ACTA: International Harmonization at What Cost?

Legal Analysis by Gwen Hinze

The next round of negotiations on ACTA start today in Guadalajara, Mexico. This week’s negotiations will apparently focus on civil enforcement, border measures, and enforcement procedures in the digital environment, and briefly, transparency.

One of the main goals of ACTA is creating new harmonized international IP enforcement standards above those in the 1994 TRIPs agreement. Thirty-seven countries with 37 different national laws are negotiating ACTA, so reaching agreement on new substantive IP enforcement standards will inevitably involve compromises. Some countries will be required to change their national law to bring them closer to other countries' approaches to IP regulation. Since two of the major powers negotiating ACTA are the US and the European Union (and its 27 Member States), there is much scope for different approaches and disagreements to arise. This is particularly true for Internet intermediary liability — where laws in the US and the various EU Member States take quite different approaches.

Which country prevails in this battle of legal wills will have tremendous consequences for citizens' access to knowledge and the future of the Internet as a powerful tool for communication, cross-border collaboration and a platform for innovation.

The EU has indicated that it is unwilling to agree to anything that requires changes to European Community law. EU negotiators would probably not be able to do so under their (still secret) negotiation mandate. On January 14, EU Commissioner-delegate for the Digital Agenda, Neelie Kroes stated that "The objective of ACTA negotiations is to provide the same safeguards as the EU did in the telecoms package... So we stick to our line and that's it."

For its part, the USTR has repeatedly said that ACTA will only "color within the lines of existing US law". Indeed, this is the justification for negotiating ACTA as a sole Executive Agreement, therefore bypassing the checks and balances of the usual Congressional oversight process applied to other recent free trade agreements, such as the US-South Korea FTA.

Given this, it is interesting to reflect on the leaked European Commission’s analysis of the US's Internet Chapter. Although draft text of the Internet chapter has not yet surfaced, the EU analysis discloses what the chapter covers: increased Internet intermediary liability, three strikes Internet disconnection obligations for ISPs, and civil and criminal technological protection measure laws modeled on the US DMCA.

(More after the Jump ..)

[Read the entire post]

MPAA and RIAA Seek Net Neutrality Copyright Loophole

Legal Analysis by Fred von Lohmann

Last week the MPAA and RIAA submitted their comments in the FCC's net neutrality proceeding. As anticipated in EFF's comments, the big media companies are pushing for a copyright loophole to net neutrality. They want to be able to pressure ISPs to block, interfere with, or otherwise discriminate against your perfectly lawful activities in the course of implementing online copyright enforcement measures.

Of course, the MPAA and RIAA couch this in language intended to sound inoffensive. The RIAA says "the perfect should not be the enemy of the good" and "justice often takes too long." The MPAA chimes in that "it is essential that government policies explicitly permit—and encourage—ISPs to work with content creators to utilize the best available tools and technologies to combat online content theft."

But here's how it would work in practice. The proposed FCC net neutrality principles include a loophole for "reasonable network management," which is defined to include "reasonable practices employed by a provider of broadband Internet access service to...(iii) prevent the transfer of unlawful content; or (iv) prevent the unlawful transfer of content." That means that so long as your ISP claims that it's trying to prevent copyright infringement, it's exempted from the net neutrality principles and can interfere with your ability to access lawful content, use lawful devices, run lawful applications, or access lawful services.

This is not about protecting copyright infringers—the FCC's proposed net neutrality principles expressly do not apply to unlawful content or unlawful transmissions. So you don't need a "reasonable network management" loophole to go after illegal conduct. The loophole that the RIAA and MPAA are after is about giving the green light to overbroad copyright enforcement measures that inflict collateral damage on innocent conduct.

The proposed copyright loophole is reminiscent of the RIAA's response when asked about innocent people mistakenly sued for file sharing: "When you go fishing with a driftnet, sometimes you catch a dolphin." Unlike the MPAA and RIAA, EFF doesn't think that ISPs should get a free pass for sideswiping innocent activities if they implement shoddy copyright enforcement systems. And neither do Public Knowledge, the Consumer Electronics Association, CCIA, NetCoalition, or the Home Recording Rights Coalition.

Allowing ISPs to jeopardize perfectly legal activities in the name of "copyright enforcement" is a bad idea. Let the FCC know that you oppose any copyright loophole that would allow the RIAA and MPAA to pressure ISPs into catching your "dolphins" in their poorly designed fishing nets.

[Permalink]

Order to Shut Down Websites Critical of Apex Technology Group is Dangerous and Wrong

Legal Analysis by Kurt Opsahl

Over the holidays, a New Jersey court issued an order requiring upstream providers to shut down three anti-H1-B websites that is deeply dangerous and wrong. The order not only tries to remove allegedly defamatory messages but also requires a complete shutdown of the websites and even purports to require the cooperation of the hosting companies and domain registrars of the websites to do so and for other service providers to identify anonymous speakers.

