Copyright Dogma and the Denied Google Books Settlement

Last month Judge Chin denied the proposed Google Books Settlement (the Amended Settlement Agreement, or ASA). While I’m pleased with the outcome, I’m troubled with some aspects of the opinion.

Setting aside the issues of adequacy of representation, notice, privacy, and whether a class action settlement should be used to establish future and ongoing arrangements, etc, I’m pleased with the outcome because in my view, the main problem with the ASA was its potentially anti-competitive outcomes. The forward-looking element of the ASA consisted of two separate parts: first, it created what is, effectively, a new Collective Rights Organization for works whose copyright owners are known and locatable (I’ll call them the “locatables”), aka The Book Rights Registry (“the Registry”), then it offered a solution to the orphan books problem (books whose copyright owners are unknown or unlocatable). The first part involved a problematic solution to a problem that does not exist. The second part, however, involved a solution, albeit flawed, to a real problem. Both aspects were subject to an opt-out rule, namely they would not apply to copyright owners who actively opt-out of the arrangement.

The first aspect of the ASA, dealing with the sale of locatables, is difficult to justify. The Registry would be essentially a new CRO with the power to set collectively the prices and terms for the sale of digital books. The traditional justification for CROs is a perceived market failure arising from the presumed impracticality of direct contacts between users and owners and the prohibitive costs of monitoring and enforcement, that is why Richard Posner calls them “benign cartels”. I have written several papers questioning the validity of this justification, but in any event I think this justification is clearly inapplicable to the Google case for two main reasons.

First, there’s no problem for even the smallest copyright owner to locate and contact Google if she wants to (if she doesn’t know how to find Google, she can google Google). That means that there aren’t prohibitive transaction costs preventing individual copyright owners and Google from transacting over the terms of commercializing their works. Second, it’s very easy for a copyright owner to monitor whether Google used the work without authorization (she can google the work and immediately find out; she can even create an alert), and there’s no problem to enforce the copyright if it has been infringed (proof: the class action). It turns out that the perceived market failure which may justify CROs (and the associated reduction in competition) in some other cases simply doesn’t exist in the case of in-print works (or out-of-print but whose owners are known and locatable). Therefore, the ASA offered an anti-competitive solution to a problem that doesn’t exist. The sale of such books should be subject to copyright law’s ordinary opt-in rule.

Unlike the first part, the second part, dealing with orphan works, addresses with a real and substantial problem. The problem stems from the fact that the market system that copyright law envisages fails when it becomes impossible or impractical to obtain the copyright owner’s prior permission. The orphan works problem is anything but new, and one could just shrug and say that the problem only reflects the fact that we’re living in a world of second best: it’s the price we have to pay if we want to have a copyright system at all. This might have been true when the opportunities for reuse of out-of-print works are limited, but the scale of the Google Books project clearly demonstrates that it is a problem that our society cannot be sanguine about–we would be losing much if we ignore it. There’s a serious market failure here and ASA tried so solve the problem. It would permit Google to sell copies of orphan works until their owners show up and demand otherwise. Meanwhile, Google will transfer 63% of its earnings from the sale of such books not to the copyright owners (because they are unlocatable), but to the Registry, which, with the aid of the Unclaimed Works Fiduciary, will attempt to locate these owners and transfer the money to them (after deducting its expenses). If those owners cannot be found, the money will eventually be given to literary based charities.

One of the main objections to this aspect of the ASA was that it would effectively turn Google into a monopoly over the commercial use of orphan works. This is a serious concern, but we must admit that if the alternative to monopoly (one seller) is “nullopoly” (no seller), we’re better off with a monopoly, but this choice is false. Note, however, that the monopolistic solution doesn’t really solve the problem of orphans, it only pretends to. It creates a mechanism for collecting money for the use of orphan works, but it doesn’t solve the problem of distributing the money to the copyright owners, which are, by definition, unknown or unlocatable. So we end up with a monopolistic solution that doesn’t really solve the problem. Even if Google’s competitors were permitted to strike similar deals with the Registry/Fiduciary for the use of orphan works the monopolistic outcome would be the same: rather than the Google becoming the monopoly and sharing its profit with the Registry, the Registry alone will become the monopolist.

But on a more fundamental level, however, the problem with the ASA, is that it is captive of a dogmatic view of copyright that inevitably leaves us only with the choice between monopoly and nullopoly. As we shall see below, Judge Chin is also captive of the dogma, yet he falls on the nullopoly side. Let’s talk about the dogma first.

The dogma

At its core, copyright law is based on a very simple logic: market logic. The law grants certain exclusive rights in creative works for a limited time. The expectation is that such rights will then be voluntarily exchanged in a decentralized market place, and that this exchange will provide the financial incentives for creating the works in the first place.

