I have been asked to give an overview of the economic and legal issues involved in the protection of databases along an intellectual property type of model. First, what is a database? It has really not been very well defined. We have the European Union Database Directive definition: “a collection of independent works, data, or other materials arranged in a systematic or a methodical way, and individually accessible by electronic or other means.”1 Examples of what would be included in this definition are tables of isotopes, journals, meteorological data, examples that we have heard this morning, gene sequences, medical records, stock reports, game scores, phone books—really, just about anything. That, of course, is one of the problems with this definition. Do we want a regime that is protection for almost anything at all? Well, perhaps yes; perhaps no. I think this is a major research question in and of itself: What is it that we actually want to protect?
What is the problem here? To a great extent, the problem came about in the Feist case. The Feist case, which Commissioner Dickinson talked about this morning, was a finding that a telephone phone book did not meet the threshold standard for creativity for copyright protection. If that ruling had gone the other way, either we wouldn't be here at all or we would be here with a very different problem statement to deal with. I will talk about Feist later. Particularly, I want to contrast it to issues about protection of similar types of property in the European Union.
Problem number one is that protection for sweat-of-the-brow work will impede scientific progress on database development through, among other things, rebalancing the financial locus of funding for scientific work, raising various types of hurdles to the aggregation of databases, and so forth. Problem number two is that there is a lack of protection for sweat-of-the-brow work, and that is impeding database development. Problem number three is a variant of problem number two, which is that there are existing protections now for these types of works, but they distort the investment in and the use of databases. I am referring specifically to various self-help types of provisions that might lead to different types of protections. Problem number four is that there is no problem at all; we would all rather be doing something else, but we are forced to be here, in part, through the E.U. Directive. I am going to rely on an excellent paper by Steve Maurer [Appendix C in this volume], which he has prepared for this conference, and I recommend it to you very highly.
Let's go on to the origins of intellectual property protections. I think it is helpful to understand how intellectual property protection came about and what the different models were long, long ago. There was the rights model—the idea that inventors, in the case of patents, and
NOTE: This session of the workshop consisted of a paper prepared for the study committee by Stephen Maurer, entitled “Raw Knowledge: Protecting Technical Databases for Science and Industry,” and a presentation by Richard Gilbert, a professor of economics at University of California at Berkeley. Mr. Maurer's paper is Appendix C of these proceedings.
1 See Directive 96/9/E.C. of the European Parliament and of the Council of March 11, 1996, on the legal protection of databases, 39 O.J.L. 77/20, March 27, 1996.
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS 4 Economic Factors in the Production, Dissemination, and Use of Scientific and Technical Databases Richard Gilbert I have been asked to give an overview of the economic and legal issues involved in the protection of databases along an intellectual property type of model. First, what is a database? It has really not been very well defined. We have the European Union Database Directive definition: “a collection of independent works, data, or other materials arranged in a systematic or a methodical way, and individually accessible by electronic or other means.”1 Examples of what would be included in this definition are tables of isotopes, journals, meteorological data, examples that we have heard this morning, gene sequences, medical records, stock reports, game scores, phone books—really, just about anything. That, of course, is one of the problems with this definition. Do we want a regime that is protection for almost anything at all? Well, perhaps yes; perhaps no. I think this is a major research question in and of itself: What is it that we actually want to protect? What is the problem here? To a great extent, the problem came about in the Feist case. The Feist case, which Commissioner Dickinson talked about this morning, was a finding that a telephone phone book did not meet the threshold standard for creativity for copyright protection. If that ruling had gone the other way, either we wouldn't be here at all or we would be here with a very different problem statement to deal with. I will talk about Feist later. Particularly, I want to contrast it to issues about protection of similar types of property in the European Union. Problem number one is that protection for sweat-of-the-brow work will impede scientific progress on database development through, among other things, rebalancing the financial locus of funding for scientific work, raising various types of hurdles to the aggregation of databases, and so forth. Problem number two is that there is a lack of protection for sweat-of-the-brow work, and that is impeding database development. Problem number three is a variant of problem number two, which is that there are existing protections now for these types of works, but they distort the investment in and the use of databases. I am referring specifically to various self-help types of provisions that might lead to different types of protections. Problem number four is that there is no problem at all; we would all rather be doing something else, but we are forced to be here, in part, through the E.U. Directive. I am going to rely on an excellent paper by Steve Maurer [Appendix C in this volume], which he has prepared for this conference, and I recommend it to you very highly. Let's go on to the origins of intellectual property protections. I think it is helpful to understand how intellectual property protection came about and what the different models were long, long ago. There was the rights model—the idea that inventors, in the case of patents, and NOTE: This session of the workshop consisted of a paper prepared for the study committee by Stephen Maurer, entitled “Raw Knowledge: Protecting Technical Databases for Science and Industry,” and a presentation by Richard Gilbert, a professor of economics at University of California at Berkeley. Mr. Maurer's paper is Appendix C of these proceedings. 1 See Directive 96/9/E.C. of the European Parliament and of the Council of March 11, 1996, on the legal protection of databases, 39 O.J.L. 77/20, March 27, 1996.