The plaintiff in the lawsuit, Apex Technology Group, is a staffing and consulting services company. Apex describes itself as "delivering sophisticated technology-enabled solutions to maximize complex business needs." The dispute apparently started when someone uploaded a document purporting to be an Apex employment agreement to docstoc.com, and noted several terms the poster considered unfair to H1-B workers (copy of original post). The H-1B is a non-immigrant visa that allows U.S. employers to temporarily employ foreign workers in specialty occupations. The defendant websites allegedly linked to this post and document, and Apex demanded its removal. Curiously, Apex simultaneously claimed that the document defamed them and that they were its copyright owners. This is unusual, since people rarely defame themselves with their own copyrighted works.

The document and the surrounding controversy prompted further heated discussion in which the websites allegedly accused Apex of being a “bodyshop" that engaged in bad practices while employing H1-B visa workers from India. According to papers Apex filed with the court, at least one website claimed that its members provided evidence of widespread visa and labor fraud by Apex, which they apparently reported to the government. Apex denies any wrongdoing.

Rather than responding to the substance of the criticisms, Apex took the matter to court to try to remove them from the internet. On December 23, Judge James Hurley issued a prior restraint against endh1b.com, itgrunt.com and guestworkerfraud.com, ordering the websites to remove all postings about Apex Technology Group or its President, Sarvesh Kumar Dharayan, until further order of the court. The court also ordered the sites’ ISPs/domain name registrars (DiscountASP.NET, GoDaddy.com, Domains By Proxy and Network Solutions) to stop hosting and “immediately shut down and disable” the websites. Finally, the order requires the ISPs to provide identity information about their customers.

This order dangerously overreaches. By restricting access to entire websites, it places a prior restraint on all of the speech on the websites, even if that speech is unrelated to Apex or Mr. Dharayan. Imagine if a court could order Amazon.com or Yelp.com shut down because of a disparaging review of a single product.

Prior restraints are improper in cases such as this due to the obvious First Amendment problems they pose. Courts limit such injunction to the rare circumstances when (1) the activity to be restrained poses either a clear and present danger or a serious and imminent threat to a protected competing interest, (2) the order is narrowly drawn and (3) less restrictive alternatives are not available. Instead, damages are the preferred sanction for defamatory speech. Here, Apex says it is not even seeking damages. And even if Apex had a valid defamation claim, the wholesale shutdown of a website is not a narrow remedy for a few allegedly defamatory postings.

The New Jersey court’s overreaching order shutting down these websites also is inconsistent with federal law to the extent that it holds service providers to account for user posts. Among other claims, Apex complains about several postings by a anonymous posters that were "allowed to remain public" on Apex’s comment sections. Yet, section 230 of the federal Communications Decency Act protects websites that host content posted by users, providing immunity for a website from state law claims (like defamation) based on the publication of "information provided by another information content provider."

Additionally, the order has troubling implications about the longstanding constitutional right to anonymous speech. In fact, New Jersey was one of the first jurisdictions to apply the right to anonymous speech to the Internet, rendering a decision that has been widely followed over the last decade. Nevertheless, the court ordered Comcast, Yahoo and Facebook to provide identifying information on the anonymous speakers despite the fact that neither the speakers nor the service providers where present at the hearing. Comcast, which is covered by the Cable Privacy Act, was to provide 14 days for the user to challenge the order, while Yahoo and Facebook were ordered to provide the information in 5 days (over the Christmas weekend), without a provision for challenging the order. Facebook has stood up to the order and has not provided the requested information.

Ordinarily, in order to safeguard this First Amendment right, a litigant seeking to unmask an anonymous speaker would need to obtain a subpoena from an appropriate court (i.e. Santa Clara county in California for Yahoo) and serve the service provider. Then the service provider would provide adequate notice to the user, and the user could move to quash the subpoena, asserting whatever defenses the user may have. These procedures are vital to protecting speech rights, and it was inappropriate and unnecessary for the New Jersey court to short-cut that process, especially over a holiday period when its is all the more difficult to obtain emergency legal assistance.

Finally, it was wrong for the court to require the upstream providers to unplug the website. Under New Jersey law, injunctions should only reach those who engage in "active concert or participation" with the person who acted wrongly. There’s no indication that the upstream providers or domain name registrars for the websites even knew about the postings in question, much less acted in "active concert" with them. Requiring domain name registrars to turn off websites in litigation about the website is a tactic that has already been rejected.

The New Jersey court order is therefore wrong in at least four ways: (1) it creates a prior restraint that takes down too much speech, (2) it wrongly punishes websites for the speech of their commenters, (3) it wrongly requires the identification of anonymous speakers without sufficient opportunity to challenge the disclosure, and (4) it wrongly enlists out-of-jurisdiction upstream providers who did not act in concert with the websites in taking down speech. We hope the parties and the upstream and domain name hosts involved will seek to overturn it.

