Google Book Search and the Transaction Costs of Consent

Larry Lessig posted this week a synchronized set of slides and audio addressing Google Book Search as fair use. The talk runs for 30 minutes but ultimately the core point is simple and I think relatively conventional: transactions costs matter for the scope of allowed fair use. The suggestion here is a transaction-cost based opt-out model: for low-cost of consent holders, GBS has to ask permission, but for high-cost of consent holders, GBS need not ask permission, but instead those holders have to opt out.

 

Larry splits the universe of books for Google book Search into three categories: books in the public domain, thought to be roughly 16% of the relevant works; copyrighted works not in print (75%); and copyrighted works in print, the remaining 9%. Google need not get consent for public domain works (though, as I suggested in my last post on this, Google seems to be taking a conservative approach on public domain works (and also perhaps an economic one as Saul Levmore suggested in a comment on that post)).

 

Google seems to be working with copyright holders on the works in print, and Larry doesn’t really address whether fair use might allow Google to do more than it is currently doing with those works, even without the consent of the copyright holders. As Larry notes correctly, fair use is not based on consent; fair use instead is based on a bundle-of-sticks model of property rights. We give some of the property right sticks to the copyright holder and other sticks to the public. The public gets to pick up its sticks without asking permission from the copyright holder. That said, fair use is notoriously hard to pin down, so Google might prefer to get the consent of copyright holders to use their works. License rather than litigate.

 

But it is the intermediate category where matters gets sticky and fair use needs to do its work. Needs to do so because Google does not have unfettered access, as it does for public domain works, and cannot negotiate for access straightforwardly as it can with many copyrighted works. There is where fair use comes in.

 

One fair use tradition is highly economic and focuses on the transaction costs of fair use. If negotiating for use is very difficult—if the transaction costs of consent are high—we can’t rely on voluntary licensing to facilitate use. We need to assign use, either facilitating use as a fair use or limiting use. Given the costs of asking permission, the use won’t happen if we assign control over the use to the copyright holder.

 

An assignment of a use right to GBS need not last forever. Transaction costs can change, and authors and publishers can take steps to lower the transaction costs of consent and limit fair use by GBS (and others who might want to build these databases). Notification by an author or publisher with regard to a particular work would change the transaction costs sufficiently so that use that started out as fair might cease to be so after notification. This tracks the notice-and-takedown procedure we see in section 512 of the copyright law for Internet service providers and also matches the approach taken to filtering in the Napster case, where the copyright holders ultimately provided lists to Napster for filtering.

 

Should this mean that GBS should be able to use all works at least until GBS receives a notice from the copyright holder? I don’t think so. GBS need not get consent for public domain works; should, at least in a transactions-cost framework, get consent for low-cost of consent works; and should be engaging in fair use for high-transaction cost consent works. Once GBS has received a notification from a copyright holder of a high-transaction cost work, the costs of consent have dropped, and the fair-use right should end with it. If the number of book search sites proliferate, that will increase the burden on authors, though the technology will matter here too, as the Author’s Guild could jump in and provide a book-search site search engine for their members to reduce the cost of opting out from multiple sites.

 

How should Google present these intermediate works? It is currently giving the least amount of access to these works: not the full access it gives to public domain works or even the multiple-page access it is giving for the negotiated copyrighted works. Instead, Google is currently presenting these works in its snippet view. Instead, GBS should track the deals that it has struck with copyright holders. On the transaction costs view, we should implement the deal that would be struck hypothetically if the costs of consent didn’t prevent that deal from happening. And that would mean that if most of the low t-cost copyright holders are exiting the program, we should probably exclude these high t-cost works as well, but I doubt that we are seeing much full exit from GBS.

 

Do note also that this might be seen by many as a very narrow approach to fair use. We might think that a lot more is at stake than just the transaction costs of consent, but even in that framework, we should be able to move GBS forward very aggressively.

 

Comments

[Doug Lichtman]

Randy -

Can you clarify something that your remarks and Larry's slides both seem to leave a little vague?

