Plaintext
VOLUME 53 | 2008/09
JAMES GRIMMELMANN
The Google Dilemma
ABOUT THE AUTHOR: James Grimmelmann is an associate professor of law at New York Law School.
This essay is adapted from talks given at the Horace Mann School on January 30, 2008 and at the New York
Law School Faculty Presentation Day on April 2, 2008. Ken Liu and Frank Pasquale provided comments on
earlier drafts. Shaliz Sadig provided research assistance. Unless otherwise specified, all URLs were last
visited November 26, 2008. This essay may be freely reused under the Creative Commons Attribution 3.0
United States license, http://creativecommons.org/licenses/by/3.0/us/.
939
Web search is critical to our ability to use the Internet. Whoever controls search
engines has enormous influence on us all. They can shape what we read, who we
listen to, and who gets heard. Whoever controls the search engines, perhaps, controls
the Internet itself. Today, no one comes closer to controlling search than Google
does.
In this short essay, I’ll describe a few of the ways that individuals, companies,
and even governments have tried to shape Google’s results to serve their goals.
Specifically, I’ll tell the stories of five Google queries, each of which illustrates a
different aspect of the problems that Google and other search engines must confront:
“mongolian gerbils” shows their power to organize the Internet for us; “talentless
hack” shows how their rankings depend on collective human knowledge; “jew” shows
why search results can be controversial; “search king” shows the tension between
automatic algorithms and human oversight; and “tiananmen” shows how deeply
political a search can be. Taken together, these five stories provide a snapshot of
search and the interlocking issues that search law must confront.
I. SEARCH: “MONGOLIAN GERBILS”
Suppose you wanted to learn about Mongolian gerbils. Where would you look?
There are decent online sources of information on them, such as Peter Maas’s The
Mongolian Gerbil Website.1 But how would you find them? There’s no obvious central
authority that someone who didn’t already know about gerbils would know to consult.
Maas’s site is at the not-easily-guessed URL http://www.petermaas.nl/gerbils/
english.htm, whereas mongoliangerbils.com is an unhelpful, ad-laden page offering
“Cheap Stuff for Hamsters” and “Gerbil Ringtones.”
If you looked for gerbil information by clicking randomly from page to page,
matters would be even worse. If you started clicking now, reading a random new
web page every second, you’d get through a few thousand pages in an hour, a few
million pages in a month, and a few billion pages in a century. That may sound like
a lot, but it pales in comparison to the size of the Internet. The Internet isn’t infinite,
not even close; but it’s still immense on a scale that’s hard to imagine. Estimates
vary, but one study claims that the world’s information production in 2006 was 161
exabytes.2 That’s over 1,000,000,000,000,000,000,000 individual bits of information,
a number so large as to be a meaningless abstraction.
As more and more of this growing flood of information finds its way online, the
Internet itself is becoming absurdly huge. One study claims the Internet has 168
million web sites, 3 and Google claims to have indexed over a trillion individual web
pages.4 Simply put, no one knows how insanely large the Internet is.
1. Peter Maas, The Mongolian Gerbil Website, http://www.petermaas.nl/gerbils/english.htm.
2. Frederick Lane, IDC: World Created 161 Billion Gigs of Data in 2006, Top Tech News, Mar. 7, 2007,
http://www.toptechnews.com/story.xhtml?story_id=01300000E3D0; see also Peter Lyman & Hal R.
Varian, How Much Information? 2003 (Oct. 27, 2003) (unpublished manuscript), available at http://
www2.sims.berkeley.edu/research/projects/how-much-info-2003/.
3. May 2008 Web Server Survey, Netcraft (May 6, 2008), http://news.netcraft.com/archives/2008/05/06/
may_2008_web_server_survey.html.
4. See Jesse Alpert & Nissan Hajaj, We Knew the Web Was Big, Google Blog, July 25, 2008, http://
googleblog.blogspot.com/2008/07/we-knew-web-was-big.html.
940
NEW YORK LAW SCHOOL LAW REVIEW VOLUME 53 | 2008/09
It’s also utterly disorganized.5 Those billions of web pages aren’t conveniently
sorted into precise categories. People pretty much just throw stuff up online as the
mood strikes them, linking to whatever they feel like, in no particular order. Every
corner of the Web is equally disorganized. Your random-click quest for gerbil
information is like looking for a needle in a field full of haystacks.
