Last week, U.S. District Court Judge Denny Chin held a much-anticipated “fairness hearing” during which a chorus of interested parties spoke both for and against Google’s proposed settlement to its Book Search litigation. My own view aligns closely to the view articulated there by the Department of Justice: The proposed settlement cannot stand because its provisions are fundamentally inconsistent with not just copyright law but also antitrust law and the class action mechanism itself.
Here, however, I want to look beyond the fight over the proposed settlement and instead think about what seems increasingly likely: that the settlement will be rejected and Google will have no choice but to head back to court to actually litigate the question of whether its practice of scanning copyrighted works and then using those scans to produce on-screen “snippets” is in fact fair use.
Section 107 of the Copyright Act empowers a court to excuse, on public policy grounds, acts that would otherwise be deemed to impermissibly infringe a copyright holder’s exclusive rights. At a high level, the purpose of this flexible “fair use” doctrine is to allow courts to waive off copyright infringement in instances where the costs of protection seem to outweigh the benefits. One way to think about fair use is to recognize that there are a large number of rights and revenues that could plausibly be assigned to authors; and so, if the end goal of the copyright system is only to move a certain amount of value anyway, copyright can and should choose the subset of those rights and revenues that will transfer whatever value is necessary from an incentive perspective but do so at the lowest external cost in terms of avoidable, undesirable third-party implications.
A common misconception is that the fair use doctrine excuses any infringing use that is sufficiently valuable to society. But that is not the case. The litigation involving Michigan Document Services provides a helpful example. (Princeton University Press v. Michigan Document Services, Inc., 99 F.3d 1381 (6th Cir. 1996)). The infringing products in that dispute were packets of photocopied materials. The packets were made up of excerpts from articles and books, those excerpts having been chosen by university professors for use in their specific university classes. The accused infringer was the copy center that duplicated the excerpts and ultimately sold those packets to students.
Clearly, the infringing products in that fight were socially attractive. They were products that facilitated classroom teaching, and they were produced at the direction of university faculty. Yet, the copy center that produced the packets was found guilty of copyright infringement and specifically had its fair use defense rejected.
Why was the copy center denied the protection of the fair use doctrine? Because fair use is not simply an inquiry into whether the accused use is valuable. Instead, it is an inquiry into whether the owner of the infringed copyright should have influence over when and how the accused use takes place. To deny fair use in the Michigan Document Services dispute, then, was not to in any way speak ill of the infringing products at issue. Photocopied university materials are tremendously worthwhile products, and no one disputes that fact. To deny fair use was instead to decide that these beneficial but infringing products ought to fall under copyright holders’ sphere of influence, with the relevant copyright holders having the right to influence who produces the packets, under what terms, and how much everyone profits from that interaction.
Two intuitive considerations guided the court in Michigan Document Services and indeed more generally seem to helpfully frame fair use analysis. The first of these intuitive considerations is the degree to which a finding of fair use would undermine the incentives copyright law endeavors to create. Copyright law in general recognizes rights in authors in order to motivate authors to create, disseminate, and in other ways develop their work. Fair use is unattractive to the extent that it interferes with that goal. Put differently, the issue here is whether repeated findings of fair use in a category like the one at issue would over the long run reduce author motivation to do things like create their work, share their work publicly, and search for new related projects. If so, fair use is on this ground unattractive, as it undermines the very incentives copyright law endeavors to create.
In Michigan Document Services, this first consideration clearly cut against fair use. Works that would be included in university course packets would often also be works whose primary audience would be the university students. If the authors of these works could not profit from their use in class, it was not clear from where profit would otherwise come. The prospect of fair use, then, in this case came with an obvious impact on the core incentives copyright law attempts to create.
The second intuitive consideration relevant to fair use analysis is the degree to which uses like the one at issue would thrive even without the protection of fair use. In the Michigan Document Services example, there were two plausible concerns along these lines: It might have been that it was just too expensive for the copy center to identify and contact each relevant copyright holder, and thus enforcing copyright would mean the death of university course packets; or it might have been that, once contacted, the copyright holders would ask so high a price that an inefficiently small number of course packets would end up commissioned.
Neither concern, though, ended up resonating with the court. The costs of identifying and then contacting the relevant copyright holders seemed likely to be adequately addressed by licensing intermediaries like the Copyright Clearance Center. These entities reduce costs by offering licensees one-stop shopping for a large number of titles and offering licensors a convenient way to approach and collect from a large number of would-be licensees. Prices, meanwhile, seemed adequately constrained by market competition. After all, no specific author has much market power vis-à-vis academic users, because a faculty member can always assign different reading if the originally chosen work is available only at an unreasonable price or subject to unreasonable terms.
Return now to Google Book Search. To the extent that Google invokes fair use to defend the entire Google Book Search program, that defense must fail. With respect to the first intuitive consideration, a finding that Google Book Search is fair use would clearly undermine the incentives copyright law endeavors to create. After all, copyright’s incentive system only works if authors have confidence that their rights will stay relevant over time.