[Permalink]

EFF Submits Brief in Key State Secrets Privilege Case

Legal Analysis by Cindy Cohn

EFF filed an amicus brief in the Ninth Circuit's en banc review of Mohamed v. Jeppesen, a case brought by the ACLU challenging the CIA's extraordinary rendition program. A panel of the Ninth Circuit Court of Appeals had rejected the government's argument that the case had to be dismissed at the outset due to the state secrets privilege. The panel decision is now being considered by a larger, en banc panel of the Court.

EFF notes that the government has made the same dangerous and overreaching state secrets arguments in the domestic warrantless wiretapping cases handled by EFF. The brief begins:

This case is another in a set of post-September 11, 2001 cases in which the Executive, having made new and tremendously broad assertions of its unilateral power, seeks to prevent the Judiciary from adjudicating the lawfulness of those new powers. To do so, the Executive skews the relevant caselaw on the state secrets privilege, attempts to rely on a case in which the privilege was not even the basis for the decision and claims that the court must blind itself to credible, admissible, nonsecret evidence because the Executive has determined that it cannot confirm or deny a particular fact. Adopting the government’s position would abdicate the Judiciary’s Article III responsibility to adjudicate the constitutional and statutory limits on Executive authority.

Oral argument is scheduled in the case in San Francisco on December 15, 2009. EFF has been urging Congress to reform the state secrets privilege.

[Permalink]

Latest Bogus DMCA Takedown Award Winner: Yahoo!

Legal Analysis by Fred von Lohmann

"Yahoo isn’t happy that a detailed menu of the spying services it provides law enforcement agencies has leaked onto the web." That's how WIRED's Threat Level blog put it when describing Yahoo's recent effort to censor its own law enforcement compliance guide off the Internet using a bogus DMCA takedown demand.

The trouble all started when Yahoo stepped in to block a FOIA request for its law enforcement compliance "price list" (i.e., what it charges to law enforcement and spy agencies when responding to requests for information about Yahoo users). Shortly thereafter, a copy of the document, entitled "Yahoo! Compliance Guide for Law Enforcement," appeared on Cryptome.org.

Here's where the bogosity begins in earnest. Yahoo sent a formal DMCA takedown notice to Cryptome.org, demanding the removal of the compliance manual. In the letter, Yahoo's lawyers allege that posting the manual infringes Yahoo's copyrights (the only proper basis for a DMCA takedown), as well as claiming that it's a trade secret (absurd for a marketing document) and that posting it constitutes "business interference" (huh? informing customers about Yahoo's disclosure practices "interferes" with business?).

This should earn Yahoo a place in the Takedown Hall of Shame (we'll be updating our list of inductees soon). Posting the compliance manual is a clear fair use. Consider the "four factors" that courts examine in fair use cases: (1) publication is clearly for a transformative purpose (criticism, public debate); (2) publication does not harm the "market" for the original (since Yahoo doesn't sell copies of the manual); (3) the nature of the publication is factual, not highly creative; and (4) while the whole manual was published, that was necessary for the transformative purpose. And, perhaps most important, a federal court has already ruled in favor of fair use on nearly these same facts, when Diebold Election Systems was sued for trying to censor embarrassing internal documents off the Internet using bogus DMCA takedowns.

This brings up another important point: the DMCA does not require service providers to comply with bogus takedown notices. The DMCA offers a "safe harbor" from money damages for copyright infringement, but you only need a "safe harbor" if the activity in question might be infringing in the first place. Where (as here) the activity is clearly not infringing, a service provider doesn't need the DMCA for protection, and can just deposit takedown notices in the trash (as YouTube did a few months ago in the face of another obviously bogus takedown notice).

[Permalink]

Google Books Settlement 2.0: Evaluating Censorship

Legal Analysis by Fred von Lohmann

This is the fifth in a series of posts about the proposed Google Book Search settlement.

As we've explained in earlier posts, when it comes to evaluating the proposed Google Books settlement, the principal potential benefit to the public (increased access to books online) must be weighed against the potential drawbacks (impediments to competition, inadequate protection for privacy). Another potential downside for the public in the proposed settlement is the risk of censorship.

To understand the importance of this risk, keep two things in mind. First, while bookstores are entitled to pick and choose their inventory, Google Books hopes to be much more than a simple bookstore. In the words of Google's CEO Eric Schmidt: "Imagine one giant electronic card catalog that makes all the world's books discoverable with just a few keystrokes by anyone, anywhere, anytime." In other words, Google Books will have many characteristics that we associate more with the research libraries from which its books are drawn than with traditional bookstores. Second, as Prof. Geoffrey Nunberg reminds us: "This is almost certainly the Last Library, after all. There's no Moore's Law for capture, and nobody is ever going to scan most of these books again."

If Google's scans under the proposed settlement are likely to be the only chance millions of books will have for a digital life, then the potential for censorship is something to be taken very seriously indeed. If the books can't be found by researchers, it will be as though they were cast down the Memory Hole.