The reason that we think of the Internet not as a chaotic wasteland, but as a
vibrant, accessible place, is that some very smart people have done an exceedingly
good job of organizing it. The Internet today is usable because of search engines. A
good search engine is, in effect, a card catalog for an infinite library. You say to
Google, “I want information on Mongolian gerbils,” click, and it points you to exactly
the page you’re looking for.6
II. SEARCH: “TALENTLESS HACK”
In 2001, a college student named Adam Mathes noticed something interesting
about Google, almost accidentally.7 One of his friends, Ben Brown, had started
calling himself an “Internet rockstar.”8 Mathes noticed that even though Brown was
the number-one Google hit if you searched for “Internet rockstar,”9 that phrase never
appeared in the text of Brown’s web page itself.10 It only showed up in the links
other people made to Brown’s page, i.e., they’d describe him as an “Internet rockstar”
and link to him.11 Thus, Mathes reasoned, Google must figure out what a page is
about by looking at the pages that link to it, since there’s no way it could have learned
that Ben Brown was an Internet rockstar by reading only benbrown.com.
In fact, that’s exactly how Google operates. The genius of Google is that its
creators didn’t come up with a great organizational scheme for the web. Instead, they
got everyone else to do it for them. The heart of Google’s system—called
PageRank—is that it looks at who links to whom online.12 Every time you create a
link to another web page, you’re in effect telling the world that the web page has on
it something important, or interesting, or useful, or funny, or whatever matters to
you. Your link is a kind of vote; you want people to pay more attention to that page.
Very loosely put, Google goes around the web, counting links. Pages with more
links pointing at them have been “voted up” more often, so they must be more
5. See James Grimmelmann, Information Policy for the Library of Babel, 3 J. Bus. & Tech. L. 29, 38 (2008).
6. Google search: mongolian gerbils, http://www.google.com/search?q=mongolian+gerbils.
7. Adam Mathes, Filler Friday: Google Bombing, Über, Apr. 6, 2001, http://uber.nu/2001/04/06/.
8. Ben Brown, Internet Rockstar, http://benbrown.com/; Wikipedia: internet rockstar, http://en.wikipedia.
org/wiki/Internet_Rockstar.
9. See Google search: internet rockstar, http://www.google.com/search?q=internet+rockstar.
10. Though Brown’s web page now contains that phrase in text visible to search engines, it didn’t in 2001.
See Ben Brown, Writer, New Zealand, http://web.archive.org/web/20010401034517/http://benbrown.
com/.
11. Mathes, supra note 7.
12. See Sergey Brin & Lawrence Page, The Anatomy of a Large-Scale Hypertextual Web Search Engine
(1998) (unpublished manuscript), available at http://infolab.stanford.edu/~backrub/google.html.
941
THE GOOGLE DILEMMA
important, and therefore Google displays them higher in its results. What’s more,
important pages—those with lots of inbound links and therefore lots of votes—are
presumably also trusted and influential. Thus, Google counts their own outbound
links as being “worth” more.
The math involved is at once elegantly simple and extremely hard to compute.
Every page’s rank potentially depends on every other page on the Internet. To run a
serious search engine, you need a huge server farm to calculate all of those
interdependencies. Google has over 450,000 servers; that’s millions of dollars a
month in electricity bills alone.13 Its major competitors have similarly extensive
computational infrastructure. That’s one reason that the search engine landscape is
dominated by just a few companies—Google, Microsoft, and Yahoo! together account
for four out of every five searches that Americans run online.14 Of them, Google is
the undisputed champion, with an absolute majority of all searches.15 And since
Americans run something like ten billion searches a month, even Google’s 60% of
that is still a very big number.16
What Mathes realized was that Google used links not just to learn how important
a page was, but also what it was about. Each link to benbrown.com using the magic
phrase was saying two things: both “benbrown.com is important,” and
“benbrown.com is about an Internet rockstar.” The more pages that link using a
given phrase, the more Google will think that the phrase accurately describes the
page. But that would mean that Google could be tricked; all you would have to do is
get a lot of friends to create links using particular words. Thus, Mathes wrote a blog
post asking everyone to link to the home page of another friend, Andy Pressman,
using the phrase “talentless hack.”17 Within a few weeks, Pressman had zoomed to
the number-one hit if you searched for “talentless hack.” Mathes called the technique
Googlebombing.18
It started as a joke, but Googlebombing has been used to send some serious
messages. So many people use search engines, and their results appear to be so
sophisticated, that any message you can sneak through in the form of a Googlebomb
has at least a veneer of legitimacy. Thus, the most famous Googlebomb of all time is
probably the one many Democrats launched in 2003 to link to George W. Bush’s
official biography page with the phrase “miserable failure.”19 This is a significant
new form of politicking. Land a bomb like this and you can convince the world that
13. David F. Carr, How Google Works, Baseline, July 6, 2006, http://www.baselinemag.com/c/a/
Infrastructure/How-Google-Works-1/.
