US Dept. Of Justice: Court “Lacks Authority To Approve” Google Book Search Settlement
With a final “fairness hearing” scheduled for February 18, the US Department Of Justice has filed briefs (PDF) in opposition to the Google Book Search Amended Settlement Agreement (“ASA”). The DOJ commends the parties for their efforts to reach an amended agreement that addresses some of the problems with the original settlement but concludes that […]
With a final “fairness hearing” scheduled for February 18, the US Department Of Justice has filed briefs (PDF) in opposition to the Google Book Search Amended Settlement Agreement (“ASA”). The DOJ commends the parties for their efforts to reach an amended agreement that addresses some of the problems with the original settlement but concludes that there are still fundamental anti-trust issues with the ASA:
Despite the commendable efforts of the parties to improve upon the initial Proposed Settlement, many of the problems previously identified with respect to the original settlement remain in the ASA [Amended Settlement Agreement]. The United States remains committed to working with the parties on the settlement’s scope and content.
The Department Of Justice’s filings argue, effectively, that class action litigation is the wrong mechanism to address the myriad business and copyright questions at issue in the sweeping agreement. It also presents the related question of whether the court, in fact, has the actual authority to approve the ASA in the context of the specific federal statutes raised and adjudicated in the case.
Here’s a excerpt from the government’s papers, which captures the essence of the objections that are then laid out in a technical-legal way throughout the brief. Key parts have been bolded:
Despite this substantial progress, substantial issues remain. Although the United States believes the parties have approached this effort in good faith and the ASA is more circumscribed in its sweep than the original Proposed Settlement, the ASA suffers from the same core problem as the original agreement: it is an attempt to use the class action mechanism to implement forward-looking business arrangements that go far beyond the dispute before the Court in this litigation. As a consequence, the ASA purports to grant legal rights that are difficult to square with the core principle of the Copyright Act that copyright owners generally control whether and how to exploit their works during the term of copyright. Those rights, in turn, confer significant and possibly anticompetitive advantages on a single entity – Google. Under the ASA as proposed, Google would remain the only competitor in the digital marketplace with the rights to distribute and otherwise exploit a vast array of works in multiple formats. Google also would have the exclusive ability to exploit unclaimed works (including so-called “orphan works”1) without risk of liability. The ASA’s pricing mechanisms, though in some respects much improved, also continue to raise antitrust concerns.
We’ll provide more analysis and context later after we’ve had more time to digest the arguments and objections of the US government. Also see this press release from the US Department Of Justice and this AP article on it. For further background on the case and the proposed amended settlement, also see:
- Revised Google Book Settlement Filed & Live Blogging The Press Call
- Google’s Schmidt To Book Settlement Critics: What’s Your Solution?
Postscript From Danny Sullivan: I’m now reading through the filing and effectively doing a fast live blogging of it. Here are sections that stand out to me. I’ve bolded especially interesting parts. From Page 2:
As a consequence, the ASA purports to grant legal rights that are difficult to square with the core principle of the Copyright Act that copyright owners generally control whether and how to exploit their works during the term of copyright. Those rights, in turn, confer significant and possibly anticompetitive advantages on a single entity – Google. Under the ASA as proposed, Google would remain the only competitor in the digital marketplace with the rights to distribute and otherwise exploit a vast array of works in multiple formats. Google also would have the exclusive ability to exploit unclaimed works (including so-called “orphan works”1) without risk of liability.
Also from page 2:
The United States recognizes that the parties to the ASA are seeking to use the class action mechanism to overcome legal and structural challenges to the emergence of a robust and diverse marketplace for digital books, including through the adoption of an “opt-out” system to address the complexity of managing millions of third-party copyrights. Under existing law, copyrighted works typically cannot be exploited in all of the ways the ASA contemplates without the prior permission of the rightsholders. But for many works, especially out-of-print works, rights clearance may not be possible as a practical matter. Even if the rightsholders can be located, it may not be clear (as between author and publisher, or as among many competing claims to rights in the work) who “owns” or controls the digital licensing of particular works. This is especially likely where publication predated, and contracts did not anticipate, the digital era. Finally, there are no major licensing systems in place by which good faith users can efficiently secure permission from, and render payment to, authors, publishers, and the other rightsholders implicated by the ASA.
