103
and intellectual property, 187–88 patent law and, 259–61
new technologies confronted by, tension between freedom and control
108–10, 190 in, 177, 249
and patent law, 207, 259 content scramble system (CSS), 189
and phone company, 148 contract law, 58–60, 97, 186, 257
rebuilding the creative commons, control:
255–56 architecture of, 34, 151, 176
and recording industry, 201–2, 254 balance of freedom vs., 14–15, 97, 139
and spectrum, 224, 255 compensation without, 201–2
connectivity, as Internet goal, 36 constitutional limits on, 105–6
consent decree, 64 of content, 7, 107–8, 110, 156–58, 176,
Constitution, U.S.: 179–80, 183, 248
Copyright and Patent Clause in, 177, in creative process, 107–10
197–98, 237 in cyberspace, 237–38
Dormant Commerce Clause of, 114 filtering as, 157
Exclusive Rights Clause of, 116 as good, 13, 15, 86
constraints: imposed via copyright, 180–217
of architecture, 120 imposed via norms, 96, 146, 151
on creativity, 7, 8, 11, 104, 202, 216 imposed via technology, 96–97, 140
on markets, 114–15 increasing, 99
in scarcity-based economy, 103 by lawyers, 4, 11, 58–60
consumption vs. production, 13 local, of worldwide technologies, 192
content: of networks, 150–51, 156–58
code as, 50 perfect, 116, 217, 250
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 340

3 4 0 I n d e x

control (cont’d): narrow scope of, 8, 106–7
pressure toward, 168 noninfringing use under, 195–96
private incentives for, 168, 174 old vs. new in, 202–5
of property, 22, 170–71 potential use under, 195–96, 200
of resources, see resources and site shutdowns, 183
scarcity and, 47 as state-backed monopoly, 250
by state vs. market, 12, 14 “thick” vs. “thin” protection in, 202–3
techniques of, 96 time limits in, 106, 107, 188, 197,
trends toward, 249 251–52
in vertical integration, 165–66 unenforceable violations of, 181, 182
via firewalls, 172–73 unlicensed uses in, 180–81
of wires, 26, 147–76 Corel Corporation, 262
Cook, Gordon, 267 Cox, Alan, 57
copy protection system, 253 CPHack, 184–87, 188, 190
copyright law: creation vs. discovery, 106
as agent for control, 180–217 creativity:
aim of, 98 in adaptations, 8
anticircumvention program in, 187–90 in the arts, 104–12, 200
application of, 58 before the Internet, 8, 104–19
in art industry, 3 centralization of, 44
bots applied to, 180–83 in commerce, 112–19
broad application of, 4–5 constraints on, 7, 8, 11, 104, 202, 216
censorship via, 184, 186–87 control in process of, 107–10
commercial exploitation limited in, with digital tools, 9
258–59 in filmmaking, 235
in Constitution, 177, 197–98, 237 free resources as basis of, 12–15, 44
constraints of, 8 incentives for, 4, 21, 59, 109
content protected in, 10–11, 183, 185, and innovation, 10
250–59 intellectual property in, 203
contract law and, 58–60, 186 in the Internet, 6, 9, 14, 23, 76
contributory infringement under, 196 legal control of, 4, 58–60
control uncontrolled in, 183, 200 licensing and, 216
for derivative works, 106–7, 123, 198–99 new built on old in, 9, 13, 202, 204, 214
DMCA, 187, 88, 190 potential for, 9–11
exceptions built into, 181 rebuilding commons of, 255–56
expansion of, 98, 106–7, 110, 196–97, and subsidiarity, 121
216, 264 tangible forms of, 58
fair use of, 105, 108, 181, 185, 188, Crocker, Steve, 36
190, 195, 254 CSS (content scramble system), 189
false claims under, 256 culture, production of, 263–64
in film industry, 3–4, 11, 200 culture databases (CDDB), 124–26
and First Amendment, 198, 264 CyberPatrol, 184–86
foreigners excluded from, 106, 185, cyberspace:
190–91 architecture in, 120–21, 140
incentives provided by, 59, 98, 107, character of, 121
197, 201 code as law of, 35
intent of, 11, 187 control in, 237–38
on the Internet, 98, 178–79, 190–92, as nonrivalrous, 116
200, 216 vs. real space, 103–4, 121, 181, 182
in music industry, 3, 192–94, 254–55 world of ideas in, 116
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 341

