More Google Print: Fair Use and Inefficient Bundling

Yesterday, Doug Lichtman asked the question “how should we decide when a copyright holder is entitled to earn revenue from a new technology.” He said that he didn’t have an answer yet, and I don’t either, but I do think that we need to talk about fair use and how it can operate as an inefficient bundling of rights.

Consider this set up. We have a book published on paper. All consumers value the paper copy of the book at $8. Some consumers are happy as clams with just paper. Other consumers would love to have a digital, searchable copy of the book to go along with the paper copy. Those consumers would value bundle of a paper copy and a digital copy at $12.

If the copyright owner didn’t fear copying—either of the physical book or a digital copy—what would she do?

The answer is straightforward: sell physical copies of the book for $8 and digital copies for $4 (or for the proverbial epsilon less than that if you like). Old-school consumers would just buy the physical copies, newbies would buy both. We would sell to all consumers, and, on this setup, the copyright holder would capture all of the value associated with the work. We would also not leave any social value sitting on table. No transactions that we would like to see take place will have been missed.

Now consider a possible fair use doctrine. Suppose that we announced a doctrine that said that any consumer was entitled to the work in any medium, once the work was purchased in some medium. The proposed Digital Consumer’s Bill of Rights would seem to do just that. I suspect that there are some who believe that current fair use doctrine suffices for this result.

A digital copy is produced—perhaps through an online open source-type process where individuals contribute chunks of typed text. Consumers who wanted the digital copy would no longer need to pay $4 to get it. Where would this put our copyright holder?

Before the copyright holder could sell different products to different consumers. The sensible approach was to sell just paper copies to the dinosaurs and sell paper and digital copies to the younger consumers. Now the rights regime makes it such that the copyright holder can sell only one product, but some of those consumers will value the product at $8, while other consumers will value the product at $12 (knowing that once they have the paper product, they can get a digital product they will value at $4 for free).

What will the copyright holder do? She can sell the paper copy for $8 or the paper copy for $12, but she can no longer sell any digital copies for a positive price. What she will do depends on the numbers (and to simplify matters, treat the costs of making physical and digital copies as being 0). If the copyright holder sets a price of $8 for the physical book, she receives revenues of 8*(O + N), where O is the number of old-school consumers and N is the number of digerati. If instead she sets a price of 12, her revenues are 12*N.

Try an example. Set N = 10 and O = 4. Total revenues from selling at $8 are then $112, from selling at $12, $120. In the face of free digital copies—whether from the open-source text collective or from Google Print—the copyright holder will jack up the price of the physical book. She can no longer treat the two customer types differently.

Note that the copyright hold is much worse off than before. Without the free digital copies, on these numbers, the copyright holder sold to everyone and took in revenues of $152. But, and this is the important point, society is much worse off too. All consumers were served before. Now the increased price for the physical book—caused by the inability to charge separately for the physical book and the digital book—has priced some consumers out of the market.

I don’t begin to think for a second that this captures the full complexities posed by Google Print, and you shouldn’t either. We are going to have to describe the relevant markets in ways that are richer than this, but that said, this kind of effect is clearly a piece of it.

Fair use is a form of bundling: one right necessarily comes with some other right. Not all consumers will want the “extra” bundled rights. Copyright holders will take into account in their initial pricing decisions for a work the full uses being conveyed, including those conveyed as a result of fair use.

Google Print makes it much less likely that consumers will buy digital copies from, say, Amazon. The Amazon Upgrade program (see my post on this) is exactly buy paper, get a digital searchable version, and it is hard to see how you sell the digital copies if they are freely available on Google Print. Sure, sometimes free can compete with fee, but it is odd to think that there aren’t spillovers of exactly the sort that we see in this example.

Comments

[Ed Felten]

You say this model doesn't "capture[] the full complexities posed by Google Print". It seems to me that this model isn't like Google Print at all.

The value of Google Print comes not from searching within a single book you own, but from doing a single search that spans a huge number of books you don't own.