If I am a copyright holder with my work in print, can Google do it's "snippet" view without my permission? I think not. I think I heard Larry say not. I think I hear you say not. That's a pretty important point to the litigation. So can you clarify?

Next, if I am a copyight holder with my work out of print, but "reasonable efforts" would succeed in finding me, again are snippets okay without my permission? (Again, I think not. I think I heard Larry say not. I think I hear you say not. Agree?)

Note that these two categories are substantially important. For one thing, the copyright holders I describe are likely worried about security issues. If Google can do snippets, that means that Google will be storing digital copies of these works, and that storage could be compromised.

These obviously are not the only categories in dispute, and gosh it's going to be hard to define "reasonable efforts". But wow. If consensus is forming on these points, maybe the parties will be able to get this thing settled soon, as Ed Felten has (optimistically) predicted over at Freedom to Tinker.

[Randy Picker]

To take your questions:

On snippet use of works-in-print: To be clear, I think the relevant category isn’t in print, but instead goes to the transaction costs of negotiating consent, which may or may not match well with in print.

But on the low t-cost of consent category, my post doesn’t take a position on whether snippet use without consent would be fair use. I haven’t thought that through yet fully. (So I hope that my post isn’t vague on this: I don’t mean to be addressing this question yet.) We think that quoting from those works would probably be fair use—Larry makes this point in the talk—and certainly some limited uses of the in-print works would be fair. I haven’t figured out yet what I think defines that boundary. I also don’t think that Google is pushing to do snippet use for these works, but instead is moving through the process of consent, so this isn’t the core issue at stake. Yes?

On the second question, on reasonable efforts: Same answer, I guess. So I separate into low-cost and high-cost of consent copyright holders, and that will match up somewhat with your articulation of reasonable efforts. Certainly hard to define and we might look for a different formulation. But again Google itself must be implementing some version of this, as they want to work with copyright holders. And again, I haven’t figured out yet whether snippet use would be permitted even without permission.

My post tries to push Google forward on the intermediate category. Google seems to have embraced a license model for low t-costs work, and are doing that even if fair use might let them do more (so figuring out the outer boundary of fair use for these works doesn’t seem to matter so much to this case). On the high t-cost works, we would do better if Google would just apply to those the access rules that it is negotiating for the low t-cost works.

[geoffrey manne]

I agree that the dividing line between fair use and not fair use is essentially a function of transaction costs. But, as you say, determining which side of the line copyright holder X is on is a difficult matter. I have a potential (imperfect) solution.

The gist is this: Google announces today, at t0, that for all books not in the public domain, it will run its GBS scheme with an opt-in license -- meaning it will only copy, scan and make searchable (or readable) copyrighted books for which it has negotiated a license or for which permission has been explicitly granted without negotiation. It also announces that at t1 (say, in 5 years) it will switch to a quasi-opt-out system, at which point Google will exercise its fair use right to use all remaining works unless Google agrees to forbear at the urging/payment of the copyright holder.

With what result? The low transaction cost copyright holders (and here the really big cost is determining who, in fact, owns the right) will come forward during the opt-in period in order to negotiate a royalty for the use of their work (or explicitly to deny access to Google for the opt-in period). The high transaction cost owners will no do so because the cost of determining the extent of the right and engaging in negotiation will exceed the value of the expected royalty (or the value to the owner/s of declining permission). Because all owners/potential owners know that at t1 they will lose their right (to say nothing of 5 years worth of royalties), there will be a strong incentive for the low transaction cost owners to come forward during the opt in period.

This means the dividing line between high transaction cost/low transaction cost will be determined in large part by the length of the opt-in period. I think Google has some reputational incentives not to set this period too short, but there is some argument that the period should be set by a third party (perhaps through legislation (the drafting of which would itself be a substantial problem)).

There is much more to hash out, of course (like the last period problem with respect to Google's incentive to negotiate), and I am working on an essay on the topic, but I thought I would just lay out the barest basics here.