I n d e x 3 4 1

dark fiber, 245 of content, 238
data: costs of, 7, 126, 138
balance of power changed by, 128, 133 of media, 119, 234
and code layer, 138 models of, 7
different classes of, 46–47 of music, 119, 123–24, 127–29, 131,
real space costs of, 138 200–202
Davies, Donald, 31 new means of, 126–29
DeCSS, 187–90 DMCA (Digital Millennium Copyright
Defense Advanced Research Projects Act, 1998), 187, 188, 190
Agency (DARPA), 242 DOCSIS standard, 153–54, 161–62
Defense Communication Agency (DCA), Dormant Commerce Clause, 114
33–34 DOS (disk operating system), 62–66
Defense Department, U.S., 27, 30, 32, DR-DOS, 62, 63, 64, 65
44, 79 DSL (digital subscriber line), 154–55,
democracy: 161–62, 248
blocked transactions in, 93 Dunkin’ Donuts, 182
centralization as opposed to, 112 DVD disks, 188–90, 256
free resources in, 12–13, 92–93
derivative works, 8, 106–7, 123, 198–99 Easterbrook, Frank, 186
DES (digital encryption standard)- eBay, 169–71
encrypted message, 136 economics, and scarcity, 103, 115
design evolution, 92 Economist, The, 206
Devil’s Advocate, The (film), 4 Edelman, Ben, 191
Dickinson, Q. Todd, 210, 211 education, and filmmaking, 235–36
digital age, 14 efficiency, 92
Digital Millennium Copyright Act of 802.11b protocol, 81
1998 (DMCA), 187, 188, 190 Einstein, Albert, 12, 13, 14, 20, 21
Digital Research, 62 Eldred, Eric, 122–23, 196–99, 202, 264,
digital subscriber line (DSL), 154–55, 265
161–62, 248 Eldritch Press, 123
digital technology: electricity grid, as end-to-end, 39
competition in, 201 Electronic Frontier Foundation, 196
compulsory licensing rights to, 109 Emacs, 54
in content layer, 23, 120–21 e-mail, 10
for copying, 98 EMI Publishing, 183
falling costs of, 8, 235 encryption systems, 188–90
in film industry, 9, 235 end-to-end (e2e) argument, 34–39, 88–89
music processor, 8–9, 96 central control disabled in, 40, 167–68
potential of, 7, 9 compromises to, 172–73
for telecommunications, 31 consequences of, 41–44, 48, 58
DirecTV, 153 costs of, 46
discovery vs. creation, 106 design choice of, 121
discrimination, technologies for, 173 on different layer, 45–46
disk drive industry, 90 discrimination in, 173
Disney Corporation, 234 importance of, 91
distributed.net, 136 and innovation commons, 36–37,
distributed processing, 136 40–41, 48, 58, 147, 156, 210, 238–39
distributed technology, 136 intelligence in, 149
distribution: neutral network of, 58, 61, 161
architecture for, 119 Enlightenment, 94, 95
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 342

3 4 2 I n d e x

Epstein, Richard, 170 forking, 52, 67–68
essential volunteerism, 56 Fourneaux, Henri, 108
Ethernet, 77–78, 231 Fox Studios, 181, 182
European Union, 64 Frankfurter, Felix, 74–75
exclusivity, 116, 128 Franklin Benjamin, 206
FRC (Federal Radio Commission), 73
fair use: free, meanings of the word, 11, 12, 15, 20
reverse engineering and, 185 free culture, 9–10
rights of, 105, 108, 181, 188, 190, 254 freedom:
VCR and, 195 in the architecture, 35
Fanning, Shawn, 130 vs. communism, 12, 13
Fano, Robert, 42, 112–14 in computer world, 52
fat pipes, 151–53 as constitutional question, 11
Faulhaber, Gerald, 176 at content layer, 50
Federal Communications Commission vs. control, 14–15, 97, 139
(FCC): as enhancing the social value of the
and AT&T, 27, 30, 81, 221 controlled, 48
and broadband, 159, 164 of the press, 197
and cable, 151–52 of speech, 116, 140, 178–79, 184,
and future of technology, 263 197–98, 224
and spectrum, 74, 75, 80–81, 218–21, to tinker, 61, 68
223–27, 244, 246 Free Software Foundation, 12, 53–54, 59
Federal Radio Commission (FRC), 73 Frost, Robert, 197
Federal Trade Commission (FTC), 166 FTC (Federal Trade Commission), 166
Feely, Pat, 234
Ferguson, Warren, 195, 196 Gates, Bill, 62–67, 206, 214
fiber optic cable, 245 Gaynor, Mark, 89
fiber optic technology, 29, 47 General Public License (GPL), 44, 54,
File Synchronization, 60–61, 66 59–60, 97, 185–86
film industry: geography, constraint of, 114
concentrated ownership in, 117 Gilder, George, 47–48, 83–84, 227–28, 232
copyright law applied to, 3–4, 11, 200 Gilmore, John, 265
digital technology in, 9, 235 Gingrich, Newt, 118
distribution in, 119, 234 GNU C Compiler (GCC), 54
DVD disks and, 188–90, 256 GNU/Linux OS, 54–55, 56, 57, 60, 62, 69
education and, 235–36 GNU OS, 53–54
lawyers in control of, 4–5, 11 Gnutella, 137, 138
production costs in, 8, 124 Goldsmith, Jack, 192
and VCR, 195 Gone with the Wind (Mitchell), 198–99
filtering, 157, 184 Gopher, 43
Firefly, 133 government:
firewalls, 172–73 avoiding control by, 80–81
First Amendment: commons created by, 45, 155
and code layer, 112, 190 control of, 14, 92, 93
Congress restricted by, 187 exclusive rights granted by, 7
copyrights as immune from, 198, 264 in open IP service, 176
free speech in, 116, 197–98, 224 phone companies regulated by, 27, 30,
protest under, 93 45, 81, 148, 161, 221
First Year, The (film), 3 spectrum controlled by, 73–76, 79–81,
foreign attachments rule, 30 83, 232, 246, 255
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 343