Censorship by Rightsholders

The biggest censorship risk created by the proposed settlement is from copyright owners. The proposed settlement gives rightsholders (until April 2011) the power to "Remove" their books from the Google Books corpus altogether. Once a book is removed, not only won't you be able to read it online, you won't even be able to find it using full-text search. In short, these books would simply cease to exist as far as users of Google Books are concerned, despite the fact that courts have ruled that indexing copyrighted works is a perfectly legal fair use. Moreover, even the libraries who contributed the book for scanning wouldn't have a digital "backup" in their collections, as these removed books would also vanish from the digital copies that Google gives back to the research libraries (the "Library Digital Copies" and the "Research Corpus," in the lingo of the settlement agreement).

Why would a rightsholder want to self-censor? First, remember that the author of a book is often not the rightsholder. As a result, the copyright in a book can be purchased and then used to suppress further publication (a trick Howard Hughes tried). Moreover, sometimes the author or author's heir (or corporate successor) wants to suppress a work (Prof. R. Anthony Reese describes a number of historical examples of post-publication suppression efforts by authors and rightsholders in this article).

In the world of research libraries, of course, this kind of censorship is impossible—no research library would pull cards from the catalog and destroy copies of published works at the behest of those who own the copyright in those books. Yet this is exactly what the proposed settlement would permit for the "Last Library." And most galling is that the settlement does not even require that a complete list of these "Removed" books ever be made publicly available (in Google's web search, in contrast, Google includes entries for results that would have appeared, but for DMCA takedown demands, and makes those demands publicly available through Chilling Effects).

At a minimum, books that are "removed" should remain in the database for full-text search, and Google should remain able to offer a "Library Link" (i.e., a link that directs a researcher to a library where the book can be found).

Even more troubling is the possibility of selective alterations of the texts of the books themselves. In Section 3.10(c)(i), the settlement forbids Google "except as expressly authorized by the Registered Rightsholder" from altering the text of scanned books when displayed to users. That's certainly a good thing, as far as it goes—we shouldn't want Google to be able to go in and selectively edit books. But Google is allowed to selectively edit if "authorized" by the copyright owner. Why is this permitted? And if the rightsholder "authorizes" Google to make changes, can Google refuse to do so? Will the fact of alteration be publicly visible to the reader? The answer is not clear. But clearly the better rule is a prohibition on anyone making editorial alterations in the text of scanned books (again, no library would allow a copyright owner to selectively blackline books in the stacks). Any other option creates the chilling prospect of "revising history" as imagined in Orwell's 1984.

Censorship by Google

The proposed settlement also gives Google a troubling degree of discretion when it comes to choosing which books will be publicly accessible. For example, Section 3.7(e) makes it clear that Google can exclude any scanned book it likes from public access "for editorial or non-editorial reasons." If it excludes a book for "editorial reasons," it must notify the Registry (but not the public), and the Registry may look for an alternative partner ("Third-Party Required Library Service Provider") to host the book. There is nothing that requires the Registry to do so, nor any guarantee that such a partner will step forward.

In addition, in order to meet its obligations under Section 7.2(e) of the proposed settlement, Google need only make 85% of the books it scans from its library partners publicly accessible through full-text search, consumer purchase, or the institutional subscription database. Assuming that Google has already scanned approximately 8 million books that are in-copyright, that means Google can make more than 1.2 million of these books disappear from its publicly accessible services for any reason and still meet its obligations under the settlement. And, again, nothing in the settlement requires Google to make the list of omitted books available to the public.

Censorship by Government

Finally, it's worth noting that governments will doubtless exploit the leeway that the settlement gives to both rightsholders and Google to pull books off the digital shelves of Google Books. It's all too easy to imagine foreign governments pressuring their citizens to "remove" books from public access on Google. It's also likely that foreign governments will pressure Google to omit books from Google Books. If that comes to pass, neither Google nor the rightsholders will be able to say that they are legally constrained by the settlement from complying short of legal process. Had the settlement agreement been written to forbid this kind of censorship, both rightsholders and Google could have responded to censorship demands by saying "come back with a court order."

And, finally, remember that Google may, under the settlement, sell off the entire Google Books project. So even if you believe that Google would never cave to foreign governments or engage in selective censorship, keep in mind that 10 years from now, Google Books might be owned by an entirely different corporate master.

[Permalink]

Google Books Settlement 2.0: Evaluating Privacy

Legal Analysis by Fred von Lohmann

This is the fourth in a series of posts about the proposed Google Book Search settlement.

We have now examined the chief promised benefit (increased public access) of the proposed Google Books settlement, as well as one of the chief potential drawbacks (impaired competition). Another down-side to the proposed settlement is its lack of adequate protections for reader privacy. And although EFF has repeatedly written about the privacy problem and outlined specific steps that could be taken to address it, as have the ACLU, CDT, EPIC, library associations, and academic authors, the revised Settlement 2.0 still does nothing new to address the serious privacy concerns raised by the Google Book Search services.

