Skip to content

transaction costs and fair use

As Mike noted in an earlier post, Larry Lessig’s Google Print presentation has sparked some interesting conversation about fair use and the transaction cost explanation for some of the uses that copyright law deems fair. The explanation is rather simple – when transaction costs are prohibitive, an efficient deal will not be struck and uses that should happen, don’t happen. So fair use sidesteps dealmaking altogether and leaves the public free to engage in the uses without transacting with the copyright owner. I think it is important to make clear that the transaction cost explanation only captures a portion of the fair use space and does not, and should not, fully define fair use. (Many people have made this point before, but it seems worth reiterating.)

The transactions we tend to think about when discussing fair use and the transaction cost story tend to be transactions between the copyright owner and the user (e.g., Google), and in evaluating the story (are transaction costs high or low?), we tend to retain our focus on these parties. This narrow focus misses spillover effects and fails to consider the possibility that the user may decide not to seek a license or may take a license under conditions or with restrictions that restrict or eliminate benefits that spillover to third parties. Fair use deems fair some uses that yield benefits to third parties, not because the transaction costs between the copyright owner and user are necessarily high but rather to sustain the flow of spillovers to third parties. As I explain in many other places (e.g., here or here), in some contexts, the user’s/licensee’s willingness to pay for access to and use of the copyrighted material may understate social demand for that party’s access and use because the user/licensee fails to observe, appreciate, or capture spillovers that flow to third parties. One lesson from this is that limiting fair use to force low cost transactions does not universally or necessarily improve net social welfare.

While transaction costs could be defined to encompass any barrier that would preclude reaching a socially optimal outcome, I think such a broad vision of transaction costs conflates much too much. We could say that when a user fails to capture spillovers that result from his use of copyrighted expression, transaction costs in a secondary “market” are prohibitive – the user cannot capture the benefits in a transaction with the person (public) that realizes the spillover benefit. The problem with this view, I think, is that in many, if not most, cases no transaction is even possible. This might seem like an argument that wins the day for fair use because transaction costs are high and cannot / will not be reduced. But that would go too far – the mere existence of a use that generates third party benefits over which no transaction can/will take place does not mean that the use should be deemed fair. (The high/low transaction cost trigger doesn’t help in this context.) The spillover benefits realized individually may be small in scale (magnitude), which explains perhaps why the benefits tend to be ignored in a transaction cost analysis. Of course, in some cases, the spillover benefits may add up to a considerable amount of value, e.g., when small scale benefits are widely dispersed. (Of course the benefits of fair use are only part of the equation; the costs matter as well. I don’t have time just now to deal with arguments re. the costs of fair use, but as discussed elsewhere (e.g., Lemley), the impact on copyright incentives is [at times] [often] overblown.)

This post is getting long, so I’ll wrap up with the basic point: some fair uses depend on whether transaction costs are high/low; other fair uses exist whether or not transaction costs are high/low. The latter category includes some uses that generate substantial spillovers. And bringing it back to Google Print, the search capability that Google Print brings the public has the potential to yield substantial spillovers to the public.

5 thoughts on “transaction costs and fair use”

  1. Bret:

    I wonder about this:

    “And bringing it back to Google Print, the search capability that Google Print brings the public has the potential to yield substantial spillovers to the public.”

    Does it? What evidence do you have that this proposed search function will be done well, yield relevant and appropriate results, and benefit people more than the existing catolog services?

    I think it’s important to ask what standards and principles Google will be using for this project. I have been asking it for months. No one seems willing to answer me.

    Why does everyone assume that Google’s library project will actually work for people and benefit research?


  2. Siva,

    You ask important questions that I don’t have very good answers for at this point. I don’t yet know how well the search function will work and the extent to which the results it will yield will be relevant and appropriate. My primary point was that the transaction cost analysis for fair use is overly narrow (generally and in the context of analyzing the Google Print case); with respect to the last sentence that you noted, I should probably emphasize the word “potential”.


  3. Siva,

    Couldn’t you ask the same or similar questions about almost any proposed “fair” use? Can you prove — ahead of time — that the use will in fact yield the social benefits that fair use proponents claim?

    Sometimes, I think, the answer is yes; often the answer is no. Yet in the latter case the use may be noninfringing nonetheless (example: time-shifting in the Betamax case), because the court’s judgment is that the harm of erring on the side of the copyright owner is greater than the harm of erring on the side of the alleged infringer.

    I should emphasize that, like Brett, I am framing a general point and not an argument about Google Book Search in particular. But I fail to see how the possibility that Google Book Search will be less effective than existing catalog services is an argument in favor of infringement and against fair use. I don’t know that “the market for existing catalog services” is the right baseline for assessing the legality of Google Book Search, but if it is, why isn’t competition from Google Book Search, on balance, a good thing?


  4. It seems quite obvious that the basic Google magic potion won’t work with books. Most books are not built around links (maybe with the exception of law review articles).

    But whatever Google uses to calculate BookPageRank, there will be many search results in the back pages no one ever looks at. And those will probably come from the majority of books out of print.

    The books that are out of print are there for a reason. They don’t generate enough interest anymore.

    That in turn means that the benefit for the searching public from including millions of obscure books no one knows or is interested in reading might be quite limited. Google might get not much of a return for most of the $750 million scanning operation.

    (Mirrored on my blog, where I am writing quite a lot about the library case).

Comments are closed.