I n d e x 3 4 3

spectrum hoarded by, 242 modular design in, 92
see also Federal Communications and open code, 69–71
Commission OS/2 computers, 28, 65
GPL (General Public License), 44, 54, IBM-compatible PCs, 13, 62, 65, 159,
59–60, 97, 185–86 189
graphical user interface (GUI) operating icons, use of term, 60–61
system, 62, 63 iCraveTV, 190–92
Guggenheim, Davis, 3–4 ideas:
GUI-OS, 62, 63 copyright of, 98, 187
expression of, 188
Hamilton, Alexander, 265 free exchange of, 49, 94, 95
Hand, Learned, 106 nonrivalrous character of, 94
Hardin, Garrett, 22, 168, 175 taken for granted, 5, 13, 15
hardware, code of, 35 as thinking power, 94
Harry Fox Agency, 183 world of, 116
Hart, Michael, 123 iMac computers, 124
Hartford Courant, 118 impact statement, need for, 211
Hatch, Orrin G., 262–64, 267 infringement, contributory, 196
Hayek, Friedrich von, 212 innovation:
Hazlett, Thomas, 74, 75–76, 84, 218–19, advantages to, 229
225, 229–30, 241 and architecture, 35–36, 92, 139, 140,
Heller, Michael, 214, 215 162
Hemingway, Ernest, 252–53 centralized control of, 32
Hendricks, Dewayne, 80–81, 96 changes in the environment for, 236–37
highways: in commerce, 6, 10
access to, 14 commons of, see innovation commons
as commons, 20, 76–77, 87, 174–75, cost of, 57, 162, 209–10
244–46 and creativity, 10
as end-to-end systems, 39 decentralization of, 85, 139–40
as free resources, 12–13, 20, 228 defense of, 6
privatization of, 244 in end-to-end networks, 36–37, 156
production costs of, 13 entrepreneurial, 6
value of, 87, 173 in environment of minimal control,
history, access to, 13–14 140
Hollywood, see film industry from experiments, 10
home networks, 157–58 freedom for, 11, 39, 139
Hoover, Herbert, 73, 74 free resources as basis of, 12–15, 71,
HTML (hypertext markup language), 41, 76, 91, 156
43, 57–58 incentives for, 71, 203
HTML books, 122–23, 196–99 legal control of, 4, 32, 139
HTTP (hypertext transfer protocol), 41, 43 licensing as deterrent to, 76, 216
Huber, Peter, 227–28 new and unimagined, 37
Hughes, David, 79–81, 96, 223, 226 old vs. new, 6, 199–201, 202–5, 236–39
“Hush-a-Phone” device, 30 protection of, 253–54
HyperCard, 42 relocation of, 140–42
hypertext, 42, 57 in stupid networks, 38
in technology, 5, 9–11, 48, 200
IBM: in telecommunications, 45, 176
and Apache server, 69–70, 71 tragedy of the commons in, 175–76,
and Linux, 69 177
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 344

3 4 4 I n d e x

innovation commons, 49 government-funded research for, 44,
at code layer, 58, 85, 103, 138, 167, 175 45–46
e2e and, 36–37, 40–41, 48, 58, 147, growth as open in, 37
156, 210, 238–39 HTML books on, 122–23
Internet as, 23, 26, 40–41, 48, 85, 167, as innovation commons, 23, 26, 40–41,
175, 238–39, 266 48, 85, 167, 175, 238–39, 266
Innovator’s Dilemma, The (Christensen), layers of, 23, 25
89–91, 139, 210 and mass media, 178–79, 200
Intel chip architecture, 62, 63, 66 and Microsoft competition, 65–66
Intel Corporation, 27 name space for, 172
intellectual property: network of, 26, 34, 147, 157–58
in the creative process, 203 neutral platform of, 10, 37, 91, 161,
as property, 237 167, 174, 175, 266
protection of, 204, 217 new applications of, 40, 122
intellectual property law, 7, 57, 97, 177, new demand created by, 132–34
187–88; see also copyright law; old regime threatened by, 6, 16,
patent law 199–200
Internet: on-line services of, 163
advertising on, 7 P2P networks on, 134–37
and AOL Time Warner, 7, 163, 165 pornography on, 178–79, 184, 199
architecture of, 7, 15, 16, 23, 35–37, protocols (IP) of, 35,36, 37, 41, 42, 48,
40, 44, 119, 135, 140, 210, 238, 264 122, 176
and AT&T, 154 public domain and, 56
backbone providers to, 157 search engines on, 137
birth of, 44–46, 148–49 service providers (ISPs) for, 34, 82,
blocking access to, 184 148, 157, 176, 246
building control into, 156–58 TCP/IP, 41, 56, 149, 248
and cable broadband, 153–54, 155–58, television content on, 162, 190–92
167 threats to future of, 5–6
capacity of, 46, 47, 229 trespass law and, 170
changes in, 7–8, 15, 16, 23, 25, 41, 46, unpredictability of, 7, 39–40, 88, 91
99, 135, 139, 140, 146, 156, 175, 176 vertical integration and, 165–66
code layer of, 48, 49–50, 56, 138, 161 video limits on, 156–57, 158, 166
as commons built on control layer, 176 weaknesses of, 46
as communication system, 25 and wearable computing systems,
connections made by, 7, 36, 41–42, 44, 81–82
48, 82, 130, 147–49 Web as separate from, 41
constraints on, 175, 216–17 wires of, 26, 34–37, 44–46, 79, 148–49,
copyright on, 98, 179–80, 190–92, 200, 151–53, 167
216 worldwide audience of, 181–82, 191
creativity in, 6, 9, 14, 23, 76 Internet Explorer, 41
dynamic content of, 137 inventions, see patent law; patents
early promise of, 7, 14, 121 inverse commons, 68
as end-to-end network, 34–39, 48, 149, IP (Internet protocols), 35, 36, 37, 41, 42,
156, 210 48, 122, 176
entrepreneurs of, 6 IPSec technology, 173
file sharing on, 157 IPv4 of IPv6 name space, 172
filtering on, 157, 184 Isenberg, David, 38, 89–90
free resources of, 14–15, 40, 44, 48, 50, ISPs (Internet service providers), 34, 82,
56–57, 86–99, 254 148, 157, 176, 246
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 345