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS authors, in the case of copyright, have a right to their creative work and have a right to be compensated for its use. Another model, the incentive model, says that society will gain by offering a reward for the creation of intellectual property. That incentive model is the way that most economists think about intellectual property rights. I want to say something about the true believers. “True believers” is the term I am using for people who believe that property rights enhance efficiency. That is fine when you are talking about certain types of goods and services but not when you are talking about intellectual property. A concept discussed in the NRC Bits of Power report [National Academy Press, Washington, D.C., 1997] is depletable goods and services versus nondepletable goods and services. Depletable goods and services are things like apples—when you consume them, there is nothing left. A personal computer is a depletable good; only one person uses it, at least at a time. The value of a nondepletable good, such as intellectual property, is not diminished by use. It has essentially no opportunity cost associated with distributing the good to others. Hence property rights on a nondepletable good, such as intellectual property or databases that have a very low reproduction cost, have potential costs because they can lead to restrictions in the exploitation of the good. The efficient usage price of a good or service is its marginal cost, which is zero, or at least close to zero, for some databases and other forms of intellectual property. There is a trade-off between setting efficient usage prices and setting prices that provide incentives to create the good or service in the first place. Protection provides an incentive for investment by reducing the risk of misappropriation, but it also creates a potential disincentive for its use. Too much protection runs the risk of creating artificial economic barriers to the use of the database. Second, there is also a risk that database costs will discourage the use of complementary databases. This is the aggregation problem that we have heard so much about this morning. Sometimes this goes under the related topic of the “anti-commons,” which means that by trying to exert property rights, the transaction costs that are imposed by those property rights ultimately diminish the value of the product for all. Third, even if stronger database protection is appropriate, there is a risk that funding agencies will not adjust to the change in the costs and the values of the databases. Increased protection for databases likely would require that funding be increased to the users of databases, possible offset to some extent by lower funding for the suppliers of databases. A good example from the Bits of Power report is the National Oceanic and Atmospheric Administration (NOAA), whose total budget more than doubled from 1980 to 1994. However, funding for the National Data Centers has actually declined. I think anybody who deals with public agencies is familiar with this type of problem. You can get money for new buildings, but you can't get your windows cleaned. Maintaining a database is, to some extent, similar to cleaning the windows. What are the problems with too little protection? Too little protection creates the risk that there is not enough incentive to create or maintain a database. If the incentive is insufficient, the databases may not be generated, even though users would be willing to pay the costs. A different type of risk involves too little protection for the underlying database information, which may distort the type of product that is created. For instance, there might be too much investment in certain types of software or certain types of copyrightable software or certain types code or
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS structure as a surrogate way to protect the underlying information. We don't see a lot of evidence that this is a problem, but it certainly could be a problem. Now, I want to ask a question: Do strong property rights provide an incentive for innovation? What I want you to think about is property rights for databases that are comparable in some respects to patent rights. Although we are not anticipating patent-like protection for databases, experience with patents in certain industries is useful to gain some knowledge of the impact of strong intellectual protection. Experience shows that patents provide important incentives for innovation in some industries, but can diminish innovation incentives in other industries. I want to read you a quote from someone working in the microelectronic field who said the following with respect to the microelectronics industry specifically: Older, larger industrial companies often perpetuate a licensing extortion to generate needed revenue in lieu of operating profits and