14. Press Release, comScore, comScore Releases May 2008 U.S. Search Engine Rankings (June 19, 2008),
http://www.comscore.com/press/release.asp?press=2275.
15. Id.
16. Id.
17. Mathes, supra note 7.
18. Id.; Wikipedia: google bomb, http://en.wikipedia.org/wiki/Google_bomb.
19. Google search: miserable failure, http://www.google.com/search?q=miserable+failure; Wikipedia:
google bomb, http://en.wikipedia.org/wiki/Google_bomb.
942
NEW YORK LAW SCHOOL LAW REVIEW VOLUME 53 | 2008/09
Google agrees with your position.20 A successful Googlebomb doesn’t just reflect the
consensus of web users; it can help construct that consensus.
III. SEARCH: “JEW”
Still sounding a bit frivolous? Consider a more serious example: In 2004, the
number-one Google hit for a search on “jew” was jewwatch.com, which describes
itself as “An Oasis of News for Americans Who Presently Endure the Hateful
Censorship of Zionist Occupation” and features a wide assortment of paranoid anti-
semitic content.21 Upset at this placement, a Jewish activist asked people around the
web to link the word “jew” instead to a Wikipedia article. 22 This touched off a
counter-Googlebomb as some neo-Nazi sites tried to point the term back to Jew
Watch, but ultimately attempts to make “jew” point to the comparatively neutral
Wikipedia page was successful.23
There’d been Googlebombs before and there’d be Googlebombs after, but this
one did something that no other had: it drew a response from Google. At the top of
the Google results page for “jew” is a box reading, “Offensive Search Results. We’re
disturbed about these results as well.”24 The same campaign that produced the Jew
Googlebomb also included a petition to have Google remove Jew Watch entirely, so
that no Google search—and certainly not on the term “jew”—would ever return it.25
The activists certainly had a point. Google could easily have changed their software
so no trace of Jew Watch remained in its results pages, no indication that anything
other than the usual process of looking for relevant results had ever taken place.
But that’s not what Google did. Instead, it left the results alone—that is, left it
up to the back-and-forth push-pull struggle of the Googlebombs and counter-bombs;
may he who gets the most links win. Google was insistent that it wouldn’t intervene
in the results.26 The reason is a feature of the company’s corporate culture that goes
even deeper than its “Don’t be evil” motto. Google deeply believes in throwing huge
automated processes at a problem. Contrast that attitude with Yahoo!, whose
hierarchical index of the Internet is maintained by editors who catalogue and
20. See Ryan Moulton & Kendra Carattini, A Quick Word About Googlebombs, Google Webmaster
Central Blog, Jan. 25, 2007, http://googlewebmastercentral.blogspot.com/2007/01/quick-word-
about-googlebombs.html.
21. See Jew Watch, http://jewwatch.com/. See generally Wikipedia: Jew Watch, http://en.wikipedia.org/
wiki/Jew_Watch.
22. John Brandon, Dropping the Bomb on Google, Wired, May 11, 2004, http://www.wired.com/culture/
lifestyle/news/2004/05/63380.
23. See Judit Bar-Ilan, Google Bombing from a Time Perspective, J. Computer-Mediated Comm., Apr.
2007, http://jcmc.indiana.edu/vol12/issue3/bar-ilan.html.
24. Google search: jew, http://www.google.com/search?q=jew.
25. See Brandon, supra note 22.
26. Id.
943
THE GOOGLE DILEMMA
categorize web sites by hand.27 That’s not how Google rolls. Google just feeds the
Internet into its server farm and waits for a result to emerge, unsullied by mere
mortals.