I n d e x 3 4 5

Jaffe, Adam, 206, 207 Machiavelli, Niccolò, 6
Jansson, Eddy, 184 Macintosh, see Apple Computer
Jaszi, Peter, 198 McChesney, Robert, 117
Java, 65, 264–65 McPherson, Isaac, 94
jazz, 9, 13 Madison Square Garden, New York, 24, 25
Jefferson, Thomas, 94–95, 104, 206–7 Madonna, 128
jukeboxes, 109 management theory, 89–91, 92
jurisdiction, problems of, 185 Maritz, Paul, 67
markets:
Kahin, Brian, 213 as agent for control, 96, 151, 162, 165,
Kapor, Mitch, 13, 35 166
Kennard, William, 80, 223 competition in, 232
Kepler, Johannes, 13 constraints on, 114–15
Kingsbury, Nicholas C., 28 free, 71
Kingsbury Commitment, 28–29 vs. government regulation, 12, 14,
Kleinrock, Leonard, 31 75–76, 263
knowledge, commons of, 49, 55, incentives in, 71, 162
57–58 linked to commons, 138, 228
Kozinski, Alex, 203–4 networks and, 46–47, 162
Kreig, Andrew, 118 power of, 159
probable returns in, 90
Lamarr, Hedy, 79 for spectrum, 75, 228, 231,232, 242
Lange, David, 256 unidentified or undeveloped, 90
language, as nonrivalrous commons, 21 Marx, Groucho, 13
Law and Order in Cyberspace (Huber), Mattel Corporation, 184–86, 187, 188
227–28 mechanical reproduction right, 108–9
layers, 23–25 media:
content, 23 bias in, 118–19
logical (code), 23 competition in, 119
physical (wires), 23 constraints on, 216
see also Benkler, Yochai; specific content as separate in, 166
layers distribution of, 119, 234
liability rule, 110 homogeneity of output in, 118, 132
licensing: Internet and, 178–79, 200
compulsory right of, 109, 201, 255 ownership of, 116–19, 159, 173
free software, 60 real-space, 132
GPL, 44, 54, 59–60, 97, 185–86 technology and, 116, 118
innovation chilled by, 76, 216 mercantilism, 71
nature of, 59 Microsoft:
open source, 60, 67, 72, 97 code controlled by, 67, 263
and patent law, 205, 210 and competition, 28, 56, 63–66, 167,
and spectrum, 73, 223–25 267
Lightman, Alex, 81–82, 222 consent decree signed by, 64–65
Linux OS, 54–55, 56, 57, 60, 62, 69 as dominant in computer industry, 27
Linux PCs, 189 and DOS, 62–66
Litman, Jessica, 4, 258–59 government lawsuit against, 61–68,
local area networks, 77 246, 262, 264
logical layer, see code layer innovation killed by, 62
Lotus 1-2-3, 13 Internet Explorer of, 41
lyric servers, 124–26 and Mac OS, 90–91
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 346