as a barrier to competitive entry. Newer companies respond by generating patents, not on inovations involving their own products, but specifically targeted to entrap the product lines of their competitors. Additionally, exacerbating the situation, are the frequent, independent paper patent inventors, who generate hundreds of patent filings without ever designing or manufacturing a product, while abusing patent office rules, to tailor their current claims against modern products based on the specification filed decades ago. Thus, the intellectual property protection and reward system as it currently exists appears both inappropriate and inhibiting to innovation in the microelectronics industry. [Quote from discussion with Michael Rostoker, Microelectronics Research, Inc., 1998.] We might hear a rebuttal to this point, perhaps even from the Patent and Trademark Office, but there seems to be at least a current of opinion that in some industries patents are a problem. What do patents do? What sort of benefits do we get from patent protection? I want to report some work that was done by Cohen, Nelson, and Walsh (Wesley M. Cohen, Richard R. Nelson, and John Walsh, 1997, “Appropriability Conditions and Why Firms Patent and Why They Do Not in the American Manufacturing Sector,” Carnegie Mellon Working Paper, June 24). In 1994, they asked approximately 1,000 managers of R&D units to indicate the percentage of their product and process inovations for which each of the following appropriability mechanisms had been effective in protecting their firm's competitive advantage resulting from those inovations during the past three years. There were eight mechanisms: secrecy; patents; other legal mechanisms, which could include copyright, for example; lead time; complementary sales and services, such as having a large distribution or a large repair facility; complementary manufacturing; the sheer complexity of those inovations, which makes them hard to reproduce; and other. TABLE 4.1 Percentage of Inovations Protected Effectively by Various Appropriability Mechanisms. All Product I ovations All Process I ovations
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS Secrecy 51.2 51.2 Patents 34.7 23.4 Other Legal 20.3 15.2 Lead Time 52.4 38.3 Complementary Sales/Service 42.1 30.6 Complementary Manufacturing 45.5 42.7 Complexity 42.2 38.3 Other 8.4 7.7 PATENT RANK (6 of 8) (6 of 8) NOTE: This table presents results of the 1997 study by Cohen, Nelson and Walsh for all industry and represents over 1,000 research and development units in 35 industries. Table 4.1 gives the results for 35 different industries. The numbers are the percentage of inovations for which a mechanism had been effective. One column is for product inovations and the other one is for process inovations. For the mechanism “secrecy,” the first number shown is 51.2, and it is 51.2 under the second column as well. This says that if, on average, all these R&D managers had 1,000 inovations in this survey, 512 of them would have benefited from the use of secrecy in protecting the competitive edge of those inovations over the past three years, and the same number for process inovations. Lead time benefited 52.4 percent of the inovations. Complementary sales and service benefited 42.1 percent; complementary manufacturing, 45.5 percent. For product inovations, complexity benefited 42.2 percent. If you use these percentages to rank these protection mechanisms, “patents” ranks, in both product and process inovations, six out of eight, that is, five mechanisms are more important, those five being secrecy, lead time, complementary sales and services, complementary manufacturing, and complexity. The only mechanisms that “patents” beats on this list are “other legal” and “other.” So, they're evidently not very important. None of the 35 industries ranks patent protection as the highest and most important mechanism. There were three industries for which patent protection was the second most important appropriability mechanism. The drug industry is one of them. For one industry it was third; for most it was right there at number six, beating out only “other ” and “other legal.” Lead time was the most important in 13 cases, second most important in 9 cases, third most important in 7 cases, fourth most important in 5 cases, and was rarely at the bottom. Secrecy has a similar pattern. Secrecy and lead time were highly correlated in the sample. Complementary manufacturing was most important for only 2 of 35 industries, but it was quite important—ranking number two, three, or four—for roughly 26 industries. Patents don't register highly on the score card as a mechanism to appropriate the benefit—the competitive advantage—for a product innovation. Yet patents seem to cause a lot of trouble to competitors. This experience with patent protection is not very encouraging as a model for strong protection for databases.