[Note: EFF represents a groups of authors and publishers who have filed an objection to the proposed settlement on privacy grounds, arguing that the lack of reader privacy protections is also a threat to the interests of authors and publishers, particularly of books on controversial or sensitive topics. This post, however, will focus on the interests of readers, rather than authors and publishers.]

The Reader Privacy Problem

The products and services envisioned by the proposed settlement will give Google not only an unprecedented abililty to track our reading habits, but to do so at an unprecedented level of granularity. Because the books will be accessed on Google's servers, Google will not only know what books readers search for and access, but will also know which pages they read, how long they stayed on each page, what book they read before, and which books they access next. This is a level of reader surveillance that no library or bookstore has ever had.

Readers who feel surveilled will be chilled in their freedom of inquiry. As Supreme Court Justice William O. Douglas observed in 1953, “Once the government can demand of a publisher the names of the purchasers of his publications . . . [f]ear of criticism goes with every person into the bookstall . . . [and] inquiry will be discouraged.” Or as Author Michael Chabon put it: "If there is no privacy of thought — which includes implicitly the right to read what one wants, without the approval, consent or knowledge of others — then there is no privacy, period."

And it's not just Google that might want records about your reading habits. A core concern EFF has with the proposed settlement is that under it Google need not insist on a warrant before turning over this sensitive reader information to governmental authorities or private third parties. This is hardly a hypothetical risk: between 2001 and 2005, libraries were contacted by law enforcement seeking information on patrons at least 200 times. And in 2006 alone, AOL received almost 1,000 requests each month for information in civil and criminal cases.

This lack of protections for reader privacy stands in sharp contrast to the privacy protections that librarians and bookstores have been fighting for in connection with physical books for decades. Nearly every state has laws protecting the privacy of library patrons. Yet when Google scans books it got from libraries, privacy protections could be left behind at the digital threshold if Google doesn't stand up for them.

Google's Privacy Policy for Book Search

Google has announced a privacy policy for Google Books. While it addresses some of the privacy concerns EFF and others had raised, it does not go nearly far enough. As we've previously explained, the privacy policy can be changed at any time, is not an enforceable obligation tied to the proposed settlement agreement, and:

• as noted above, fails to commit to a "come back with a warrant" standard before disclosing reader information to the government;

• fails to require Google to delete logging information about users within 30 days, or any other reasonably short period of time;
• allows (albeit upon opt-in consent) Google to aggregate the information it learns about readers with other information it knows about readers from other sources, including its other services and its DoubleClick product that places cookie-traced advertising on millions of non-Google websites across the Internet;
• fails to ensure that readers will always be able to use anonymity services like the Tor network, proxy servers and anonymous VPN providers to access Google Books;
• does not offer registered users who purchase texts any equivalent of a "hiding books under their bed" to protect against parents, family members or other local users who might scrutinize their reading (we have suggested several ways that Google might implement a feature like this, and hope that Google will eventually do so);
• does not allow purchasers to cover their reading tracks by anonymously transferring or giving purchases to accounts that do not have Google Checkout or other identifying features (we've also discussed technical methods for this with Google and believe they are seriously considering it);
• fails to provide a robust, easy-to-read notice of and link to the Google Books privacy provisions on the Google Books pages themselves, rather than tucked away in a privacy policy;
• fails to promise to annually publish online, in a conspicuous and easily accessible area of its website, the type and number of requests it receives for information about Google Book Search users from government entities or third parties; and
• fails to require Google to store information about readers, who must be in the U.S. under the terms of the settlement, in the U.S. so that they will be protected by U.S. privacy laws.

For all of these reasons, in its present form and without further affirmative steps by Google either in the context of the settlement or outside it, the proposed Settlement 2.0 makes Google Books a threat to reader privacy, which in turn is a serious a down-side that must be weighed against the settlement's potential benefits.

[Permalink]

Google Books Settlement 2.0: Evaluating Competition

Legal Analysis by Fred von Lohmann

This is the third in a series of posts about the proposed Google Book Search settlement.

Now that we've described the proposed settlement agreement's biggest potential upside for the public—expanded online access to books, particularly out-of-print books—that benefit must be weighed against the potential down-sides. On that score, the settlement's potential impact on competition in the online book market has loomed large. Critics of the settlement have emphasized two principal dangers:

1. The potential for a Google monopoly over orphan and unclaimed books.
2. The potential for monopolistic pricing of the Institutional Subscription Database, particularly for higher education.

The revised Settlement 2.0 made little or no effort to address these concerns, leaving it to Congress or antitrust authorities to fix later.

A Google Monopoly on Orphan & Unclaimed Books?