3 4 6 I n d e x

Microsoft (cont’d): growth of, 199–200
monopoly position of, 64, 65, 66–67 on jukeboxes, 109
MSN of, 183 lyric servers in, 124–26
on open code values, 57 MP3, 123–24, 125, 192–94, 201
and software patents, 206, 214 my.mp3.com, 127–29, 133, 192
and Windows, 28, 62–66 Napster and, see Napster
Minix, 54 new forms of, 9
Mitchell, Margaret, 198–99 in noncommercial broadcasting, 109
mobile devices, connections of PCs and, OLGA, 182–83
81 ownership of the artist in, 128
modularity, 92 piano rolls, 108–9, 254
monopoly: production of, 127
antitrust laws and, 67 unregulated, 8
and AT&T, 28, 29, 30, 32, 148 my.mp3.com, 127–29, 133, 192
benefits of, 29, 30, 175
and cable television, 152, 153 Napster, 130–32, 194–96, 201, 254
centralized control of innovation in, 32 and album sales, 200
in closed societies, 250 bandwidth used by, 138
control built into, 32, 175 court shutdown of, 264, 265
as exclusive right, 58 firewalls against use of, 172–73
of a malevolent giant, 91–92 Hatch’s views on, 262–63
and Microsoft, 64, 65, 66–67 as P2P technology, 135, 194, 258
power over the community in, 87, 216 National Center for Supercomputing
state-backed, 71, 205–7, 211, 212, Applications (NCSA), 55
215–16, 242, 250 National Parks Service, 231
Moody, Glyn, 56 National Research Council (NRC),
Motion Picture Association, 185 34–35, 40, 165–66, 176
Mouse, Mickey, 107 Native American tribes, as sovereign
MP3, 123–24, 125, 192–94, 201 nations, 81
my.mp3.com, 127–29, 133, 192 NATs (network access technologies),
Napster and, 130–32, 194–95 171–72
MS-DOS, 62, 63, 64, 65 Naughton, John, 32, 33
MSN, 183 navigable waterways, 87
MUDs (multi-user domains), 163 NBC, and spectrum allocation, 74
Mueller, Milton, 28 NCSA Server, 55
Murphy, Frank, 75 Negroponte switch, 243
music: Net, see Internet
access to, 128, 131 Netscape Corporation, 55, 65
album sales, 200 Netscape Navigator, 41
availability of, 7, 13, 111 network access technologies (NATs),
breaking into, 134 171–72
capital markets in, 201 network design, principles of, 34–35
compulsory licensing rights to, 109 networks:
concentrated ownership of, 117 access to, 34, 45, 155, 171–72, 244–46
copies of, 130, 181 ATM design of, 149
copyright law applied to, 3, 192–94, built by communities, 162
254–55 closed, 162
digital composition of, 8–9, 96 collision on, 77
distribution of, 119, 123–24, 127–29, control of, 150–51, 156–58
131, 200–202 data transport provided by, 35, 36, 40
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 347

I n d e x 3 4 7

end-to-end argument for, 34–39, 156 open source licenses, 60, 67, 72, 97
Ethernet, 77–78, 231 operating systems (OS):
government monopoly of, 242 built on open code, 68, 247
home, 157–58 common platform of, 51
incentives to build, 174 DOS, 62–66
intelligence located in, 34 free, 52
linking of, 147–49 GNU, 53–54
local area, 77 GNU/Linux, 54–55, 56, 57, 60, 62, 69
market influences on, 46–47, 162 GUI, 62, 63
neutral platforms of, 37, 45, 61, 161 IBM OS/2, 28, 65
as open resource, 45, 61, 88 incompatibility of, 51
organization models for, 167 licensing of, 60
P2P, 134–137, 159, 194, 258 Mac OS, 60–61, 63, 90–91
real options theory applied to, 89 as nonrivalrous resource, 96
smart, 38 Unix, 51–52
speed of, 47 Oracle, 208
stupid, 38 Oram, Andy, 137
vertical integration in, 165–66 O’Reilly, Tim, 210–11
wires of, 26 Osterman, Jack, 32
neutral platforms: Ostrom, Elinor, 95
increasing controls vs., 175 overconsumption, prevention of, 22
of Internet, 10, 37, 91, 161, 167, 174,
175, 266 P2P networks, 134–37, 159, 194, 258
of networks, 37, 45, 61, 161 packet-switching technology, 31–34, 40,
of open code, 72, 246–49 46, 79, 157
public benefits of, 168 Parker, Sean, 130
Newton, Harry, 38 parks:
Newton, Sir Isaac, 13 as commons, 19, 20, 21, 87
Noam, Eli, 219, 226, 231 as free resource, 13, 20
noise, communities built by, 133 patches, 55–56
nonexclusive rights, 86 Patel, Marilyn Hall, 196
nonrivalrous resources: patent bar, 211–12
nature of, 21, 58, 94–95, 115–16 patent law, 205–15
need for creation of, 95, 96 as barrier to idea theft, 211
system of control for, 95–96 and content layer, 259–61
norms: damages under, 260
control via, 96, 146, 151 duration and scope of, 207–8, 212–13,
protection by, 58, 161, 167, 170 214, 216, 259
NRC (National Research Council), “holdup problem” of, 214
34–35, 40, 165–66, 176 innovation affected by, 205, 206,
209–11, 214
object code, 50 and licensing, 205, 210
old vs. new, 6, 16, 139, 145–46, 199–201, and litigation, 213
202–5, 212, 236–39, 265 progress as goal of, 205
OLGA, 182–83 reform of, 260–61
1-Click technology, 211 as state-backed monopoly, 205–7, 211,
on-line bulletin boards, 79–80 212
on-line communities, 162–63 technological competition and, 206
on-line services, 147–48, 163 Patent Office, U.S., 207, 208, 213,
open access movement, 164, 165 259–61
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 348

