Washington, D.C.
October 15-17, 1997

Preservation of Digital Information

The Legal and Policy Implications of CMI

Julie E. Cohen, Assistant Professor of Law
University of Pittsburgh Law School

At the outset, I think it is useful to differentiate between two things: CMI, Copyright Management Information, which is information attached to a digital file that would tell you, for example, who the author is, who owns the copyright, and what the terms of access and use are; and CMS, Copyright Management Systems, which are technological protections that enforce the terms of use. CMS can be designed pretty much to measure, meter, and charge in whatever way one wants, and can also be designed to capture reader use information for billing and other purposes, such as direct marketing, perhaps.

You have already heard, and I think it was probably immediately obvious to most of you, that, if implemented in their full capability, these systems force us to really rethink Sections 107 and 108 of the Copyright Act, the fair use and library copying privileges provisions.

It is technically true that if you look to how fair use is defined in the Copyright Act, it is defined as a defense. And that makes some practical sense, if you think about how a fair use issue might arise. It wouldn’t really arise until somebody is sued for alleged infringement, and then could be used an argument to establish why the suit is meritless. The advent of CMI and CMS forces us to think about that categorization of fair use as a defense, and whether that is all that it is or that we want it to be. If you look at the case law on fair use and the history of the fair use doctrine, there is real tension, at least in United States law, between two competing visions of what fair use is.

Under one vision, it is just a mechanism that we have developed to cure market failure caused by the fact that transaction costs become high in connection with particular kinds of uses, such as private or home copying, parodies, use in classroom settings, and so on. If that is how you see fair use, then the minute there is a fairly inexpensive mechanism, such as automated CMS, to take care of all these transactions cheaply and easily, the rationale for fair use disappears.

But it is not so simple, and if you look at the entire history of the doctrine and of copyright, there is as much support for quite a different vision of fair use, a vision which one might say starts to look like an affirmative right in some circumstances. There is a very strong argument to be made for the proposition that copyright wasn’t ever intended to apply to things like private home copying and private home use in the first place.

Similarly, as to the Section 108 library privileges, obviously, if you have the possibility of charging fractional royalties so that the expense of using works that one wouldn’t ordinarily purchase goes down, then, if you subscribe to the “market failure” notion of what copyright is all about, it is really not so important to give libraries the ability to make copies of works for their patrons. They can just serve as information facilitators, helping patrons find their way to various works and sources of information. But the reader’s actual access to that information would occur in the context of a licensing transaction between the reader and the content owner. The role of the librarian really has changed very significantly in that vision.

If you look to other aspects of the history of copyright, you might note, for example, that we have always had a requirement that works be deposited with the Library of Congress. That requirement has been modified in some respects recently, to take into account things like computer source codes that contain trade secrets. Still, by and large, we do have a deposit requirement, and implicit in that requirement is that works become publicly accessible by means of that deposit.

We have also had, implicit in the structure of our information markets, a practice that libraries follow of buying works once they are commercialized, once they are made available on the market. And so, in practice, what that has meant is that once a work was made available for general readership, people could have access to it, if not through their local library, then via interlibrary loan, or, in a case of last resort, through the Library of Congress.

Thus, I think one could fairly argue that Section 108, like Section 107, reflects a principle of access that is also present in the copyright law, as developed by our courts over time and as reflected in some aspects of the legislative history of the copyright statute.

This may all become a moot point, as Cliff Lynch told you earlier. If there is a rights management system that prohibits you from exercising your right of fair use, or prohibits you, if you are a librarian, from making a copy that your patron has requested, it is really beside the point whether you have these rights—, because they are affirmative rights, that also means that content owners can’t make it more difficult to exercise them. That’s a difficult subject.

It has been my contention in my writings in this area that if fair use is an affirmative right, then it has to be expected and it has to be legal for readers to exercise technological self-help in instances in which they would like to make fair use of published materials. One might think that a market would develop, as it did a decade and a half ago for systems to defeat copy-protection technologies to enable people to make archival copies of their software. The market might provide, for example, software tools that would help people make fair use, private home use, of reading material.

That may be a somewhat Panglossian view of the way the world should be, because currently pending in Congress is a bill sponsored in the Senate by Senator Hatch and in the House by Representative Coble to protect rights management systems against so-called “circumvention” by users and other types of people who are generally not to be trusted.

Also pending is a different bill, sponsored by Senator Ashcroft, that would focus more specifically on the act of circumvention only in cases where it would defeat a right that the copyright owner actually has, as opposed to a right that the copyright owner is trying to aggrandize through a technological protection mechanism. It is vitally important for the future of fair use and the future of library copying privileges that the library community take an interest in those two bills, or if a compromise between the two is reached, take extra care to ensure that it looks more like Ashcroft’s bill than Hatch’s.

To throw another acronym into the mix, at lunch you heard about UCC 2B, and we also need to think about UCC 2B and how it relates to copyright. If content providers can contract around the rights that otherwise would be established by copyright, then copyright isn’t terribly relevant. It is quite simple to set up a rights management system in such a way as to structure every transaction in terms of what you could call a “click-wrap” contract, where you click on little boxes to manifest your assent to various things, such as the surrender of your fair use rights if you’re an individual, or surrender of your Section 108 rights if you are a librarian.

Traditionally, courts have said that contract is not preempted by copyright, because contract has a sort of extra element to it as a matter of law. Contract is a specific relationship that is established between the two parties, whereas copyright is a right against the world, and so it is okay if two parties want to bargain to particular terms, because the rest of the world will still have the rights that copyright gives pertaining to that work.

