Plaintext
SSRN Considered Harmful*
James Grimmelmann
February 26, 2007
The Social Science Research Network (SSRN) has adopted several unfortunate policies that
impair open access to scholarship. It should enable one-click download, stop requiring papers to
bear SSRN watermarks, and allow authors to point readers to other download sites. If it does not
reform, those who are serious about open access should not use SSRN.
How SSRN Works
The Social Science Research Network (SSRN) is a scholarly repository for economics,
accounting, law, and several related disciplines. It hosts papers for free download and provides
an email abstracting service. When an author posts a paper, SSRN creates for it a “download
page” containing the paper’s abstract, the author’s contact information, and links that any
Internet user can use to download the paper from SSRN. For a time, SSRN required that users
create an account and log in to download papers under many circumstances. Following protest
from open-access advocates, SSRN revised this policy. It now allows unrecognized users to
download papers, but encourages them to create accounts. SSRN counts the number of times
each abstract is viewed and each paper downloaded. These counts are used for a number of
purposes, including generating lists of the most-downloaded papers, authors, and institutions.
Papers do not “go live” immediately upon posting; they are instead reviewed by SSRN staff, a
process that takes several days and can involve edits to the abstract. The staff will occasionally
reject a paper. One ground for editing or rejection is that SSRN does not allow non-SSRN URLs
to appear in an abstract. The justifications offered vary. I have been told that the reason is
SSRN’s inability to provide download support. Another scholar has been told that the reason is
to prevent links to commercial sites, and that SSRN bans all external URLs rather than attempt to
distinguish commercial and non-commercial sites. This policy is strict, and applies even if the
URL is accompanied by an appropriate disclaimer.
The email abstracting service is partly subscription-based. Some abstract series are sponsored by
schools and journals, and contain abstracts of new papers from their faculty or published in their
pages. These series are free to subscribers. Other series are produced by professors for SSRN
itself. A subscription to all of SSRN’s topical series in one discipline costs between $35 and $90
a year, depending on the discipline; many institutions purchase site licenses for their members.
Each emailed abstract contains links to the author’s SSRN page and to the article’s download
page.
What’s Right with SSRN
*
This essay is licensed under the Creative Commons Attribution 3.0 United States License,
http://creativecommons.org/licenses/by/3.0/us/. It is available for download from multiple locations.
SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=965633
BEPress: http://works.bepress.com/cgi/viewcontent.cgi?article=1013&context=james_grimmelmann
Author’s homepage: http://james.grimmelmann.net/files/SSRNConsideredHarmful.pdf
1
The key goal here is open access to scholarship; anyone should be able to find and read all
scholarship of interest to her, for free and without difficulty. Open access helps scholars, who
depend on other scholarship in their own work. Open access also helps the public (who
subsidize scholarship by paying scholars’ salaries) by assisting the diffusion of scholarly ideas.
SSRN facilitates the reading of scholarship by enabling free download of any paper posted by its
author. It facilitates discovery of scholarship by offering search of the text of abstracts and
several kinds of browsing. Its abstracting service, while partly restricted to paying customers,
also enables the discovery of scholarship.
What’s Wrong with SSRN
Some SSRN policies have unfortunate collateral consequences for open access:
The download-page system prevents direct one-click download. I cannot give you a URL that
points directly to a PDF, only one that points to a page with a link to the PDF. This is a small
hassle for users, but completely disables automated search tools. Since SSRN itself does not
provide full-text search, this means there is no way to search the text of papers posted to SSRN, a
substantial barrier to locating scholarship. (Consider the value of full-text search in Lexis and
Westlaw, or the usefulness of Google in locating papers from just a distinctive phrase.) In
addition, the lack of direct download impedes the development of even better search and
indexing technologies. SSRN clams to be working with Google to enable search, but the results
are not yet apparent and a truly open strategy would work with all interested search engines.
SSRN now also watermarks all uploaded papers with the URL of their download pages; each of
the first two pages in the PDF is stamped with a line reading, for example, “Electronic copy of
this paper is available at: http://ssrn.com/abstract=962596.” These watermarks are ugly; they
can cause trouble for automated text-processing tools (such as screen readers for the blind); and
they distort a scholar’s preferred presentation of her research. Further, they create lock-in.
SSRN’s URL scheme may change; SSRN may fail to exist; scholars may decide that another
repository better serves their needs. The watermarks on already-downloaded copies will
continue to point to the old, obsolete URLs.
The editorial review process itself is a minor speed bump on scholarship, but the use of that
process to discourage competition is worse. Some scholars prefer to use SSRN’s abstracting
service but host their papers elsewhere (e.g. at a school or personal site, where no watermark
would be applied). By going out of its way to remove references to non-SSRN download sites,
however, SSRN leverages the popularity of its abstracting service to promote its download
service at the expense of other repositories and the scholars who would prefer to use them.
The old policy of often requiring login to download was also a serious drag on open access; it
made anonymous reading difficult or impossible. The right to read anonymously is an important
principle. Not all SSRN papers will be highly sensitive, but some will. (Consider Tim Wu’s The
World Trade Law of Internet Filtering, at http://ssrn.com/abstract=882459.) Fortunately, SSRN
has revised this policy.