3 4 8 I n d e x

patents: Posner, Richard, 202–3
on algorithms, 207–8 preference matching engines, 133–34
as anticommons, 214–15 press, freedom of, 197
of AT&T, 27 Prince, The (Machiavelli), 6
on business procedures, 207, 208–9, 210 privacy, loss of, 133, 140, 213
Congress and, 207, 259 Prodigy, 147–48
costs of, 213 production:
Franklin on, 206 consumption vs., 13
Jefferson on, 94–95, 104, 206–7 costs of, 7, 8, 13–14, 124, 203
limited rights of, 71 incentives for, 14, 21, 71, 95, 97, 203,
nature of, 212 241
power of, 94–95 of music, 127
on software, 206, 207–8, 213, 214 Project Gutenberg, 123
Supreme Court and, 207 property:
transaction costs of, 210 control of, 22, 170–71
Patterson, L. Ray, 258 as exhaustible resource, 230
PC-DOS, 62 incentives to produce, 97, 203, 241
Peacefire.org, 184 intellectual, 7, 57, 97, 177, 187–88,
peer-to-peer (P2P) networks, 134–37, 203, 204, 217, 237
159, 194, 258 open to the public, 86, 228
Perl, 56 and prosperity, 5
permission: protection of, 20, 187, 236–37
absence of need for, 20 with public interest, 87
granting of, 11 resources as, 13
necessity of, 12 scope of, 6, 11, 94
neutral granting of, 12, 20 trespass on, 170
physical layer, 23–25 property right, exercise of, 20
access to, 138 property rule, 110
in the arts, 110–11 protocols:
changes in, 240–46 ”Bluetooth,” 81
of computer-communications 802.11b, 81
architecture, 113 HTML, 41, 43, 57
control at, 110–11, 238 HTTP, 41, 43
creativity at, 110–11 open and free, 57
free spectrum at, 86, 241–44; see also for sharing resources, 97
spectrum TCP/IP, 41, 56, 149, 248
innovation at, 103 see also IP
wires of, see wires public domain:
physics, different, 104 free content in, 20, 50, 106, 108, 110,
piano rolls, 108–9, 254 123, 177, 196, 256
Pitofsky, Robert, 166 Internet protocols in, 56
plasticity, 39 scope of, 8
Platt, Charles, 176 publicity, right of, 203
player pianos, 108–9 publishing industry, see books
poetry:
new forms of, 9 Quality of Service (QoS) solutions, 46–47
new markets for, 126
Poetry Daily, 126 radio:
policy-based routing, 156 commercials on, 74
pornography, 178–79, 184, 199 early programming on, 74
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 349

I n d e x 3 4 9

noncommercial broadcasting, 109 free vs. controlled, 11–15, 22, 44, 47,
smart, 78 48, 69, 71, 86–99, 156, 228
software-defined, 242 incentive for production of, 14, 71, 95
Radio Act (1912), 73 “in common,” 19–21, 86–99
Radio Act (1927), 73 privatization of, 227, 244
radio spectrum, 218–26 production costs of, 13
allocation of, 74, 218–21, 227, 229, 233 protocols for sharing of, 97
amateur use of, 225 real cost of, 120
AM stations, 242 recreational, 20
community-owned, 223–24 rivalrous vs. nonrivalrous, 21, 87,
concentrated ownership of, 117, 159 95–97, 115–16
FCC and, 74, 218–21, 223–27, 244, 246 scarcity of, 75
FM stations, 223–25, 232 state vs. market control of, 12, 75–76,
and freedom of speech, 224 228
nature of, 73–75 unfair capture of, 87
as property, 225–26, 229 unknown applications of, 88–89
social value in, 227 unused, 230
technical interference with, 225 value of, 87–88
transmission on, 79 RFC 1958, 36–37, 40
see also spectrum Rheingold, Howard, 136
radio waves, 73 RIAA (Recording Industry Association of
Randall, Alice, 198–99, 202 America), 130, 192–96, 200, 215
Rand Corporation, 26 Ritchie, Dennis, 51–52, 55
rap music, 9 rivalrous resources:
Raymond, Eric, 68 depletion of, 21–22
RCA, 234 nature of, 21, 95, 115–16
Real Audio, 127 need for control of, 95, 115
real options theory, 89 Road Runner, 154
real space, 15 roads, see highways
barriers to innovation in, 120, 138 Robertson, Michael, 127–28, 133, 201
vs. cyberspace, 103–4, 121, 181, 182 Rose, Carol, 13, 20, 86, 87–88, 96
friction of, 181, 183 Rosen, Hilary, 200, 215
law of, 181, 199 RSA Labs, 136
media in, 132 Rubinfeld, Daniel, 166
Recording Industry Association of America rules, control via, 151
(RIAA), 130, 192–96, 200, 215
recordings, see music Saltzer, Jerome, 34, 37, 38, 156
Red Hat, 213 satellite TV, 153
Reed David P., 34, 35, 37, 38, 39–40, 77, scarcity:
90 and control, 47
regulator, and target for influence, 74 economics of, 103, 115
relativity, Einstein’s theory of, 12, 14, 20, 21 search engines, 137
Reno v. ACLU, 179, 180, 265 “sendmail,” 56
resources: servers:
access to, 13–14 centralized, 131
clear use of, 89 competition among, 82
community relationships to, 21 open code projects, 56
depletion of, 21–22, 95, 96–97 patches to, 55–56
as foundation to participation in restrictions on, 157
society, 93 use of term, 55
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 350