This is interesting because opt-out is exactly how web indexing works. Google and other major search engines do not gain the explicit permission of rights holders before making a copy of pages in order to index them for searching purposes. Here, the DOJ seems to suggest that opt-out isn’t good enough and also notes that there are no major licensing systems. In the web word, one could argue that the robots exclusion protocol effectively works as a licensing system. Even if an author isn’t explicitly know, Google can still automatically ask the hosting web server for permission.
Search Engines, Permissions & Moving Forward In Copyright Battles is a primer I wrote that explains these issues for, especially comparing web search to book search. In inability for book authors to automatically opt-out is one reason why I suggested in it that Google not scan copyright books without explicit permission.
This is also a good time to note that many in-copyright books have been scanned with permission. And any that you can read via Google Book Search are there because authors or publishers did grant permission. There’s a great deal of confusion on this issue.
From page 3:
These realities make it difficult if not impossible to engage in lawful large-scale book digitization projects, thereby denying the public the full corpus of 20th
century books and, perhaps, unknown benefits of future creativity and economic innovation.
Is this the DOJ saying it doesn’t view the scanning Google has done to be lawful? Not quite, as I’ll get into further below.
Also from page 3:
Despite this worthy goal, the United States has reluctantly concluded that use of the class action mechanism in the manner proposed by the ASA is a bridge too far. The Supreme Court has cautioned that “Rule 23, which must be interpreted with fidelity to the Rules Enabling Act and applied with the interests of absent class members in close view, cannot carry the large load” of restructuring legal regimes in the absence of congressional action – however sensible that restructuring might be. Amchem Prods., Inc. v. Windsor, 521 U.S. 591, 629 (1997).2 That caution should be heeded here. Indeed, it applies with particular force because the legal and structural changes the parties seek to accomplish would confer on one entity a level of market dominance that other competitors without access to the ASA’s special rules and procedures will be hard pressed to challenge for the foreseeable future. For these reasons, the ASA is not the appropriate way to achieve the laudable goals the parties seek.
As I read it, since the settlement involves books from many parties that can’t be found (“orphan works” that are in copyright but where it is difficult or impossible to currently find the rights holders), a class action suit can’t involve them.
From page 4:
At this time, in the view of the United States, the public interest would best be served by direction from the Court encouraging the continuation of settlement discussions between the parties and, if the Court so chooses, guidance as to those aspects of the ASA that need to be addressed. The United States is committed to working constructively with all stakeholders on the scope and content of an appropriate settlement of this matter, and on legislative or market-based solutions to ensure a robust marketplace for digital works.
I’m at a loss as how this is supposed to magically happen. The DOJ has indicated that it feels many books were illegally scanned. At the same time, it finds the settlement an attempt to a “laudable goal” of getting around that problem. But if you don’t have a class action suit — and you can’t find the rights-holders of some of those books — what then?
It may be that the two main parties who brought the suit against Google, the American Association of Publishers and the Authors Guild, will only be able to strike a settlement for books that have known rights-holders. As for orphan works, potentially those could be excluded — and Google could potentially continue to use them for searchable purposes unless the actual rights holders come forward. Or, potentially the US government might take action against Google over those works. We’ll see what shakes out as this filing is digested.
Indeed, later on page 4 is this:
The United States accepts the proposition that a properly defined and adequately represented class of copyright holders may be able to settle a lawsuit over past conduct by licensing a somewhat broader range of conduct.