The analogy to Google's web search is instructive. If the law had required search engines to get explicit permission in advance from every site they indexed, today's popular search engines would never have come about. Transaction costs would have killed them before anybody saw their value; and we probably wouldn't know what we were missing.

If you're going to argue that it's good policy to require Google to get advance permission from publishers, then I think you'll need to have an argument that holds for books but not for web sites. I don't know if that's where you're headed, but I think the analogy to web search is valuable nonetheless.

[Doug Lichtman]

Ed -

A few thoughts.

1. If we want to tie directly to Google Print, I take it Randy will say that -- if Randy's argument holds -- it would tell us that GP should not be allowed to offer within book searching. That is, even though GP will have the right data to provide that service, we should not permit it as fair use.

2. Your move -- "this is the same as web searching" -- is just not right, even though lots of folks have made it. Two among many differences are: (1) people who are in Google's normal web engine are people who put their stuff online and in digital form willingly, whereas authors did not and have not; and (2) the economics of these industries (ie how people earn returns on their investment) differ tremendously, which means the implications of a fair use argument might also differ, given that fair use explicitly factors in questions about economic harm.

3. Finally, what is interesting about Randy's post is that it introduces yet another angle to think about the question of how new technologies alter the value of existing copyrights. I had talked about new techs displacing old revenue streams; Randy points out that there are pricing externalities such that new techs can change the ability of a seller to extract value from the old medium. I don't see that as a slam-dunk answer to my original inquiry; but nor does Randy. It's just a helpful addition insight into how to frame this question.

ps. I am now involved in the case, so let me insert the standard disclaimer here that my views are my own, and they in no way rely on or reveal the views of, or any information provided by, any client. If I'm wrong, it's me and me alone. (Sadly, it is often also true that when I'm right, it's still me and me alone, but that's an entirely different issue!)

Have a great weekend.

[Tim Lee]

Another way that Google Print differs from Prof. Picker's analogy is that any author can request his or her book be removed from the Google Print index. So from an economic standpoint, the issue isn't so much whether publishers are allowed to sell print and electronic versions separately. Rather, the argument is about which way the default should be, and who should bear the transaction costs of opting in/out: Google or the publishers.

Keep in mind that the vast majority of books now on library shelves have passed their sell-by date, and are of historical and academic interest only. Many of them are "orphaned works" whose owners would be expensive and difficult to track down. They are also books for which Prof. Picker's argument doesn't apply at all: if a book is no longer available for sale, there are no revenues for Google to harm. For books that are no longer making much revenue for the publisher, it's obviously a pareto improvement to have them in an index where the general public can benefit from the ability to find them easily.

Yet it is precisely those works that would be most effected by the rule the publishers are advocating. For books still in print, it doesn't matter a whole lot whether the model is opt-in or opt-out. Either way, publishers will make sure the books aren't in the index if it's in their economic interest to do so. But for orphaned works, the default is likely to be decisive. Because the economic value of each work is pretty small, and so if the copyright system makes clearing the rights to them too cumbersome, they simply won't be included at all, even if the copyright holder would have consented if he'd been asked. (as the owner of an orphaned work almost certainly would)

[Ed Felten]

Doug,

I agree that Randy's argument is useful in thinking about efficient rules for new technologies that use copyrighted works. But I don't think they speak directly to the Google Print situation. Probably they can be extended to Google Print, but when that happens I expect other factors will come into play.

I'm not arguing that Google Print is exactly like web search. What I am saying is that an argument that Google Print should be illegal must hinge somehow on differences between Google Print and ordinary web search. You point to some differences.

The web search analogy works differently depending on which point in the history of web search we are comparing to. We can analogize to web search back when Larry and Sergey were just grad students and nobody imagined that web search would ever be big business. People writing things on the web in 1995 were almost as clueless as book authors of the 1950s about the possibility that online search of their work could eventually be monetized. Or we can analogize to web search now, where people know that their writings will be crawled and indexed, and they have the choice of opting out. I think both analogies are useful.