3 5 0 I n d e x

servers (cont’d): licensing of, 73, 223–25
Web, 41, 56, 134 managing policies of, 232
see also ISPs market for, 75, 228, 231, 232, 242
SETI (Search for Extraterrestrial nature of, 73–75
Intelligence), 135 as nonrivalrous resource, 230, 231
Shapiro, Carl, 214 optimal use of, 231
Shavell, Stephen, 205 overlay of, 76
Simpsons, The, 180, 181, 182 pollution of, 242–43
Singer, Hal, 166 as property, 221, 242
Skala, Matthew, 184 radio, see radio spectrum
Smith, Adam, 71 as rivalrous resource, 230–31
software: shared resources of, 76, 78–79, 96,
on business methods, 208–9 219–20, 230–31
code of, 35 spread, 76, 80
commons of, 49, 69, 85 sunk cost in, 232
content protection in, 11, 59, 252–53 tragedy of the commons in, 229
copyright of, 252–53 unlicensed, 82–83, 226, 244
development of, 70 unused, 230
Free Software Foundation, 12, 53–54, wireless, 82, 154, 161–62
59 speech, freedom of, 116, 140, 178–79,
host “IMP,” 36 184, 197–98, 224
licensing of, 59–60 spiders, 168–71
open code of, 51, 59, 68–69 spreadsheet, 13
patents on, 206, 207–8, 213, 214 spread spectrum, 76, 80
restrictions on copying of, 59 Sprint, 38
shrink-wrapped licenses with, 185–86 Stallman, Richard, 12, 52–54, 59, 69,
source carried with, 53, 55 208, 213, 267
Somers, Daniel, 158 Starbucks coffee shops, 70
Sony, 195, 196 Story, Joseph, 198
source code, 50, 60, 67, 72 storytelling, creativity in, 8, 9
Soviet Union, collapse of, 145–46 streaming technologies, 127, 137
Speakers’ Corner, London, 24, 25 streaming video, 156–57, 158, 159, 190–91
spectrum, 73–84, 218–33 subsidiarity, 121
access to, 80–83, 161–62, 232–33, 241, Supreme Court, U.S.:
244 on copyright law, 105–6, 108, 109,
allocation of, 76–77 195–96, 197
auction of, 75, 225–27, 231–32 on patent law, 207
bottom-up sharing of, 78–79 on sovereignty of Native American
as commons, 83, 84, 86, 221–22, tribes, 81
228–30, 241–44 on spectrum regulation, 74–75, 255
competition in use of, 222, 232, 245 Sykes, Alan, 192
congestion in, 241 synchronicity, 10
cost of, 231
digital processors on, 79 Tannenbaum, Andrew, 54
DOCSIS standard for, 153–54, 161–62 TCP/IP, 41, 56, 149, 248
dual strategy for, 242 technology:
government control of, 73–76, 79–81, ”ah-ha,” 130
83, 232, 246, 255 balance of control vs. freedom in, 99,
government hoarding of, 242 201–2
innovation in use of, 84 changes in, 176
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 351