So, while there were in the past authors who wrote their novels before the modern motion picture was invented – and while copyright law could have denied those authors any rights over motion picture versions of their works – copyright law instead kept pace with technology. In that way it made clear that even if times change, and even if authors cannot precisely predict how they will ultimately want to exploit their work, the relevance of the promised rights will persist. That same concept governs here. It is entirely plausible that, tomorrow, a substantial portion of the economic value of books will derive from various types of digital exploitation. Google’s fair use argument threatens to take that burgeoning value away; that clearly weighs against a finding of fair use.
With respect to the second intuitive consideration, meanwhile, a finding of fair use is not critical in terms of facilitating the creation of the Google search engine, because a great deal of the project could be accomplished through negotiated, consensual transactions. Each major publisher, for example, could act as a helpful intermediary, negotiating terms with Google on behalf of all the authors still under contract with that particular publisher. And even individual authors could opt into the program, for instance if Google were to create a website where interested authors could agree to participate or even upload electronic copies of their work.
Similarly, there is no reason to believe that licensing rates would be inefficiently high. As I have pointed out already, Google can build a tremendously useful resource even if at the start it has only 30 percent of the world’s books. That is important because it means that no single author has significant market power vis-à-vis Google; an author who demands a disproportionate share of the project’s profit or undue involvement in the project’s design can simply be left out of the database until that author makes a more reasonable offer.
Were Google to concede infringement for many of the works at issue but invoke fair use only to more narrowly excuse its use of books in instances where the costs of identifying the copyright holder is prohibitive, however, Google’s claim would be strong. It is enormously difficult to acquire permission with respect to books that are significantly old, or books for which the current ownership of rights is hopelessly unclear. As applied to that class of work, Google might be right that the only way to use those books is to invoke fair use.
Google could also fairly point out that the harm to that subclass of authors is small, because authors who are so difficult to identify are likely also not authors who are actively profiting from or marketing their work. The main weakness with this argument is that Google in practice makes no effort to distinguish these so-called “orphan” works from the many works for which permission would be practical. A court might require Google to undertake reasonable efforts along these lines as a condition of any fair use finding, or, similarly, might limit the fair use defense to cover only books that fit one of these troubling categories.
Google ands its supporters hear analysis like this and contend that fair use ought to excuse the project anyway. One argument they make is that Google’s project is good for authors because the existence of a comprehensive search engine will likely increase demand for books. I hear that point, but I doubt that this argument would much move a court. After all, this argument only tells us that authors are likely better off in a world where Google’s project is fair use as compared to a world where no one builds search engines at all. That, however, is not the relevant comparison. Denying fair use here is not tantamount to banning search engines, now and forever. Denying fair use instead forces search engine providers to negotiate with copyright holders and then come to agreement about how to share revenue, how to protect the book database from hackers, and so on.
A second argument that Google and its supporters put forward is the argument that Google’s use should be deemed fair because Google allows copyright holders to opt out of the program. Specifically, a relevant copyright holder can notify Google that it does not want a particular book included in the database, and Google has promised to respect that request. This opt-out provision certainly makes the Google project more attractive than it would otherwise be, but again my view is that this feature does not significantly affect the overall fair use analysis.
The reason is the fundamental insight that fair use considers “not only the extent of market harm caused by the particular actions of the alleged infringer, but also whether unrestricted and widespread conduct of the sort engaged in by the defendant would result in a substantially adverse impact on the potential market for the original.” (Campbell v. Acuff-Rose Music, Inc., 510 U.S. 569, 590 (1994)). An opt-out works well in a world where Google is the only infringer. Authors could in that case at low cost find out about the Google project and communicate their desire to be left out if need be. This would be efficient, in fact, because the costs to authors in finding Google would likely be much smaller than the costs Google would incur were it required to find each individual copyright holder.
When the analysis shifts to focus on the possibility of countless Google-like opt-out programs, however, the conclusions reverse. In a world with a large and ever-changing list of opt-out projects – today a search from Google, tomorrow some new startup from Microsoft, the next day some fan site based in Boston, and on and on – authors would be forced to invest substantial sums finding each new project and notifying each as to whether they desire to participate.
The problem would be even worse if some of those opt-out programs were designed strategically to make things difficult on authors, for instance imposing high standards of proof before acknowledging that an opt-out really came from the correct copyright holder. (Infringers have an incentive to do just that, because in an opt-out system infringers benefit if authors find it too expensive to actually engage in the mechanism of opting out.)
So where does all that leave us? For now, my comments are admittedly premature. The Google Book Search case is still today a case about the parties’ proposed settlement; the fair use doctrine will become relevant only if that settlement is scrapped and the parties as a result have no choice but to return to the copyright case that first launched this fight. However, if we get there – and I hope we do – I think the good news for Google is that the fair use doctrine is up to the task. But the bad news is that their current practices reach far beyond anything that doctrine should or will cover.
* By way of disclosure, Professor Lichtman prepared an expert report on the issues discussed in this Essay and filed it as part of the Google Books litigation in France. Nevertheless, the views expressed here – and there, frankly! – are entirely his own.
Want to comment on this article? Submit your comments below.
(Please note: All comments are screened for language prior to posting.)