As a general rule in copyright, a user should contact the owner and ask permission before using the work. Let’s call it the permission-first rule. Generally in copyright, permission-first makes lots of sense. It makes sense because ordinarily it is much less costly for a prospective user to approach the owner and ask permission before using, than for an owner to locate and all actual or potential users and ask them to pay. In the paradigmatic case, the owner is one, locatable and stable, whereas users could many and varied, scattered and fleeting. Moreover, because the work is non-rivalrous, the user can obtain the work without negotiating with the owner at all, meaning that under a permission-later rule, many users would use without ever getting permission, hoping that their use would go unnoticed, and using any tactic to delay payment once contacted.

But generally is not always, and not every case is paradigmatic. A permission-first rule only makes sense if the cost of getting permission is indeed low, or at least low relative to the expected value of using the work. If the cost of obtaining prior permission becomes prohibitively high, then a system demanding prior permission would result in a market failure—the very market the copyright law seeks to create. In such cases we need an alternative rule.

The problem with many mass digitization projects, especially when they involve orphan, partly orphan, or quasi-orphan works, is that seeking permission first is either impossible, or at least prohibitively costly. That’s why for many of these cases, permission-first is not a workable rule.

The reason that so far solving the orphan works problem is that we often treat the permission-first rule as a dogma, as a hard and fast rule, not to be ever disputed, doubted or diverged from. For some copyright purists, the rule should be adhered to regardless of the consequences, because “permission-first” is the essence of copyright, anything else would be “turning copyright on its head.”

The ASA slightly deviated from the dogma, but it’s still its captive. It recognized that an orthodox adherence to the permission-first rule would lead to an orphan works nullopoly, and offered a modified rule. Rather than seeking permission from the owner, Google would be able to use the works by paying an organization representing other copyright owners, who will try to locate the owner or give the money to literary based charities. But as I noted above, the solution doesn’t really solve the problem, it only shifts the problem of locating the unlocatables from Google to the Registry or the Fiduciary. Moreover, it raises the following question: if it’s OK to use a work without getting permission first, as long as you’re willing to pay later, there could be a much simpler solution: Google could simply announce that it will pay the unlocatable owner directly should she ever show up, and we could create some incentives making sure that it will actually do it (more on which below). Such a rule could be applicable not only to Google, but to any other user thus reducing the orhpan-works monopoly concerns.

The logic of the ASA seems to be the same logic that led the widespread sale of indulgences by the Catholic church in the middle ages. Rather than dealing with the fact that people do commit sins, the Church absolved catholics of their sins, provided they paid the right amount to the Church. This proved an easier solution. After all, if people commit sins all the time there might be some problem with the dogma itself or with the institutions that are aimed at discouraging them from doing that, and of course, selling indulgences incidentally helped the Church fund itself (as well as the lifesytle of some of its priests).

Similarly, the ASA remains true to the copyright dogma by accepting that it’s a sin to use a work without the copyright owner’s prior permission (regardless of whether permission can be obtained or not), but then absolving the sinner as long as he pays money to a the established copyright owners’ church.

Therefore I’m pleased with the rejection of the ASA, which provided one anti-competitive solution to a problem that doesn’t exist and another anti-competitive fictional solution to a problem that does exist.

What’s next?

Now, after the rejection of the settlement, the question is what’s next? Judge Chin leaves the door open to an approval of another future-looking settlement that would be based on opt-in rather than opt-out. That’s absolutely fine with regard to the locatables, except that an opt-in system for such works does not require any settlement, or any other type of intervention. Google and every copyright owner can just do it if they wish (and many of them have already done so).

An opt-in solution, however, is inapplicable to orphan works, or at least runs to risk that the majority of such work would remain unused for many years to come. In fairness, Judge Chin is aware of that but he takes the position that Congress rather than courts should solve the orphan works problem. Unfortunately, some aspects of the decision could make it more difficult to devise a sensible solution to the problem.

Recall that the orphan works problem has been difficult to solve because of the dogmatic adherence to the permission-first rule, and the difficulty of distinguishing between the cases in which the rule makes sense and those in which it doesn’t. The decision doesn’t attempt to make such distinction either. Rather, it is rife with pronouncements of very strict adherence to the dogma. At one point Judge Chin states that

Certain objectors contend that the ASA’s opt-out provisions would grant Google the ability to expropriate the rights of copyright ownerswho have not agreed to transfer those rights. … The argument may have merit.