I n d e x 3 5 1

competition in, 206, 214 modems connected to, 45
compression, 123 packet-switching technology for,
control of, 232, 234 31–34, 46
controls imposed via, 96–97, 140 patents of, 27
costs of, 113 security of, 26–27, 31–34
digital, 7–9 selective interconnection in, 28, 29
disruptive change of, 89–90, 91–92, universal service in, 28
146, 175 wires of, 147, 148–49, 152, 154–55
distributed, 136 television:
experimentation in, 10 and AOL Time Warner, 7
falling costs of, 7, 8 cable, see cable television
innovation in, 5, 9–11, 48, 200 competition and, 132
for local control, 192 freedoms in, 111, 112
and markets, 114–15 and iCraveTV, 190–92
and media outlets, 116, 118 and Internet content, 162, 190–92
new, congressional dealings with, pay-per-view, 153
108–10, 190 satellite TV, 153
possible but not profitable, 48 as spectrum guzzler, 243
potential for, 10, 263 and VCR, 195
probable returns of, 90 wires of, 152
progress in, 5 Thompson, Ken, 51–52, 55
protection by, 256 Time Warner, AOL merger with, 7,
retrofitting by, 139 164–66, 267
scarcity of, 47 Titanic, 73
streaming, 127, 137 Tonga, high-speed Internet access in,
unpredictable future uses of, 39 80–81
wireless, 82 Torvalds, Linus, 54–55, 57, 62
see also specific technology town square, as commons, 87
telecommunications: trademark law, 204–5
alternative systems of, 31–34 tragedy of the commons, 22, 23, 48, 168
and Communications Act, 155 in innovation, 175–76, 177
end-to-end network in, 149–51 in spectrum, 229
government intervention in, 27, 45, 81, transaction costs, 210
148, 161, 221 transistor, invention of, 29
innovation in, 45, 176 transparency, nature of, 71
research in, 29–30 trespass:
wires of, 26–34, 45–48, 88, 147, by bots, 169–70
148–49, 152, 154–55, 163–64 classic definition of, 170
telephone system: Turner, Ted, 152
AT&T, 24, 26–34, 44, 62 Twelve Monkeys (film), 4
Bell system of, 27 2001, 196
and birth of the Internet, 44–46,
148–49 uncertainty:
circuit-switching technology for, 33 and free vs. controlled resources,
competition in, 27, 28–29, 32–33, 158 88–89
DSL in, 154–55, 248 plasticity and, 39
independents in, 28 Uniform Computer Information
innovation in, 30 Transactions Act (UCITA), 257
Internet wired to, 44–48 United States v. Microsoft, 61–68, 246,
layers of, 24, 25 262, 264
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 352

3 5 2 I n d e x

Unix operating system, 51–52 neutrality of, 149
BSD Unix, 53, 67 as physical layer, 23–25
command line interface of, 62 as property, 241
as commons, 52, 54 Quality of Service solutions, 46–47
GNU and, 53–54 as rivalrous resource, 96
licensing of, 53 sharing of, 33
USENET, 96 of telecommunications system, 26–34,
45–48, 88, 147, 148–49, 152,
Vaidhyanathan, Siva, 202
154–55, 163–64
Valenti, Jack, 195
and voice call vs. Internet use, 148
value, dimensions of, 93
woods, cathal, 183
VCR, 195
WordPerfect, 262
Verser, Rocke, 136
World Wide Web:
vertical integration, 165–66, 176
auction sites on, 169
video:
audience of, 181–82, 191
development of, 234
bots of, 169–71, 182
digital forms of, 96
browsers on, 41, 43, 134
limits on, 156–57, 158, 166
code layer of, 57, 85
streaming, 156–57, 158, 159, 190–91
core protocols for, 56
VCR, 195
denial of service attack on, 169
VisiCalc, 13
emergence of, 41–44, 134
Vixie, Paul, 56
as fantasy, 7
volunteerism, essential, 56
HTML as mark-up language of, 57–58
Walker, Jay, 212, 267 HTML books on, 122–23
Wall, Larry, 56 indexing of, 168–69
Walzer, Michael, 93 interconnectivity in, 43
Wayner, Peter, 60, 68, 213 Internet as separate from, 41
Web, see World Wide Web many-to-many publishing of, 88
WebSphere, 70 open architecture of, 169, 170
Western Union, 29 as “out of control,” 37
Wheel of Fortune, 203 patent law and, 213
White, Vanna, 203–4 pornography on, 184
wideband technologies, 76, 78 potential for, 182
Wind Done Gone, The (Randall), 198–99 privacy on, 213
Windows, 28, 62–66, 189 programming language for, 56
wired, use of term, 49, 178 search engines of, 169–71, 182
Wired magazine, 49 spiders in, 168–71
wireless, see spectrum as universal resource, 37, 44
wires, 26–48 Web servers on, 41, 56, 134
of AOL, 163–64 writers:
constraints from, 47 as authors, 111, 204
control of, 26, 147–76 royalty expenses of, 203
end-to-end design, 34–44, 46, 88, 156 Wu, Tim, 47–48
fiber optic cable, 245 WWW, see World Wide Web
government-created commons on, 45,
155 Xerox PARC, 235–36
of Internet, 26, 34–37, 44–46, 79, Yahoo!, 28, 167, 169, 183
148–49, 151–53, 167 Young, Bob, 213
monopoly control of, 32
network connections on, 26 Zittrain, Jonathan, 129, 130
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 353

A b o u t t h e A u t h o r

Lawrence Lessig is a professor of law at the Stanford
Law School. Previously Berkman Professor of Law at
Harvard Law School from 1997 to 2000 and professor at
the University of Chicago Law School from 1991 to
1997, he is a graduate of the University of Pennsylvania,
Trinity College, Cambridge, and Yale Law School. He
clerked for Judge Richard Posner on the Seventh Circuit
Court of Appeals and Justice Antonin Scalia on the
United States Supreme Court. He is a monthly colum-
nist for The Industry Standard, a board member of the
Red Hat Center for Open Source, and the author of
Code and Other Laws of Cyberspace.
less_0375505784_4p_01_r1.qxd 9/21/01 13:50 Page 354