That seems to say if all these unknown rights-holders and their works are excluded, then a class action settlement can go forward. You get more of this on page 5:
In previous submissions to this Court, the United States (and other interested parties) discussed the Rule 23 limitations expressed in Amchem, 521 U.S. at 620, 628-29, see, e.g., U.S. SOI at 6-8 (D.E. 720), which suggests that absent class members cannot be adequately represented as to uncertain injuries or rights that are far removed from the facts underlying the complaint.
Many critics of both the original and amended settlement have focused (quite rightly, I’d say) on the fact that it doesn’t actually settle what the case was all about originally — whether scanning books and showing short portions is fair use or not. Instead, the settlement created a new business arrangement that conferred those rights without exploring the legality. On page 6, the DOJ seems to suggest that the court should only rule on the scanning and short display portion — and that on this issue, it could indeed rule to cover those “absent” or “orphan” rights-holders:
The provisions that settle the specific allegations of infringement in the Class Complaint – those that allow Google to scan millions of copyrighted works and to make available small portions of such works in response to search requests – address disputes within the Court’s subject matter jurisdiction. Those aspects of the ASA are based on specific conduct that falls squarely within the scope of the case made through the pleadings. There are strong arguments that an appropriate set of publisher and author class representatives can adequately represent all members of the class with respect to reaching a settlement as to the uses of their works challenged in the litigation.
In contrast, provisions of the agreement that would allow Google to actually sell full copies of works online don’t seem to make sense, since Google didn’t do anything like this in the first place. It wasn’t sued over this:
The broader aspects of the ASA stand on a somewhat different footing. There has not been – and simply could not be – any allegation in this litigation that Google has sold full access to works for which it lacks the right to do so, or even that such activity was threatened. Indeed, selling such access would have been legally indefensible , and thus would have been at odds with Google’s entire pre-settlement book search strategy, which was premised upon staying within colorable “fair use” grounds. With very good reason, therefore, Google consciously avoided creating precisely the factual predicate that might support the settlement of book- and subscription-selling claims. The business models that the ASA authorizes therefore relate to activities in which Google never engaged or threatened to engage, and thus claims of copyright infringement that could not have been brought. Although Rule 23 does not require the Court to survey the claims of every class member to determine if they are ripe, there are serious questions about whether a settlement that resolves future claims by absent class members for activities well beyond the facts underlying the complaint can meet the first prong of the Firefighters test.
Remember earlier when I asked if the DOJ was saying that scanning to make something searchable was a copyright violation? This tells me no. This section seems to suggest that the real red flag would have been if Google had reprinted book for sale. THAT would have been legally indefensible.
From page 8:
Here, in contrast, the ASA authorizes future activities beyond the scope of the conduct alleged in the complaint that do not remedy injuries plaintiffs suffered in the past, nor do they seek to prevent future injuries. Rather, these provisions provide the defendant with benefits it could not have secured either through trial or even through normal private negotiations.
Again, more of the “this settlement goes to far” theme.
From page 9:
The ASA seeks to carve out an exception from the Act’s normal rules and presumptions, which require a rightsholder to affirmatively grant permission for the kinds of uses contemplated by the ASA. The parties claim that creating an opt-out exception would better serve the purposes of the Constitution’s Copyright Clause by promoting the progress of science and the useful arts. That, however, is a judgment better suited for legislative consideration, rather than one for courts to make in the context of approving a settlement under Rule 23.
Here, the opt-out discussion is about how the agreement would allow books to be shown or sold unless the authors opted-out (unlike the opt-out of just scanning, that I discussed before). Many critics have wondered why the agreement doesn’t go the other way — allow Google to show or sell only books with explicit permission. All the parties to the agreement have countered with so many orphan works, it’s easier to go opt-out. Then as a new registry is created to hunt down orphan rights-holders, they can choose to opt-out. Certainly switching to opt-in would have made getting some agreement in place much easier.