2
Three broader concerns link these specific problems. First, they suggest that SSRN often does
not fully understand open-access principles or appropriately value them. A sincere institutional
commitment to open access is critical for an institution that seeks to be a canonical scholarly
repository. Second, they involve anti-competitive behavior that seeks to channel all access to
scholarship through services under SSRN’s sole control. This desire for centralization runs
counter to basic lessons of open access, which favor broad and redundant distribution, and a
healthy ecology of diverse institutions and tools. Third, these measures all cost effort and money
to implement. We, the scholarly community, are paying to have our access unnecessarily
restricted by SSRN.
A Few Objections Answered
Aren’t these complaints small beer?
It is true that my objections to SSRN are individually small on the scale of open-access concerns.
But they are hardly insignificant. The lack of full-text search alone puts a significant brake on
the discovery of relevant scholarship, and many scholars seriously object to the watermarks. It is
also telling that it took significant lobbying to convince SSRN to make login consistently
unnecessary.
Doesn’t SSRN have the legal right to do these things?
Of course it does, but the point is what SSRN ought to do, and how we ought to respond when it
does not. To flip matters, SSRN certainly has the legal right to do better by open access. If it
does not, we also have the legal right to take our business elsewhere.
Isn’t SSRN a business that should rationally act in its owners’ interests?
Given SSRN’s stated commitments, I am not convinced that profit has played a significant role
in its decision-making. But if this general argument is to be used to justify the anti-competitive
aspect of SSRN’s actions, then the future of open-access scholarship should not be trusted to a
for-profit entity. The use of the business corporate form here is appropriate to the extent that the
business’s owners’ goals align with the goals of open access. If the two diverge significantly, we
as scholars should prefer the latter.
Isn’t SSRN’s continued success necessary to the cause of open access?
Actually, no. The arXiv physics preprint archive, the Internet Archive, and the Wikimedia
Foundation are all highly regarded, successful institutions that have used the nonprofit form to
place openness above self-preservation. There are also for-profit repositories that have adopted
better policies than SSRN has. Within the scholarly field, the Berkeley Electronic Press
improves on SSRN in many of the areas I have noted; it features near-instantaneous paper
upload, full-text search, one-click download, no watermarking, and (it claims) lower costs to
subscribing institutions.
Isn’t it vital to have a single central repository for scholarship?
3
No, for a variety of important reasons:
First, the question depends on a false premise. The threat of defection may cause SSRN to
improve or cause convergence on another, more open repository. Either result would further
open access without affecting centralization.
Second, as long as papers are online somewhere and in standard, indexable formats with good
metadata, readers will be able to experience them as though they were hosted centrally, thanks to
high-quality general-purpose search engines. Disaggregation of hosting tools and search tools
would encourage innovation in both.
Third, there are fewer economies of scale in centralized scholarly hosting than one might
assume. The total resource burden associated with downloads of any one school’s faculty
scholarship is well within what that school can sustain. Given that academic institutions already
invest in web hosting, and that hosting more generally is already a highly commoditized service,
centralizing scholarly hosting brings few additional savings in bandwidth and server costs. While
there are some savings involved in having a single architecture and interface design, duplication
of effort has not stopped scholarly institutions from performing many other design and IT
functions for themselves.
Fourth, monocultures are vulnerable. Trust failures, technical failures, design failures—all are
mitigated if scholarship is backed up using redundant servers, institutions, protocols, and
formats. We can always survey scattered, divergent, redundant copies after a disaster to figure
out what we have and how to reintegrate it. We cannot recover from the loss of anything kept
only in one place.
Isn’t download-counting is a necessary function that justifies restrictions?
Download-counting, while a fun parlor game, is a proxy for a proxy, not a high priority
justifying substantial investment. Readership is an inexact approximation to quality; downloads
similarly provide only a guess at readership. SSRN’s restrictions may even inhibit the
development of innovative distributed mechanisms to measure quality. Further, some of those
restrictions are actually unnecessary to the stated goal. If the problem is ballot-box stuffing, the
easiest response is simply to ignore the questionable downloads in the final counts, rather than
taking steps to make downloading harder. Conservation of resources may be a better argument
for at least some restrictions. Still, other institutions have found ways to discourage mistaken or
wasteful downloads while still allowing one-click download.
Why shouldn’t SSRN tie its abstracting service to its hosting service?
There is nothing wrong with the useful technical integration that enables an author to circulate an
abstract and host a paper with one upload. The problem is that SSRN uses its watermarking and
no-outside-URLs rule to prevent authors from combining SSRN abstracting with hosting
elsewhere. (Note that SSRN has no problem with authors who use only the abstracting, since
many papers are abstracted through SSRN but not made available online at all.) This kind of
4
forced bundling inhibits competition from other repositories and inhibits innovation in tools that
work with abstracts.
Of course, open access is imperfectly served by those abstracting series that one needs to pay to
subscribe to. Because of the network effects associated with a high-volume abstracting service,
scholars who might wish to use a different, more open system must realistically also use SSRN’s
comparatively closed system. Scholars and institutions should investigate the possibility of
creating and encouraging free alternatives.
What is to Be Done
When measured against open-access principles, SSRN falls short in several ways, none of them
justified. There are good alternatives available that do not fall short in all these ways, including
hosting by individual scholars, by schools, and by other scholarly repositories. Their existence
shows that SSRN could do better. SSRN should enable direct download, make watermarking
voluntary and opt-in, and allow references to other download sites. If it does not do so, scholars
should use these other alternatives to distribute their papers.
Revision History:
February 26, 2007: Initial draft.
February 27, 2007: Added download locations. Changed “each page” to “each of the first two
pages in the PDF” to reflect a positive change in SSRN’s watermarking implementation.
5