At the heart of the proposed settlement is a bargain that lets Google (and only Google) leapfrog the problem of "unclaimed works"—books whose copyright owners cannot be found or whose owners can't be bothered to fill out paperwork for a small payment disbursed by the Registry (consider how many "class action" notices you've tossed in the trash unread). Thanks to the magic of the class action process, the settlement solves this problem by resolving the copyright claims of these otherwise unreachable copyright owners and designating all of their works by default as available for "Display Uses" by Google. In other words, so long as no one steps forward to claim these books, Google (and only Google) has a license to make them available in all the ways the settlement allows.

Many who filed objections to the proposed settlement, including the Department of Justice, Microsoft, Amazon.com, the Internet Archive, and Public Knowledge, among others, argued that this could create a de facto Google monopoly over online use of these unclaimed works. And while the revised Settlement 2.0 creates an "Unclaimed Works Fiduciary" (UWF) to act as a guardian on behalf of owners of unclaimed works, neither the UWF nor the Registry has the power to grant a similar license to any other entity that might want to make the same kinds of uses that Google will be entitled to make under the settlement.

Nobody likes this "only-for-Google" aspect of the settlement—in fact, Google has said that it would support orphan works legislation that would empower the Registry to make the same deal (or even a better deal) with others who want to use these unclaimed works. (Where the claimed books are concerned, in contrast, the Registry will likely ask the rightsholders to appoint it to license companies other than Google. But that still leaves all the unclaimed books out.) The settlement agreement even has a provision that makes it clear that the UWF can license others "to the extent permitted by applicable law"—what amounts to an "insert orphan works legislation here" invitation.

But absent some legislative supplement to the revised Settlement 2.0, it still seems that any other company would have to scan these books, get sued, and hope for a class action settlement. That, of course, is the kind of barrier to entry that any monopolist would envy.

This raises a worthy question: if legislation is necessary to fix the competition problem posed by the settlement, then why do we need a class action settlement in the first place? Why not solve what seems like a quintessentially legislative problem with legislation, instead? (As Amazon points out, that's exactly what was done when music publishers brought a class action against the first digital audio tape (DAT) recorders).

Here's where realpolitik enters the equation. Google correctly points out that Congress has been working on orphan works legislation for years, to no avail. And none of the legislative proposals came close to the comprehensive solution embodied in the proposed settlement. So the question boils down to a political one: do you believe that approval of Settlement 2.0 will make orphan works legislation more likely, or less likely? Without a crystal ball, it's hard to know.

Monopoly Pricing of the Institutional Subscription Database?

One of the commercial services that Google is authorized to provide under the proposed settlement is the "Institutional Subscription Database" (aka "ISD"), which will provide "all-you-can-eat" access to the corpus of scanned books. The chief customers for the ISD are likely to be universities (the same folks who are providing Google with the books to be scanned), for whom instant digital access to every word in every book in Google's collection is likely to be very compelling.

The big question is whether, over time, the ISD will become the one database that no university can do without, and the one database with no market substitute (again, because Google will be the only company who can provide a comprehensive corpus without fear of copyright liability, for the reasons explained above). This, of course, is a recipe for monopolistic price gouging, as a group of academic authors led by Prof. Pam Samuelson have pointed out. Over time, universities could face spiraling prices as Google and the Registry conspire to maximize their revenues on the ISD product.

Google and its supporters respond by pointing out that the settlement requires that pricing for the ISD be set with regard to "two objectives: (1) the realization of revenue at market rates for each Book and license on behalf of Rightsholders and (2) the realization of broad access to the Books by the public, including institutions of higher education." The settlement goes on to promise that Google and the BRR "will use the following parameters to determine the price of Institutional Subscriptions: pricing of similar products and services available from third parties, the scope of Books available, the quality of the scan and the features offered as part of the Institutional Subscription."

But Google's own people have reportedly admitted that there might not be any "similar products and services" to the ISD. And the settlement does not give ISD subscribers the right to go to court to enforce these "objectives" and "parameters." Instead, Google has entered into "side agreements" with some of its major library partners (U. of Michigan, U. of Wisconsin—both of which will be receiving subsidies from Google for their ISD fees) that allow only those institutions to challenge pricing, and only under certain circumstances. So what we are left with is a "trust us" from Google, the Registry, and their biggest library partners.

Of course, the chances of this coming to pass are hard to know in advance. As we have pointed out, if many large publishers pull their books out of the ISD database, then perhaps the ISD service won't become indispensable to universities after all. So, ironically, the more successful the ISD proves to be, the more of a danger its pricing mechanism might prove to be for higher education.

Fixing the Competition Problem

Just because the proposed Book Search settlement isn't good for competition doesn't mean it's illegal. There is a robust debate going on (see, e.g., articles by Picker, Elhauge, Fraser, Lemley, and Picker again) about whether the proposed settlement might violate antitrust laws, and the Antitrust Division of the Department of Justice will doubtless continue its investigation.

But we shouldn't be satisfied with antitrust law here. This is not just a simple market transaction between commercial entities. Google is building an enormously important public resource, a task it can only undertake with the blessing of a federal court. The public deserves a solution that is not "barely legal," but that instead encourages real, robust competition. As written, without some modification or legislative adjunct, Settlement 2.0 does not do that.