Thus, the “Offensive Search Results” page that Google put up to explain why
“jew” takes you to Jew Watch is Google’s way of saying “don’t blame us, the computers
did it.” 28 It’s a point of pride at Google that the programmers stay out of politics and
keep their personal views out of the search engine: “The beliefs and preferences of
those who work at Google, as well as the opinions of the general public, do not
determine or impact our search results.”29
Is that really true? Is it really the case that search engine results are purely
automated, impersonal things that don’t reflect anyone’s opinion at all? In one sense,
passing the buck and saying “don’t blame us, the computers did it” is an uncomfortable
position for any computer programmer to take. Who, after all, gave the computer its
instructions? The programmer did. Everything that Google’s automated ranking
system does, it does because Google programmers told it to. A computer is just a
glorified abacus; it does what you tell it to.
Google’s programmers may not have specifically told its computers to rank Jew
Watch highly and to call George Bush a miserable failure. But they did feed into the
computers a list of criteria—each link from a weblog is worth so many points, pages
less than a week old get a 10% newness bonus, whatever—and those criteria lead
directly to the conclusion that Andy Pressman is a talentless hack. Google could
have changed those criteria, and if they had, he’d be a talented hack or a miserable
failure or an Internet rockstar.
And, of course, the “beliefs and preferences” of Google’s employees and users do
enter into its search results in another sense. The employees prefer that Google
return results that the users believe to be useful. They optimize their algorithms all
the time to make the results more relevant to their users’ questions. They don’t want
you to get Jew Watch if you search for “mongolian gerbils.”
This isn’t to say that what Google is doing is right or wrong, just that the
distinction it makes between objective “automatic” results from the computer and
subjective “beliefs and preferences” doesn’t really hold up.30 This tension is a sign
that Google and its employees are uncomfortable with the massive power they have
to shape online opinion. They’re searching, as it were, for principled ways to use that
power responsibly.
Call this problem the Google Dilemma. Google has the ability to shape its
search results to prefer some web sites over others. Indeed, whatever choice it
makes—alphabetical, by link count, by politics, by whatever—will result in some
27. See Yahoo!, What Are Directory Categories and How Are They Organized?, http://help.yahoo.com/l/
us/yahoo/directory/ctd/ctd-07.html.
28. See Google, An Explanation of Our Search Results, http://www.google.com/explanation.html.
29. Id.
30. See James Grimmelmann, The Structure of Search Engine Law, 93 Iowa L. Rev. 1, 58–59 (2007).
944
NEW YORK LAW SCHOOL LAW REVIEW VOLUME 53 | 2008/09
sites being on the first page of results and others not.31 So you’re Google; what
choices do you make? Do you make George Bush a mongolian gerbil? Presumably
not. Do you make Andy Pressman a talentless hack? Perhaps; it seems harmless
enough. What do you do with “jew”? Harder to say.
IV. SEARCH: “SEARCH KING”
Google isn’t the only one facing the Google Dilemma. Lawyers, judges, and
legislators deal with it, too. There have been dozens of lawsuits against search
engines. Each of those lawsuits requires the court hearing it to confront the
Dilemma.32 In order to say that a search engine did something wrong, there must be
a plausible alternative—something it could have done that was right. You can’t say
that a search engine was irresponsible unless you have a baseline of what would be
responsible.
Consider my favorite search engine lawsuit of all time. An Oklahoma web
company called Search King had a web site that was highly ranked for certain queries
on Google.33 It had “PageRank 8”: on Google’s 0–10 scale of how important a page
is, Search King was an 8. That high ranking meant thousands of people came to
Search King and its clients—local Oklahoma businesses like f lorists and car
salesmen.
Then, one day in late summer of 2002, Google utterly destroyed Search King’s
business. Search King went from PageRank 8 to PageRank 4 overnight. In the
Google economy, that kind of drop is a disaster. Suddenly, instead of showing up
high on the first page of Google results, Search King and its clients were lucky to
show up in the first few pages. For a local business dependent on Internet orders,
that’s the kiss of death. Think about sending flowers to someone in Oklahoma City,
if you’re not a local. Since you don’t know any florists there, you’re likely either to go
with a national chain or to look for a florist through a search engine. If Bob’s Flowers
doesn’t show up when you search, it might as well not exist. Too bad for Bob.
