Epstein on Google, continued

My comments on Epstein on Google Book Search: Seeing fair use as being fundamentally about the necessity of quotation for critical purposes is a mistake; there are plenty of noncritical uses that are also important to a robust speech environment, including repetition of religious texts and positive reviews.

Epstein responded that fair use for positive reviews is also good because a positive review means more if the reader knows it didn’t require the copyright owner’s prior approval, and thus we can assume a kind of implied consent for fair use for any kind of review. But that is a just-so story; it relies on an awful lot of unevidenced assumptions about consumer knowledge and behavior. The justification should not be the implied consent of the copyright owner at all.

Epstein then turned the discussion to the situation in Texaco, where he argued that individual copying was okay because it provided an indirect benefit to the copyright holder (by increasing the value of a subscription), but when individual copying was aggregated, price discrimination became an issue. You don't want a uniform price for intensive and nonintensive users, so you license copying, at least once metering becomes possible. I'm not convinced this is an answer to my point about the purpose of fair use, but okay.

Now more from me: Epstein thinks that a market for a database of digitized books can develop. But right now, blanket licensing isn’t possible. The publishers don’t necessarily own the rights to books bearing their imprints, especially the older books. The fact that the Authors Guild has also filed suit is a harbinger of ownership disputes to come, just as the record companies and the recording artists started to fight about ownership in Napster. (C.E. Petit said a similar thing about rights ownership, from a very different perspective.)

This also affects Epstein's argument that Google's business model for net search is protected by implied consent even though it's not for Book Search: Google was sued for online copying despite the robots.txt opt-out, so not everyone has gotten the message that putting material on the web implies consent to copy. On the other side, if the publishers don’t own most of the books in the University of Michigan library, as noted above, then their lawsuit isn’t evidence that most owners reject digitization.

Epstein agreed that there were big ownership problems, though he thought that made Google's opt-out policy even less sensible. Someone will pop up claiming ownership in many cases, and Google will have to invest a lot to figure out who has the valid claim (or will just take a book out whenever a claim is asserted, valid or not). This is already occurring with the non-Library Project books in Book Search. Epstein's preferred solution was more radical. For low-revenue/old works, he'd have a fixed statutory royalty for use, paid into a fund: a weak compulsory license. For high-revenue/young works, you'd need the copyright owner's consent. (I didn't ask whether there'd be a payment for digitization, a payment for showing a snippet in a search result, or a payment when a searcher clicked on the result, or some combination, but that's only the first of the hard questions.) Given the huge problem of rights fragmentation, we need a voluntary or statutory blanket licensing scheme. With recent works, rights are clear enough that voluntary transactions can go forward.

As an aside, Epstein thought we should also have much shorter copyright terms, which would go a ways towards solving this problem. Too bad it's not going to happen!

On the air rights comparison: I said that this dispute reminds me of the (relatively) old saw “The Internet perceives censorship as damage and routes around it.” Opt-outs are somewhat harmful to the completeness of the system, just as the result in Tasini was harmful to the completeness of the proprietary databases concerned, but they don’t destroy the value of the system. Indeed, since the whole is greater than the sum of its parts, opt-outs (at least at a low level) are less harmful than their individual merits might suggest. Whether opt-outs can be allowed depends on the specifics of technology and social organization; it’s not a defining feature of an anticommons problem. If the US had been settled at a uniform level of population density, airspace opt-outs wouldn’t have been as problematic as they would be when there are obvious paths flights should take. In other words, disallowing opt-out may not be a necessary part of a commons-type scheme.

Some audience questions: What about collateral uses of works? By "collateral," the questioner meant cases like Bond v. Blum, 317 F.3d 385 (4th Cir. 2003), the Fourth Circuit case finding fair use where a work was copied as evidence in a court case (it was used to show that the author shouldn't have child custody because of the bad things he confessed to doing in his unpublished autobiography, Self-Portrait of a Patricide: How I Got Away with Murder). Epstein didn't say much about this, nor do I think that Bond offers a common enough scenario to help us analyze Google Book Search. Precedents allowing copying for reverse engineering, as another comment suggested, might be more helpful. Epstein says Book Search is very different from reverse engineering. He's highly sympathetic to contractual restrictions on reverse engineering because they enable price discrimination, though the majority of copyright scholars are on the other side.

Question: Will every book eventually be searchable/available in a digital database? Answer: Yes, but when? The usual rule is that information loses 25% of its value every year. In publishing, you get 6 months of glory and an eternity of oblivion. In the end, the obvious gains to everyone will lead to the universal database; the issues involve reducing transaction costs. The publishers want to be a little better off than "better off," though they concede they'd be "better off" under Google Book Search. That is, Google's best argument is that its scheme is already Pareto optimal. Again, the intensity of use is key to the publishers' agenda. Publishers, like many owners, would like to switch from sales to licenses.

At this point, I asked Epstein how he felt about that as a libertarian. He felt fine, since licensing is just a voluntary transaction, and if licenses are more efficient than sales, they should be allowed. (The reason I asked is that some versions of libertarianism suggest moral values in individual property ownership, and a world in which most of us don't own property, only rent it, might create a different kind of character and politics.)

Question: What about Kelly v. Arriba Soft? Answer: It's pretty much the same thing. If snippets are samples for which publishers should get paid, thumbnails are too. Shrinking size, like putting a Rubens painting on a postcard, isn't fair use. (At this point, I've got to assume that Epstein's consistent choice of out-of-copyright examples is a deliberate effort to show how he's not seduced by contemporary popular culture -- but I'd still think that Hemingway and Picasso would be better examples.) The really interesting issue, he thinks, is trespass to chattels. He helped litigate the eBay case, in order to establish the basic principle that if you own a site, you can exclude others from using it. If a use is fair use, can the site owner exclude others from entering to get the work? Or is it still a trespass? This is an important ongoing issue.