[Permalink]

Google Books Settlement 2.0: Evaluating Access

Legal Analysis by Fred von Lohmann

This is the second in a series of posts about the proposed Google Book Search settlement.

The Potential Upside: Enhanced Public Access

From the public's point of view, unprecedented public access to books is the chief benefit promised by the revised proposed settlement (aka Settlement 2.0) of the Google Book Search litigation. That's the "upside" against which all the possible "down-sides" will be measured. And when it comes to enhancing public access, the proposed settlement holds great promise. Whether that promise will actually come to pass, however, is harder to predict.

Here's what we know about Google's book scanning efforts so far [revised in light of updated numbers sent by Google Nov. 19]:

• Google has already scanned more than 12 million books (for comparison, U.S. libraries hold an estimated 42 million titles total).
• Roughly 50% are in languages other than English, with more than 100 languages represented. (In the revised settlement proposal, however, the parties have tried to exclude most books published in countries other than the US, UK, Australia, or Canada, so some non-English language books may now be excluded.)
• 2 million are clearly in the public domain (i.e., published pre-1923, government works, etc).
• 2 million have been scanned with the explicit permission of copyright owners as part of Google's partner program.
• That leaves ~7 million scanned volumes that are potentially the subject of the copyright lawsuits and the proposed settlement (given the low rate of copyright renewals for works published between 1923-1963, it is likely that a substantial portion of these 7 million volumes may actually be in the public domain, in which case they would fall outside the settlement).

So how much access will the public have to the scanned books that fall within the scope of the settlement (that's the ~7m already scanned, as well as millions more Google will be scanning in the future)? The answer will vary based on their copyright status, what services Google implements, and the expressed wishes of copyright owners:

Out-of-print, in-copyright books: For these books—principally out-of-print books published after 1923—the settlement envisions Google providing access through four principal mechanisms:

• "Preview Uses" (show up to 20% of the book, for free, in response to search queries);
• "Consumer Purchase" (permanent, full-text, online access on a book-by-book basis for a fee);
• "Institutional Subscription" ("all-you-can-eat" full-text online access on a blanket basis through an institution); and
• "Public Access Service" (at least one free public terminal for public libraries).

All of these "Display Uses" will be enabled by default under the settlement agreement for out-of-print, in-copyright books. This is just a default, however; copyright owners are entitled to change the default by electing to "Remove" or "Exclude" their books from any or all of the Display Uses. Of course, where unclaimed works (books whose copyright owners cannot be located or have not bothered to sign up with the Registry) are concerned, the default will effectively be the rule, which is a good thing for public access to these works.

In-print, in-copyright books: By default "Display Uses" will not be permitted for these books. In other words, if Google scans these books, they will go into the database corpus, but will not be available for Preview, Consumer Purchase, or Institutional Subscription, unless the copyright owner chooses to enable one or more of those uses. In short, no public access unless the copyright owner chooses to allow it.

Google Partner Program books: Under the settlement, copyright owners of both in-print and out-of-print books can elect to pull their books out of Google's database corpus, choosing instead to negotiate a different deal in the Google Partner program, which gives the copyright owner more flexibility to define exactly how the book can be accessed. Some observers anticipate that many, perhaps most, major publishers will take this option and remove their works from the products and services described in the settlement.

The Potential: Unprecedented Online Access

Taken together, these features mean that the Google Books project could potentially provide Americans (and only Americans, as the settlement only authorizes Google to offer Display Uses of in-copyright books to U.S. Internet users) with unprecedented instant access to a large collection of books that previously were available only in research university libraries. In particular, like the Internet before it, Google Books could make specialized resources available to people who otherwise might never be able to access them (see, e.g., Google's agreements to digitize U. of Wisconsin's Native American collection and U. of Texas' Benson Latin American collection).

In addition to enabling search and reading, the products and services envisioned by the settlement could also unleash innovative, transformative new uses for the information inside these books. For example, the availability of all these readily citable books could radically expand and transform Wikipedia, which places a premium on citations to neutral sources to validate edits to its pages. Once every Wikipedian can do full-text searches against the research collections of major university libraries, Wikipedia should see a huge expansion of cited contributions.

The proposed settlement also offers the promise of unprecedented access for the visually impaired. The proposed settlement commits Google to offering screen enlargement, read-aloud, and Braille displays ("Accommodated Service") for the Institutional Subscription product. As the National Federation for the Blind and a coalition of other disability rights groups have pointed out, this will make a "historically unprecedented" number of books accessible to the visually impaired.