From page 10, although the DOJ recognizes that opt-out would benefit the public more, that can’t come at the expense of rights-holders:
The United States recognizes that it is the ASA’s broad grant of rights to Google, coupled with the settlement’s opt-out requirements, that allows for the use of the largest possible universe of digital works. The United States also recognizes that, although Google’s activities are commercially motivated, its business plan would generate numerous public benefits. The ASA would achieve these benefits, however, in spite of and not in furtherance of the basic premises of the Copyright Act.
From page 12, the DOJ raises the issue that the court seeks to impose a settlement on works on non-US authors who may not be fully represented:
Nonetheless, there are significant numbers of foreign authors from outside Canada, the UK, and Australia whose works were published in one of those countries or registered in the United States, and thus are subject to the ASA, even though the rightsholders may not have been represented by the new associational plaintiffs. This point is made clear by foreign governments, which object to the settlement.
Further pages get into pricing issues that frankly get beyond me. Suffice to say, the DOJ is worried there are anti-competitive issues involved.
Page 21 comes back to the issue that the settlement grants rights to Google that it wasn’t originally sued over, and which competitors would be hard-pressed to gain:
There is no serious contention that Google’s competitors are likely to obtain comparable rights independently. For example, Amazon – Google’s likely chief rival digital book distributor were the ASA to be approved – began scanning copyright-protected books in 2002, after firstsecuring permission of the works’ rightsholder(s). To date, Amazon has amassed a library of approximately three million digital titles. See Amazon.com, Inc. Obj. at 1 (D.E. 206). This impressive number pales in comparison to the tens of millions of books Google has scanned or is poised to scan if the ASA is approved. The suggestion that a competitor should follow Google’s lead by copying books en masse without permission in the hope of prompting a class action suit to be settled on terms comparable to the ASA is poor public policy and not something the antitrust laws require a competitor to do.
I love this part. It’s so blunt and straight-forward. Others could scan books just like Google did and show short snippets. If they did, maybe they’d get sued. But even then, there’s no guarantee a settlement would allow them to sell those books in the way Google will be allowed.
On page 22, the DOJ goes to Google’s core search business:
Finally, wholly apart from the new business ventures contemplated by the ASA, Google’s exclusive access to millions and millions of books may well benefit Google’s existing online search business. Google already holds a relatively dominant market share in that market. That dominance may be further entrenched by its exclusive access to content through the ASA. Content that can be discovered by only one search engine offers that search engine at least some protection from competition. This outcome has not been achieved by a technological advance in search or by operation of normal market forces; rather, it is the direct product of scanning millions of books without the copyright holders’ consent and then using Rule 23 to achieve results not otherwise obtainable in the market.
Here, the DOJ’s argument is weaker, I’d say. Amazon or Microsoft could scan books just as Google did to make them searchable. They could have been doing that for years. They might not have because they interpreted fair use differently. Or, in the case of Microsoft, simply because they didn’t think book search was commercially attractive enough, as it said when it backed out of scanning in 2008 (see Microsoft Burns Book Search – Lacks “High Commercial Intent”). The agreement wouldn’t prevent any competitor from scanning for searching purposes.
However, the agreement — in allowing Google to display books scanned — certainly would give it a more compelling book search service. And as previously covered, it would be much harder to nearly impossible for competitors to get those display rights.
On page 23, some DOJ suggestions:
The United States continues to believe that an approvable settlement may be achievable here, for example, by requiring rightsholders to “opt-in” to the settlement or by narrowing both the scope of the plaintiff class and the relief, to better align with the actual dispute underlying the case….
- Change from opt-out to opt-in
- Narrow rights granted
- Stick closer to what the case was about: can you scan for searching purposes
Also on that page, there’s a suggestion of a waiting period so as to reduce the number of works that might get used without explicit permission:
The United States believes there would be real value in creating a meaningful waiting period before Google may commercially exploit out-of-print works without the permission of the rightsholder (e.g., two years from the time the title is publicly listed in the Registry). Such a waiting period, combined with efforts of the Registry to locate rightsholders, may reduce the number of rightsholders whose works would be exploited without their knowledge
Further points also look to narrow the use of orphan works. And that’s it.