[Steve Dispensa]

Interesting post. I have a couple of concerns, though:

The definition of "in print" seems to be decreasingly relevant in a digital world. I'm not sure it makes sense to rely so heavily on that notion as a definition of t-cost. Besides, I would hope that the concepts you're developing here would apply to works that aren't published by traditional houses - i.e. blogs, etc.

The security concerns are also a little fishy to me. We're seeing the analog of this in the music world, for instance: virtually every CD I've ever tried to buy is available on one music service or another (although obviously with the consent of the rights holders). Beyond that, it is fairly clear that I can rip a music CD onto my computer - what if that computer is compromised? I think this argument has practically been conceded by the music industry, and I'm not sure it makes any sense to reevaluate it in the dead-tree world.

I realize that this is theoretical, but I can't help but wonder about the practicality of the "shades of grey" approach based on transaction costs. Does that mean that there could, in principle, be a great many lawsuits based on whether or not the transaction cost was "low" in a given case? I see the desire to define a couple of clear categories (e.g. "in print" vs. "out of print"), but I'm not sure I see any that aren't tied to physical printing analogies.

[Ariel Katz]

Here’s my concern: if the transaction costs justification for fair use in this case is taken seriously, then publishers/authors may create a copyright collective a la ASCAP and say: “Hey, we’ve set up a collective. True, if Google had to deal with each of us individually the transaction costs would be prohibitive and fair use justified. But no longer. Now Google only needs to deal with our collective, and without our permission the use is no longer fair.”
If fair used is based solely on transaction costs grounds, then we must acknowledge that the alternative to the market failure is not necessarily a permission-free world; it may also be an authors’ guild offering GBS a blanket license at a monopoly price, and perhaps requiring any blogger who uses others’ materials to pay for it in advance by logging-in to its website and buying a license (e.g., on a per-word basis). If they can do that efficiently, then even citing others may not necessarily be fair use any longer. How do we respond to that?

[geoff manne]

Off hand, I would respond by wondering what the problem is and welcoming the development of the collective. If the intention behind a doctrine of fair use is to enable valuable uses that would otherwise be precluded by operation of the copyright law, why is this a problem instead of a solution? Unless you believe "fair use" is imbued with some inherent distributive content, the transaction cost analysis gets us where we want to be, doesn't it? The objection to the notion that "even citing others may not be fair use" is unpersuasive if negotiation to obtain permission were costless or nearly so and distribution isn't a concern. Perhaps a revisiting of the BMI case and the related consent decrees might be required in the event, but not necessarily. And, of course, there's always the no-man's-land where use isn't fair but enforcement is costly that insulates many uses about which we might worry.

[Ariel Katz]

Hi Geoff,
Following your line of argument I would set aside distributional justifications for fair use, or the fact that many fair uses create positive externalities that could decrease if the use would have had to be paid for. So we basically remain with 2 options: Option 1 - fair use which is free (no permission) and free (as in free beer); Option 2 – a collective selling a blanket license that must give a license if the user pays the prescribed fee. Which of the options is better? Option 1 is the most efficient: the materials are supplied at marginal cost and there are no t-costs of licensing. Ideally, if the collective can perfectly price discriminate the amount of use would not decrease, although the collective extracts all consumer surplus. But in practice, the collective can only imperfectly price discriminate, and there are some t-costs so compared to Option 1, Option 2 necessarily creates some deadweight loss. Of course, because under Option 2 copyright owners get more, there could be gains resulting from greater incentives to create that may outweigh these costs, but we don’t really know how at the margin this cost/benefit analysis looks like.
In addition, under the collective, each copyright owner gets her share of a collective monopoly price which is more than she would have gotten by virtue of her copyright, so we can expect a demand being made that the collective would be allowed to charge only a ‘reasonable fee’ and some tribunal to set that fee (i.e., additional costs and opportunities for rent seeking). For me, these are good enough reasons to favor Option 1.