In addition, under the terms of the settlement, Google may make two copies of the scanned books database ("Research Corpus") available through university libraries for "nonconsumptive" research (i.e., you can use it to develop your new OCR algorithms, but not to extract and compile every paragraph that mentions zombies to create a "Zombies Through The Ages" book). Although use of the Research Corpus will be subject to a number of restrictions that have drawn fire from academics, the creation of such "Research Corpus" would nevertheless be an important step forward for access. Programmatic access to a large database of books is likely to open new avenues of scholarly inquiry and unleash new innovations, including better search algorithms, optical character recognition techniques, automated language translation breakthrus, and other uses that we haven't yet imagined.

The Uncertainty: Empty Promises, Empty Shelves?

But the promise of what the settlement might accomplish is no guarantee of ultimate results.

First, under the settlement copyright owners can pull their books (see Section 3.5, "Right to Remove or Exclude") out of all the products and services envisioned by the settlement, including full-text search and limited "snippet view" access. This is essentially the "take the money and run" option—the copyright owner collects a per-book payment from Google for books already scanned, but then the public gets no online access to these books unless and until the copyright owners negotiate new deals with Google or other online providers. This effectively gives copyright owners a unilateral right to trump fair use, essentially "unpublishing" their books online. Some observers expect that most major publishers will opt to "take the money and run" for both their in-print and out-of-print titles, leaving gaping holes on the virtual shelves of Google Books. If this takes place, then the settlement would only foster access to orphan and unclaimed works. Still good, but far short of full access to every book in the University of Michigan library.

Second, Google is not required to offer all the products and services envisioned in the settlement. The settlement only compels Google to offer the following within 5 years (see Sections 3.7(a), 7.2(e)(i), 7.2(g)(ii)(1)):

• Consumer Purchase (not clear what percentage of the scanned books must be made available)
• Institutional Subscription for Higher Education, including Accommodated Service (for at least 85% of books scanned)
• Public Access Service (for at least 85% of books scanned)
• free search services (including Snippet View and Preview, for at least 85% of books scanned)
• Library links that will help you find a library with hard copy (for at least 85% of books scanned)

Notably absent from this list is the Research Corpus described above (in side agreements with its library partners, however, Google has made monetary commitments toward building the Research Corpus). And if Google never gets more than 85% of eligible books online, that would represent still more gaps on the virtual shelves.

Third, the public gets only the kinds of access that Google makes available, only through interfaces that Google chooses to expose. And while this level of access is certainly preferable to no access at all, the "One Interface to Rule Them All" approach is likely to impede innovation, which ultimately means less access. It would be preferable if others had access to the underlying book scans, just as Google had access to the World Wide Web when it built its own search engine. (Google will protest that it spent the money to make the scans, and it's unfair to allow competitors to free-ride on its scanning investment. We already posted our answer to that objection.)

And Don't Forget the Down-Sides

So while the settlement does offer the exciting promise for drastically increasing public access to books, it is hard to predict whether that promise will be fulfilled. And even if the promise of access were fulfilled, there are other down-sides to the settlement, which we will take up in our next posts.

[Permalink]

Google Books Settlement 2.0: Evaluating the Pros and Cons

Legal Analysis by Fred von Lohmann

This is the first in a series of posts evaluating the proposed Google Book Search settlement.

When it announced its Book Search project in 2004, Google set for itself an inspiring and noble goal. In the words of Google CEO Eric Schmidt, "Imagine yourself at your computer and, in less than a second, searching the full text of every book ever written." What started as a dream of universal book search, however, has become something much broader: a class action lawsuit and proposed settlement that hopes to let Americans read, as well as search, millions of books online.

The fate of that more ambitious plan is now before a court in New York. In the face of opposition from many quarters (including EFF and the U.S. Department of Justice), Google and class representatives for authors and publishers recently revised the proposed settlement (aka "Settlement 2.0", 300-page PDF redline posted here). The court is expected to decide whether to approve the revised settlement sometime in the first half of 2010.

Advocates on both sides have had their say, both in court and elsewhere. In a series of posts, we will be evaluating Settlement 2.0 from a variety of angles intended to illuminate whether it is a good deal for the public, including its impact on access, competition, privacy, censorship, innovation, and fair use. Is Settlement 2.0 the best we can do? The most we can reasonably expect? Not good enough? How does it measure up against our aspirations the future of online digital libraries?

Here's a preview of the overall contours of the debate. The chief benefit of the proposed settlement is the increased public access to books (particularly out-of-print books) that it makes possible. Against this important benefit must be balanced concerns about possible detrimental effects on privacy, competition, innovation, and fair use. Complicating the overall analysis are the requirements and limitations of class action litigation, as well as the inherent difficulty in predicting how copyright owners and readers will respond to the new Google products and services contemplated in the proposed settlement.

In the meantime, if you want to keep track of the latest breaking news regarding the proposed settlement, we recommend Prof. James Grimmelmann's blog, The Laboratorium, as well as The Public Index, which includes a handy, linkable, annotatable copy of the entire proposed settlement agreement. For the latest from Google about Google Books, we recommend the Google Book Search blog and Google Public Policy blog.

[Permalink]