Businesses had gone up and down in Google’s rankings before, but Search King
made history by being the first to sue Google over it. Search King claimed, in
essence, that Google was libeling it. Demoting Search King to PageRank 4 was the
equivalent of telling Google’s users that Search King was unimportant or a bad
source of information. According to Search King, what Google had done was the
equivalent of standing outside of Bob’s Flowers saying, “Don’t shop here. They’re
terrible. You don’t want to know what they did to my sister’s petunias. And that ax
murderer they have working in the back room? Tsk. Tsk.” The way Search King saw
it, Google’s computers had calculated that searchking.com was PageRank 8, but then
31. See Eric Goldman, Search Engine Bias and the Demise of Search Engine Utopianism, 8 Yale J.L. & Tech.
188, 189–92 (2006). But see Lucas D. Introna & Helen Nissenbaum, Shaping the Web: Why the Politics of
Search Engines Matters, 16 Info. Soc’y 169, 174–75 (2000).
32. See Grimmelmann, supra note 30, at 15–51 (taxonomizing search engine lawsuits).
33. Search King, Inc. v. Google Tech., Inc., No. CIV-02-1457-M, 2003 U.S. Dist. LEXIS 27193 (W.D.
Okla. May 27, 2003).
945
THE GOOGLE DILEMMA
some evil mastermind at Google with a grudge against Search King had gone in and
changed the number to PageRank 4.34
One might expect that Google’s reply when confronted about “search king” would
be the same reply it gave when confronted about “jew”: “Don’t blame us, the computers
did it.” Instead, Google gave an intriguingly different answer. Buried beneath the
legalese, Google said something more like, “We’re not admitting to anything, but if
by some chance we were to have hand-tweaked the results a bit to punish Search
King, then we would have had a good reason to.”35
Consider the possibility that Search King wasn’t quite the innocent Internet
entrepreneur it made itself out to be. Some observers have claimed that Search King
was in fact running a large and cynical version of a Googlebomb called a “link
farm.”36 A link farm operator has decided that getting thousands of people to link to
her to raise her PageRank is just too much work. Instead, she creates those links
herself. They’re designed to look enough like real web content to trick a search
engine, but consist mostly of links to other pages in the link farm itself. It’s a web
form of spam. If she creates enough fake links to look as though tons of people have
suddenly decided this, then her previously-obscure site is now the world’s foremost
authority on mongolian gerbils. Search King’s link farm had its clients link to each
other. Bob’s Flowers would link to Joe’s Used Cars, and vice-versa, even though they
had no connection whatsoever besides both being clients of Search King. They were
paying Search King, in essence, to help them conspire to boost their PageRanks.
Web users, on the other hand, are generally unhappy if their searches lead them
to link farms: the informational food they grow is intended for search engine spiders,
not for people. When you search for a florist in Oklahoma City, you want to see the
florists that other web users liked. You probably don’t care very much whether the
f lorists were good at creating fake web pages or at exchanging irrelevant links.
Search engines take this preference to heart and try to keep link farms out of their
results. Google’s guidelines for web site creators specifically prohibit link farming
and link exchanges of the sort Search King was engaging in, and when Google
realizes someone’s been running a link farm, it retaliates and demotes the link farm
in its rankings.37
As it turned out, Google won the Search King lawsuit. The court held that
Google’s choice of how to rank its results was a form of speech protected by the First
Amendment.38 Other courts, facing similar lawsuits, have uniformly sided with the
34. See Frank Pasquale, Rankings, Reductionism, and Responsibility, 54 Clev. St. L. Rev. 115, 126–28
(2006).
35. See James Grimmelmann, Google Replies to SearchKing Lawsuit, LawMeme, Jan. 9, 2003, http://research.
yale.edu/lawmeme/modules.php?name=News&file=article&sid=807.
36. kpaul, SearchKing Sues Google, Kurohin, Oct. 19, 2002, http://w w w.kuro5hin.org/
story/2002/10/19/153131/23.
37. Webmaster Guidelines, Google Webmaster Help Center, http://www.google.com/support/
webmasters/bin/answer.py?hl=en&answer=35769.