[Michael Martin]

As Prof. Picker concedes by acknowledging that his model does not capture the full complexities, the argument that permitting price discrimination will be socially beneficial depends on a lot of details that are left out of his model. I think the ones that Tim (Bernars-?) Lee points out -- that the transactions cost of clearing the copyright in orphaned works -- are very important. We really have no idea what the magnitude of the external benefits of making that volume of information readily available are like.

Another detail to add is that permitting the kind of price discrimination that Prof. Picker describes effectively raises the barriers to entry for potential competitors in the subsidiary market (here, the market for digital books). To effectively compete in a world in which copyright owners control both paper and digital distribution, a potential new entrant will have to offer BOTH paper and digital books (like Amazon). If we think that the social benefits of competition in the digital book market are substantial enough to outweigh the consumer surplus foregone in the paper book market, permitting price discrimination may not be such a good idea.

At first glance, I'd say there's probably a lot of information that's worth publishing (or republishing) in digital format that wouldn't be worth publishing (or republishing) in print. Promoting new entry into the digital publishing market doesn't seem like such a bad idea if it will mean making a wider selection of expressive works available in that format.

On the other hand, maybe we don't need to worry about barriers to entry so much in a world where capital is so readily available to companies like Amazon.com

[Paul G]

Tim's arguments are critical here. If we distinguish the "market-life" and the "copyright-life" of books, and we acknowledge that the former is almost always shorter than the latter (e.g., eldred), then defaults really start to matter. Publishers of books in the prime of their market-life will take care of themselves and Google allows them to do just that. For the rest, the majority, we need to choose the appropriate default. When doing so we mustn't simply look to the absolute number of "switchers" (i.e. those that would opt-out if opt-in was the default and vice versa) and set the rule in favour of the predominant starting position. We must factor in those owners who would like to switch but for whatever reason don't. This seems to me to be a critical issue with regards out-of-print and orphaned works. There's no market to harm, so there's no apparent economic reason (there may be non-economic reasons) for opting-out, but there's also no guarantee that the owner will in fact opt-in. I may be assuming too much, but I believe this to be a crucial dimension to the fair-use problem.

By the way, i'm from the U.K. Under U.K. law, GP wouldn't have a hope under a fair dealing (closest thing we have to fair use) assessment. This probably explains why Google is sticking to public domain works from the Oxford University library.

[Doug Lichtman]

Several of the comments mention transaction costs. But remember: a publisher will have to "opt out" of not only Google's system, but every other system that similarly claims the protection of fair use. So, as I pointed out in my original post on all this, the transaction costs argument cuts both ways. For this project in isolation, Google likes to argue that opt-out is cheaper. But the reality is that there will be hundreds of projects that claim fair use if Google wins this case, and the costs of evaluating and opting out of all of those projects could be significant. Put differently, the key point here is to remember that this is a legal rule we are making, not a Google-specific contract. The transaction costs arguments are therefore much more complicated than the traditional conversation lets on.

[Bruce]

"But the reality is that there will be hundreds of projects that claim fair use if Google wins this case, and the costs of evaluating and opting out of all of those projects could be significant."

Interesting -- I just got done teaching a class in which we discussed whether it was fair to make consumers opposed to tracking cookies visit dozens of separate advertisers' sites, evaluate their policies, and opt out of each one, or whether the presumption ought to be one of requiring consumers to opt in. Given the relative lack of harm to consumers I tend to lean toward allowing multiple opt-outs; but does that mean I'm committed to accepting Google's fair use argument? Or vice versa, if you favor Google's argument, you're committed to supporting privacy legislation that puts most of the burden of "opting out" on the consumer? This is another area where I think copyright and privacy overlap, but in ways that tug at the usual alignment of intuitions.

[Michael Martin]

In response to Prof. Lichtman, I think the probability that hundreds of projects that claim fair use will emerge if Google wins this case is exactly the social benefit of increased competition in the digital book market that could be desirable. Why should print publishers have exclusive control over that market? There's no reason to believe that they should unless the consumer surplus foregone in the print book market by denying them the ability to price discriminate is larger than the benefit of having lots of Google-print-like projects.