At another point he states that

A copyright owner’s right to exclude others from using his property is fundamental and beyond dispute. … As counsel for Amazon argued: “[T]he law of the United States is a copyright owner may sit back, do nothing and enjoy his property rights untrammeled by others exploiting his works without permission.”

And then

[I]t is incongruous with the purpose of the copyright laws to place the onus on copyright owners to come forward to protect their rights when Google copied their works without first seeking their permission.

While these statements make a lot of sense when the owners are locatable and the market can work, strictly adhering to them when they are unlocatable no longer makes sense. Even if Congress can create a different rule for unlocatable works, its job would become much more difficult if any deviation from the permission-first rule is described as incongruent with the purpose of copyright law, as a form of expropriation, and if it is considered to be “beyond dispute” that a copyright owner may sit back and do nothing no matter what. After all, it is hard to imagine any solution to the orphan works problem that does not involve using them without getting permission first.

Another solution: remedy tweaking

I believe that there’s a much more simple way to solve the orphan works problem, one that remains true to copyright law’s principles rather than its dogma. This solution requires very little legislative reform, if any. We can address the orphan works problem by tweaking the remedies available for infringement. What I mean by that is that the fact that the owner of a copyright has been unlocatable, should be a consideration taken into account when determining the appropriate remedy in a case of copyright infringement. In other words, we do not create an exception for the use of orphan works, but allow courts do take this fact into consideration at the remedy stage.

Why is remedy tweaking a solution? The main reason why the orphan works problem is so serious is that the range of remedies and particularly the threat of injunction impose a huge risk on the user. Suppose you’re a film-maker and you want to incorporate an orphan work in your film, and you would be willing to pay license fees if you could locate the owner. If you use the work without permission and your film becomes a success, and the copyright owner suddenly shows up, then you’re in big trouble. If the copyright owner can get an injunction stopping you from distributing and showing the film, she can really hold you up and then offer to lift the injunction in return for a payment that will eat up whatever you may have expected to earn. Statutory damages pose a similar problem, and can accumulate to substantial sums, especially in large scale projects such as Google Books. Ordinarily, exposing users to this risk is a good thing because this exposure motivates them to seek permission first. But if the owner is unlocatable, this risk serves no useful purpose.

I propose that the fact that the owner was unlocatable should be a factor considered by the court when determining the proper remedy after a finding of infringement and will deny an injunction in appropriate cases. The court may still order to pay damages, or account for the profit attributable to the infringed work, but if these approximate what the parties would have agreed had a license been negotiated in advance, the orphan works problem would be seriously diminished. Obviously courts should also have some discretion to deny or reduce statutory damages for this solution to work.

Remedy tweaking will shift some of the risk of using orphan works from the user to the owner. It will adjust the user’s duty to seek ownership information and permission with a corresponding duty of the owner to provide such information. This should motivate copyright owners to make themselves locatable: to de-orphan their works. After all, the orphan works problem is a problem of those copyright owners own making, and they are the ones who can mitigate it at the lowest cost. Tort law takes into account a plaintiff’s contributory or comparative negligence; contract law demands that plaintiff take reasonable effort to mitigate their damage before they can recover its full extent. Property law sometimes imposes a duty to provide notice before a right can be enforced. The law does not recognize an unfettered right to sit back and do nothing in all circumstances and regardless of the consequences. Copyright law should not recognize such right as well.

Applied to Google Books, that’s how the case should go forward:

The original GBS (scanning and snippets) should be regarded as fair use, or the parties may settle this issue;
Next, Google announces its intention to display more than snippets and to sell copies. It invites publishers and every other copyright owner to negotiate the terms of such deals;
Google announces a date after which it may start selling copies and invites copyright owners to negotiate (opt in) or opt out;
After the deadline, Google starts selling copies under the agreed terms with respect to works whose owners opted-in, and refrains from selling copies of works whose owners opted-out. But what about the works whose owners remained silent, i.e., neither opted in nor opted out?
If it chooses to sell copies of the silent it may still be liable for copyright infringement. However, the scope of its liability and the remedy may depend on the locatability of the copyright owner. If locatable, Google should be liable just like any infringer. If the owner was not locatable (until she sues, of course), the remedy for the past infringement should be limited to actual damages (and maybe part of the work-specific profit made by Google, but not statutory damages). Prospectively, it depends on what the owner chooses to do. If the owner doesn’t want her work sold by Google, it opts out and from now on Google would be fully liable if it continues to sell copies. If she opts in, then there’s an agreement between the copyright owner and Google and the issue resolved.
Of course, the same rules apply to any company that wants to engage in the same or similar activities, not only to Google.
We get all the benefits of the settlement without the monopoly costs.

Search inside