[Doug Lichtman]

Ariel -

Your comments at the blog got me thinking, but at the end of the day
I'm not sure I'm convinced.

Start with your concern about monopoly pricing. Why do you assume that we'd see such pricing? Authors have no obligation to join any collective rights organization, so if prices got too high authors would have a strong temptation to defect and either sell on their own (ie Tom Clancy undercuts the coalition by selling his rights a la carte) or form a competing rights organization. Indeed, we might imagine each publisher offering up its slate of books as a group. That would give us competition, but also provide an economy of scale for the transaction. In short, your monopoly problem
would arise only under a poorly implemented system, and it seems wrong to tar the whole idea based on that.

As for rent seeking, meanwhile, how do you know which path produces
the least rent seeking? For instance, if we find this to be fair use, aren't we opening the floodgates for a million other technologists who will want to claim fair use for their projects too? What of frauds or malevolent copycats? And what about the copyright industry's likely
response, namely going to Congress and lobbying for more protection
and/or a clearer "fair use" line? I see rent seeking in every state of affairs, and indeed in my academic work have argued that it's almost impossible to stop these sorts of "arms races" rather than merely shifting them from one forum or issue to another. In my view, then, you assume too quickly that you can reduce rent seeking here and not as a result shift it elsewhere.

All that said, your post really did get me thinking, so my response
here is in part to say thank you for raising clever points that i had not seen before, and in part then to say that I thought
about them, am not convinced, and would love to know if maybe I'm missing something?

Many thanks.

[Ariel Katz]

Thanks Doug.
Here are some additional points that I suggest worth considering:
Authors indeed have no obligation to join any collective rights organization, but if they can get monopoly prices for their works (in excess of what they can obtain in a competitive market by virtue of their copyright alone) that’s good enough reason for them to join (just like ordinary cartels whose members aren’t obliged to join but choose to). The question therefore is whether such monopoly pricing by the collective could be stable, or whether defection would destabilize it. There are several reasons to think why a copyright collective may be less contestable than other ‘ordinary’ cartels.
One reason is that I assume that what the collective would sell is a blanket license that covers all the works of all right holders. I also assume that users need a variety of works so that more works from one author is only a poor substitute for a menu of diverse works. If that’s the case, a defecting member will not be able to attract the user away from the collective, unless there are enough defectors that the user can actually get all her needs by contracting only with the defectors.
But recall that the original assumption was that it is prohibitively costly to transact with the individual right holders. This is to say that if we accept the argument that there prohibitive t-costs out there and that collective administration and a blanket license is the solution, then we cannot simultaneously assume that potential competition from the individual right holders would erode all supra-competitive prices.
True, if a model of collective administration is adopted then we’d rather design it in such a way that right holders can always opt out and license individually (as is the case in the US, as opposed to Canada and many other countries). It may partially discipline the collective, but unfortunately only partially.
I agree with you that individual authors may reduce t-costs and create economies of scales by assigning their copyrights to publishers (and that’s in fact what they ordinarily do). But, again, I think that the underlying assumption of the t-cost argument as a justification for fair use is that even publishers that compete with each other aren’t big enough, hence the choice between fair use and collective licensing. As I argued in a recent paper about performing rights organizations I’m not at all convinced how serious the t-cost problem is, so personally I’d rather look for other, or additional, justifications for fair use if we think it's important.
I liked your point about rent seeking, but still, I would fear that once a collective is set up and has some monopoly profits under its control it gets life of its own in the rent seeking game. It is much easier to set up a monopoly than to get rid of one.

[wasiu smart]

i think GBS need to be able to sell license of the copyright to companies on behalf of the copyright holders base on an agreement between the GBS and their copyright holders for example i am a student in Nigeria and the government in my country has paved way for student entreprenuership so i plan with my team to establish a student entreprenuer network that would produce software that contains books in the financial sector and their policies so i intend to partner with google so taht learning teching and research development can an easy access for my community. i need googles response.