Isn't the question really whether we want
(a) more NYT bestsellers cheaper or
(b) a wider variety of Google-print-like archives on the internet?

If that is the right question, the choice is clearly (b) for me.

[Tim]

Prof. Lichtman,

It's true that the presence of more search engines raises the opt-out costs to publishers. But don't the opt-in costs rise in precisely the same proportion? If there are 100 search engines, each of them would have to contact and seek permission from each and every publisher. That, it seems to me, would create a large and unnecessary barrier to entry to the book-search-engine business.

If opt-in is made the default, that could have the effect of dramatically reducing the number of firms that enter the book-search-engine market. Although it's true that that would reduce the opt-out costs of publishers that wished to opt out, it seems to me that the much more important result will be that the potential innovation that those competitive search engines might bring would be squeezed out. Especially in a young and vibrant technological field, it's important not to put unnecessary barriers in the way of new firms.

Moreover, even if there were hundreds of search engines (which seems unlikely in practice) I don't think the burden on publishers would be unacceptably high. Consider that the major cost of opting out is in compiling the list of books to be opted out complete with the required information (ISBN number, author name, etc). Once the list is compiled, it's a trivial matter, on the margin, to send that list to an additional search engine. Indeed, it's easy to imagine a firm specializing in delivering these notifications on the part of publishers. Since the process could be almost entirely automated, the cost could be as low as a few dollars per book.

[Randy Picker]

Lots of interesting comments; I will probably repost to take up some of these later this week, after we have wrapped up the mobblog on Julie Cohen's paper. See, if you are interested, http://picker.typepad.com .

[Doug Lichtman]

Randy is right -- lots and lots of interesting comments. Thank you all so much.

Let me pick up on a few more ideas.

1) Michael, you are certainly right that some rules will lead to a huge number of fair-use-based services, and other rules will lead to a different number of permission-based services. I'm not sure which services I value more, which will be better run, and so on; but I agree that we should think about that as one relevant consideration in our analysis. This is just like Grokster, where (in part) the choice was whether we would funnel cash and innovation into services like iTunes (permission-based) or services like Grokster (not-permission-based). I don't want to pretend that to be an easy issue; both sides are important, I think.

2) Tim, you are right to think about opt-in and opt-out costs, and again that was exactly what I urged in my first post on all this. But now that we are focusing on both costs, I guess we have to think about which default is more likely to match what people want. That is, if publishers are going to opt out of 99% of these services -- and my guess is they will, given that most will be amateur services that are insecure and poorly run -- then an opt-out rule is likely more costly than an opt-in rule, since opt-out will often be the choice. Also, as you point out, none of this analysis can be done without thinking about middlemen who can make this stuff cheap and easy. So, for example, ASCAP coordinates an opt-in regime for the playing of copyrighted music on the radio; why not a similar approach here?

3) Several of the comments took my point about "if Google can do this, lots of people will claim they can do it" very narrowly, imagining only that those "lots of people" will also do search engines. But why should the floodgates be so narrow? Just like Napster opened the door to all sorts of services, I take it that a Google win will inspire hundreds of different things, all trying to say that their "fair use" is analogous to Google's. That would be a nightmare for copyright holders who would have to find, evaluate, and respond to all sorts of different technologies, some of which would be created in bad faith (ie like Grokster) and all of which, taken together, could substantially undermine the value of copyright protection.

Again, thanks so much for the comments and feedback. The whole idea of this blog was to open up the conversation exactly as we are doing here. I'll keep reading, and I hope you will, too.

[Jule]

It is interesting to me that Randy's post dealt with the economics of Google Print relative to its competitors and the choices faced by authors/copyright owners for monetizing digital copies of their works, yet all of the comments revolve around the control aspect of copyright -- whether Google Print should be an opt-out or opt-in system. But the "fair use" analysis traditionally has been pretty blunt -- if it's fair use, neither control nor compensation is due the owner. And though Google has offered opt-out, it has not said anything about payment to authors, I would guess because they are taking the position that they provide more benefit to authors, so if anything, authors should be paying them rather than vice-versa.