38. Search King, 2003 U.S. Dist. LEXIS 27193, at *11–12.
946
NEW YORK LAW SCHOOL LAW REVIEW VOLUME 53 | 2008/09
search engines on similar grounds. 39 This seems like the right result; if search
engines didn’t have a fairly free hand to demote sites in their rankings, the Internet
would be completely overrun with junk from link farms. Still, the First Amendment
reasoning is worrisome. If taken at face value, it would suggest that the First
Amendment provides an absolute shield for search engines to re-rank sites at will, for
any reason at all, even a deliberately malicious reason. What’s more, it would suggest
that the government is powerless to make the search engines stop, even in cases
where the re-ranking was obviously malicious and dishonest.40
V. SEARCH: “TIANANMEN”
Evidence from around the world, however, suggests that governments are far
from powerless when it comes to controlling search engines. In Europe, for example,
many countries have strong laws against hate speech. The German criminal code
bans “Volksverhetzung,” which translates to “incitement of the people” or “agitation
of the populace.”41 Specifically, it’s illegal to incite hatred against segments of the
population or to call for violence against minorities. In the United States, these
forms of hate speech are usually protected from government restriction by the First
Amendment, but the legacy of Nazism and the Holocaust has made many European
countries very nervous about ethnic hatred.
The result is that American companies—coming from a country with a very
robust tradition of anything-goes free speech—are often surprised when they
encounter European speech-restricting laws that would be unconstitutional back
home. American Internet companies—which have also at times assumed that the
Internet is somehow beyond any government’s power to regulate—are often doubly
surprised when European governments tell them to clean up their act.42 In 2000, a
French court ordered Yahoo! to remove from its auction site any auction listings for
Nazi memorabilia.43 Yahoo! fought the order for a while, but then realized it would
rather comply than shut down its entire French operation.44 It was a precedent that
most other multinational Internet companies have followed.45 The result is that
most of their European sites are, let us be honest here, censored in various ways.46
39. See, e.g., KinderStart.com, LLC v. Google, Inc., No. C 06-2057 JF (RS), 2007 U.S. Dist. LEXIS
22648, at *13–15 (N.D. Cal. Mar. 16, 2007).
40. But see Frank Pasquale, Asterisk Revisited: Debating a Right of Reply on Search Results, 3 J. Bus. & Tech.
L. 61, 83–84 (2008) (claiming First Amendment does not prevent regulation of such “speech”).
41. See James Q. Whitman, Enforcing Civility and Respect: Three Societies, 109 Yale L.J. 1279, 1337–39
(2000).
42. Jack Goldsmith & Tim Wu, Who Controls the Internet?: Illusions of a Borderless world
129–45 (2006).
43. See Yahoo!, Inc. v. La Ligue Contre Le Racisme Et l’Antisemitisme, 169 F. Supp. 2d 1181, 1184–85
(N.D. Cal. 2001), rev’d, 379 F.3d 1120 (9th Cir. 2004).
44. Goldsmith & Wu, supra note 42, at 5.
45. Id. at 8.
46. See Joel Reidenberg, Yahoo and Democracy on the Internet, 43 Jurimetrics J. 261 (2002).
947
THE GOOGLE DILEMMA
Thus, turning back to search engines, you get different results in different
countries. Let’s have a look at a familiar search from earlier. If you type “jew” into
Google’s German version, google.de, guess what site doesn’t show up at all? Jew
Watch.47 Perhaps this is just a coincidence, a fluke of the algorithm? Actually no. If
you scroll down to the bottom of this results page, you find another very interesting
notice. It reads, “Aus Rechtsgründen hat Google 3 Ergebnis(se) von dieser Seite
entfernt,” which translates to “For legal reasons, Google has removed 3 result(s) from
this page.”
Google isn’t kidding. The results really have been removed. Even a search on
“jewwatch.com” doesn’t bring it back as a result.48 As far as Google is concerned, in
Germany, JewWatch.com doesn’t exist.49 The very thing Google said it wouldn’t do
in response to a public petition, the very thing that Search King accused Google of
doing, . . . Google does when the German government says to.
Here’s a perhaps more disturbing example. Compare a google.com image search
on “tiananmen” with the same image search performed on google.cn, the Chinese
version of Google. Below is the U.S. version of that search—prominently featuring
the famous photo of an anonymous nonviolent resister standing up to a line of
tanks.