If you think about how this might apply to standard-form click-wrap contracts, to which one must agree in order to access digital material, you may think that contract will look a bit like a right against the world, the kind of sweeping right that, in essence, constitutes private legislation. If we are to pay more than lip service to the idea of having a federal copyright system, you might conclude that such contracts should not be able to override federal copyright rights.

This is going to be an issue that will need to be confronted in the courts and probably also in Congress, particularly if the UCC 2B is enacted by the states in the form that it is in now. It is an issue that is raised by the Ashcroft and Coble bills about protecting rights management systems against circumvention. Again, I think it’s something that the library community has a clear interest in, because it is really a choice of either copyright as a supreme federal law, or doing whatever you want as a matter of contract and it doesn’t matter what the copyright law says.

Let me say a little bit about the privacy half of this problem, and then a bit about what sort of approach might be taken to address the problems that I’m raising.

As Cliff has told you, clear privacy violations are threatened by systems that capture, on a very fine-grained level, details of what individuals are reading, perhaps even down to the chapter or the pages of a book, how many times they have read it, and whether they wanted to excerpt it or not. You might think that there would be some privacy law that protects against this, but, in fact, in the United States there really isn’t any comprehensive source of privacy protection concerning the collection of information like this, by private entities, although there is some protection against what government entities can do.

But, other than some very specific laws— example, we got the “Bork Bill” after we heard about what movies Judge Bork liked to rent, and there is a privacy law that governs cable TV subscriber information— than narrow provisions like those, there really isn’t anything that applies to private entities. In addition, there is the Direct Marketing Association, a quite powerful lobbying organization that for years has been maintaining that its members actually have a First Amendment right to collect and share information about what their customers buy, what their customers like, and what their customers want.

It may not trouble you much if someone knows that you bought a Black & Decker hand-held vacuum cleaner or some Tropicana orange juice, but it might trouble you more if the stored information concerns what you like to read, what ideas you consider persuasive, and, perhaps, what medical information is important to you and your family, which you have been looking for on the Web.

I have argued, and I think it is quite a legitimate argument, that one can find elements of an affirmative right to read anonymously grounded in First Amendment principles. I will just briefly explain why. The First Amendment protects against compelled speech. You don’t have to recite the pledge of allegiance if you don’t want to. You don’t have to subscribe to opinions that you don’t hold. And you don’t have to speak unless you want to. Information about what you are reading and whose ideas you find persuasive is a form of communication. It tells the person who is looking at that information something about who you are, and what you believe.

If this information were widely available, it would create the kind of chilling effect that the Supreme Court has found troubling in other contexts. It would, for example, chill the exercise of HIV positive people to go and read lots of material about HIV, if they thought that their employers could go and purchase databases of who is accessing this material. I am sure you could think of many other chilling effects that might arise.

The Association of American Publishers (AAP)— I can stand to be corrected and I hope I would be when John speaks— AAP has not been that sensitive to privacy issues thus far. Their European counterparts have been somewhat more sensitive. The library community, in contrast, has historically been very sensitive to reader privacy, and therefore it is a topic on which they can provide valuable input. It is a subject that really hasn’t had enough attention paid to it thus far, and I would hope that there would be a lot more attention paid to it before we get any kind of law enacted, defining what kind of information people can or cannot collect and share.

In closing I would just like to say a word about the title of Cliff’s presentation, “When Technology Leads Policy.” It should be clear by now to everybody here that these systems can pose very real threats to various things that historically we have valued very deeply. On the other hand, I think it is false to think of rights management systems as things that can only be implemented in one way. It is false to think of rights management as something that can only be accomplished by dividing works into the smallest fractional unit that is still possible to bill, while keeping the most information that is possible to keep. Those things aren’t necessary, and, I think, are in fact quite the opposite.

We should be focusing our efforts not on trying to identify what kinds of measurement and data aggregation are possible, but on what kinds of functions we would like to see these systems perform from the perspective of readers, librarians, and the academic community.

For example, would we want people physically present in libraries to have a right to browse? And can we design systems that permit people to browse for, say, no more than three hours and don’t allow any printing or digital copying? Would that be something that would be satisfactory, given that there is a legitimate need to protect content owners against the wholesale making of hundreds of thousands of copies of their content. That’s not something that we would like to see, either.

Maybe we want to preserve the right to— I do think it is a right— make some private home copying. We might want to disable any digital copying functions, but still allow printing to occur on a screen by screen basis to mimic the slowness and inconvenience of a photocopy machine, while still making it possible for people to engage in some copying.

You might say that, well, now that we have this wonderful capability in our hands, why do we want to mimic the inconveniences and the frustrations of older technology? Let me just give you one example of an area in which, in fact, that has been done.

You may remember that back in 1987 there was a big stock market drop that was traced to automatic computer programs that conducted program trading for large mutual funds. These funds have enormous stock holdings, and so an automatic decision to buy or sell could have enormous repercussions, sending the market into a tailspin, all just as a matter of the automated performance of computer systems.

As a response to that problem, it was decided to actually build some lag time into the program-trading computers in order to mimic, in some degree, the slower response one would get if the system were not automated, to thereby avoid crashing the entire stock market. This example is from a slightly different area, but it is a good illustration of why it might always not be such a bad thing to have real-world capabilities, rather than virtual capabilities, operative.

So, copyright management systems need not pose a stark tradeoff between efficiency on the one hand and access and total redistribution of intellectual property to the people on the other. (Though, if it were such a tradeoff, I would favor redistribution over efficiency hands down.) There is some middle ground that we can identify, and that is really the task that lies ahead for the library community. I hope that you will accept it, because you have a voice that a lot of people will pay attention to if you make it heard.

Thank you.