So, what happens to the author who wants to be part of Google Print but wants to be paid for it too, perhaps even with a "reasonable" offer -- e.g., "I want 10% of ad revenues from Search Results screens with my book on it and 50% from ads on pages devoted to my book." Is this negotiation possible if Google's activity is deemend fair use? It think Randy's answer is no, it's not, and based on the comments here and in Doug's earlier post, it seems many commenters here believe that that may be a good result -- essentially the position that authors have no claim to the economic value generated by Google through Google Print.

I think that's a very debatable conclusion -- if copyright means anything it all, it should be that an author has at least a claim to some of the value that new technology brings to the enjoyment of copyrighted works. It is also at odds with the 1976 Copyright Act, which was intentionally drafted (in defining terms like "copies" and "fixed") to not tie copyright protection to existing technologies. It's a bit like arguing that The Beatles 1963 songs should only receive compensation for sales of sheet music, LPs, and radio broadcasts, but nothing for 8-tracks, cassettes, CDs, mp3s, mini-discs, cable TV performances, satellite radio, webcasting, etc. -- makers of those technologies should be unencumbered in their use of existing works that are created before the tech's advent.

Thus, if we accept the notion that both authors and Google should share in the economic value Google Print generates, the important question becomes what is the best way to resolve their claims? Copyright, with it's opt-in approach, gives us one answer -- essentially that the user must get permission first, and the author can leverage that in exchange for payment. How does the Opt-out approach (or any other) answer that question?

[Tim]

Professor Lichtman,

The security risks of search engines is a completely separate question from the issue Prof. Picker was raising, which was whether Google Print undermines publishers' ability to sell paper and digital rights to its products separately. I was looking at it in those terms.

If a search engine has security problems that pose the risk that users will be able to get free copies of copyrighted books, that would be a matter that the publishing industry as a whole could deal with through a lawsuit. The courts should have no trouble distinguishing Google (a service that has been very careful to prevent such risks) from shipshod imitators that fail to do so. And in that case, it wouldn't be necessary for each individual publisher to opt out, since an organization like the authors' guild could likely get the service shut down.

As for the Napster/Grokster analogy, you seem to be forgetting that Napster lost its lawsuit. Yet Grokster and its ilk still were created. I think it's inevitable that we'll have a proliferation of new technologies that make use of content in unforseen ways. But it seems to me that that's precisely why a hard-and-fast opt-in rule is inappropriate--some of those new technologies could have profound social benefits, yet it might be impossible to get them off the ground if they have to negotiate hundreds of contracts with publishers before they're allowed to set up shop.

Indeed, if the Napster/Grokster line of cases proves anything, it's that the copyright industry has little difficulty evaluating individual technologies as they come along and suing the copyright-infringing ones out of existence. After Napster, should the courts have ruled that all peer-to-peer technologies are presumptively illegal to save the RIAA the trouble of evaluating and suing them individually? To the contrary, not all P2P applications are illegitimate, (BitTorrent is used in a lot of legitimate products) and so it's vital that the courts evaluate each on its merits. Is that a burden on the RIAA and MPAA? Sure, but sometimes justice costs money.

It seems equally unfair to argue that the courts should rule against Google's good-faith fair use claim because in the future some companies might cite the precedent in making fair use claims in bad faith. There probably will be some bad-faith fair use claims, but other fair use claims--like Google's--are likely to be legitimate. Separating the wheat from the chaff is precisely the reason we have judges to make these decisions in the first place.

[Michael Martin]

I think Hayek would love this problem if he were around to see it. And we would love Hayek. He would surely have some useful insights here.

To respond to the Prof. Lichtman and Jule (Julie?), I think we all agree that the issue is whether to adopt an opt-in/permissions-based or an out-out/non-permissions-based approach.

The way I understood Prof. Picker's original post, he's identifying price discrimination as one of the benefits of an opt-in approach. Copyright holders can charge lower prices in the upstream non-digital market when they can also charge for use of the copyright in the downstream digital market. The social benefit of price discrimination is the elimination of dead-weight losses in the upstream non-digital market.