Contrast that to the Chinese version of that search—principally photographs of
the Tiananmen itself, the Gate of Heavenly Peace that gives the square its name.
47. Google-Suche: jew, http://www.google.de/search?hl=de&q=jew.
48. Google-Suche: Jewwatch.com, http://www.google.de/search?hl=de&q=jewwatch.com.
49. This statement requires qualification. A German user looking for JewWatch could find it by starting
from google.com rather than from google.de. Sometimes countries restrict access to google.com to
prevent users from circumventing local restrictions; sometimes they don’t. See Philipp Lenssen, Google
Censorship FAQ, Google Blogoscoped, Mar. 2, 2007, http://blogoscoped.com/archive/2007-03-02-
n19.html.
948
NEW YORK LAW SCHOOL LAW REVIEW VOLUME 53 | 2008/09
Now, to be fair to Google, this page does display a version of that same dis-
claimer we just saw: “ ,” which translates,
according to Google, to the same general explanation we saw on the German results
for “jew”: “According to local laws and regulations and policies, some search results
are not displayed.”50 Thus, everything’s aboveboard in the sense that there’s
disclosure.
Still, this practice leaves a bad taste in a lot of people’s mouths. Reasonable
people disagree about hate speech; it seems plausible that you could have a well-func-
tioning democracy that does or doesn’t tolerate racial incitements to violence. But
when you start censoring information about democratic protests (Tiananmen Square),
or about minority religions (Tibetan Buddhism and Falun Gong), that crosses a line
that a lot of people who think about human rights, political freedom, and democracy
are concerned about.51 Nothing that Google, Microsoft, and Yahoo! have done has
drawn as much criticism as their decisions to do business in China on terms dictated
by the Chinese government.
If the Internet is a gigantic library, and search engines are its card catalog, then
Google has let the Chinese government throw out the cards corresponding to books
it doesn’t like. There may be sites with full and honest discussion of the June 4, 1989,
crackdown accessible on the Internet from China. But when those sites aren’t visible
in search engines, we’re back to our field full of haystacks. The truth about
Tiananmen Square, like the truth about Mongolian gerbils, may be out there. It’s
just not on Google.
50. Google Translate, http://translate.google.com/translate_t#zh-CN|en| ,
.
51. See generally Human Rights Watch, “Race to the Bottom”: Corporate Complicity in Chinese
Internet Censorship (2006), http://www.hrw.org/sites/default/files/reports/china0806webwcover.
pdf.
949
THE GOOGLE DILEMMA
Or perhaps it is. To many Chinese, the iconic image of Tiananmen Square
comes not from June 4, 1989, but from October 1, 1949, when Mao Zedong
proclaimed the founding of the People’s Republic from the Gate.52 To them, the
google.com association of “tiananmen” with protesters and tanks might seem like the
result of a gigantic Googlebomb perpetrated by the Western world—which, after all,
had the advantage of getting online more quickly. They’d say that google.cn tells
more of the truth about Tiananmen than google.com does. There are many (both in
China and the West) who’d disagree, but I hope at this point everyone would at least
agree on one thing: Google has a choice to make, one that matters a lot.
VI. CONCLUSION
Google has the power to make the truth accessible. It’s chosen to use its power
to help you learn about “mongolian gerbils” and to say that Andy Pressman is a
“talentless hack.” Its thoughts about “jew” have changed over time, as has its opinion
about “search king.” And it’s deliberately chosen not to tell Chinese users everything
it knows about “tiananmen.” We’re entitled to ask what makes these queries different
from each other, and whether Google made the right choices about them.
To review: We rely on search engines’ rankings to make the Internet useful.
Search engines produce those rankings by applying sophisticated computer algorithms
to the web pages made by ordinary users. People therefore create links to try to
change search engines’ rankings. If that doesn’t work, they ask the search engines to
change the results, or ask government to demand that search engines do so. Search
engines pride themselves on being automated, except when they aren’t. Search results
matter: to culture, to business, and to society. Every decision about search, and every
lawsuit about search, has these inescapably political questions at its heart. That is
Google’s dilemma, and ours.
52. See David Barboza, Chameleon Mao, the Face of Tiananmen Square, N.Y. Times, May 28, 2006, § 4, at
3.
950