The social cost of price discrimination, however, is what I have been worrying about since I made my original comment. The social cost, I think, is the cost of increased barriers to entry in the downstream digital market. In an opt-in world, to effectively compete with existing copyright holders new firms will have to enter both the non-digital and the digital market. This is because the copyright holders have an artificial (government-created) economy of scale in the opt-in world. They get to use the cost of copyright in two markets, whereas a publisher that entered the digital market would only get to use them in one.

In a way, the choice between opt-in and opt-out is like the choice between a property rule and a liability rule, introducing Coase's critique of Pigou, and the ensuing discussion of transactions costs.

And so my thoughts turn to Hayek. When the question of who should be assigned the right gets murky, in addition to considering whether transactions costs are going to prevent an efficient reallocation, we should ask the larger question about who, on the street, is going to have the best information.

Here's a first guess about that: it isn't the existing copyright holders. They don't know anything about the digital market. I would rather give a federal judge the chance to consider the question under the fair use rubric than assign the rights to the existing copyright holders.

[Tim]

Jule,

You're wrong to suggest that Google hasn't offered to share revenue with authors. By default, Google Print displays no ads on pages containing copyrighted books. However, if publisher opt into the Google Print for Publishers program, then ads are displayed and the revenue from those ads are split with the copyright holder.

Indeed, this characteristic of Google Print is likely to be an integral part of Google's fair use claim--that it's a fair use precisely because they don't generate any ad revenue without the permission of the publisher, and therefore can't be said to directly benefit from the economic value of the publishers' copyrighted materials without the publisher's consent.

May I suggest everyone reading this exchange spend some time perusing Google's help pages? It would clear up a lot of confusion about what Google Print actually does.

http://print.google.com/intl/en/googleprint/about.html
http://print.google.com/intl/en/googleprint/publisher.html

[Joe Liu]

Doug,

I agree that the transactions cost question is complicated and warrants a lot more consideration. But I guess it's not so clear to me that opt-out will cost more than opt-in. Is it really so clear that publishers would opt out of 99% of these services, if the default rule were changed?

And even if they would, we still need to consider the non-publisher copyright owners (for example, I own copyrights in some works, which I would be happy to have Google scan) and the unidentified owners of "orphan" copyrighted works. With respect to these interests, an opt-out rule would be less costly (obviating the need to identify hard-to-find copyright owners).

Finally, the potential presence of transactions-cost-reducing middlemen cuts both ways. As Tim points out, intermediaries could be used to efficiently administer both opt-in and opt-out regimes. In my mind, the larger source of potential transactions costs comes in the form of tracking down the owners of orphaned works, and these costs are less amenable to these kinds of solutions.

Ultimately, this is an empirical question and I find the evidence so far to be very ambiguous.

[Bruce]

Joe, I think you're admitting this, but there's going to be hard-to-find people no matter which way you arrange the rights. On the copyright owners' side, there's going to be hard-to-find sites that are using digitized works for some purpose, perhaps one not as intuitively appealing as creating an index, and claiming fair use from it. E.g., a library of famous sports plays, most of which are only a few seconds long, but which (like quotations from J.D. Salinger's letters) comprise a qualitatively significant portion of those works. Making the right "opt-out" puts an enormous burden on the rightsholder to locate, evaluate, and communicate its preference to the site owner -- which in the privacy arena is the main complaint many consumer-side folks have with "opt out" privacy regulations.

With respect to third party rights aggregators, there's a happy medium somewhere concerning the number of parties you'd want to deal with. Too many, and the transaction costs are too high. Too few, and there may be an anticompetitive impact. Last I heard ASCAP and BMI are both still operating under consent decrees reached with the DOJ.

[Michael Martin]

Very interesting news from the Official Google Blog: Google is changing the name of Google Print to Google Book Search.

http://googleblog.blogspot.com/2005/11/judging-book-search-by-its-cover.html