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS The next question is, Do sufficient protections already exist for sweat-of-the-brow works? The types of protections (I think Steve Maurer's paper is very helpful in this area) are self-help legal protection, like bilateral contracting; shrink-wrap or click-wrap licensing; and Internet copying rules, which may emerge from a consensus, if we get one, on the national information infrastructure. As many have pointed out, there are likely difficulties with bilateral contracts. They don't work if you want to disseminate a product widely, because they risk piracy and high transaction costs. Most of these other provisions either don't exist yet or are unproven. We don't know how well they will work; they have been largely untested. So, the extent of protection through these means is uncertain. I want to mention another type of protection, which I will call “self-help, other.” This category includes software support, documentation, customer support, frequent upgrades, quality certification, and timely data —all the value-added services that go into databases. I think there is every indication that people value these complementary activities, although there is, in theory, a risk that they can be over-provided. Does the European Union database initiative require change in the United States? It doesn't require it. If Europe chooses an inefficient level of protection for databases, we should not copy their policies and shoot ourselves in the foot as well. If the European Union decides that patents should last 100 years, I don't think we would want to match that 100-year patent term here. If they make a mistake, they are going to bear the heaviest burden. I want to emphasize, and I mentioned this in the Feist context, that the European Union has a very different approach both to competition policy and to intellectual property than we have in the United States. A very good example of this is the Magill case (Radio Telefis Eirann v. Commission [Magill], C-241/91P ECJ [April 6, 1995]). The circumstances of the Magill case are the following: TV stations broadcasting in Ireland and Northern Ireland published individual daily program listings in local newspapers. The plaintiff in this case, Magill, wanted to amalgamate these listings into a weekly TV guide format. The TV stations challenged this as a copyright violation. The case was appealed to the highest European court, which decided two things. Number one, that these TV listings were protected by copyright. I suspect that would not have been the outcome in the United States. Justin Hughes commented about what is protected by copyright under Feist , and whether TV listings would be protectable in the United States. Number two, applying European competition law, the court held that the refusal by the TV stations to license the program listings for the purpose of creating the TV guide was an abuse of dominant position. It is unlikely that the facts in the Magill case would have led U.S. courts to a similar decision. U.S. courts have been extremely reluctant to hold that a unilateral refusal to license intellectual property can be antitrust violation. Public policy with respect to intellectual property and the exercise of market power is quite different in the United States than in the European Union. You have to think about their database initiative in the context of their approach to intellectual property and to competition policy. The European Union has a safety net. If they believe that certain types of information should be more widely licensed, they can compel it under E.U. competition law. Such an outcome is less likely in the United States. The next question about convergence is, Do we have to imitate the E.U. Directive? There is a reason to imitate, which is that differences in protection can be reasons for “intellectual property wars.” There is one going on now, which is the “neighboring rights war in Europe,” because they protect audio and visual works differently than we protect those works; it raises all
OCR for page 101
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS What is the standard for creativity and effort that must be satisfied to earn protection. For a utility patent, the invention must be something that is novel, useful, and non-obvious. It doesn't matter that it was easy to do. For copyright, you have the Feist minimal creativity standard. The threshold for trade secret is just that it has to be kept secret. You can't look at these issues—length, breadth, and height—in isolation. If you are going to give strong protection it should be short because there are inevitably going to be social costs associated with that protection. For example, a patent provides strong protection for a short time. Copyright, in most cases, gives very little protection; therefore, you can afford to extend it for a very long time. (Sometimes, though, copyright gives tremendous protection.) Trade secret protection applies indefinitely, but it expires once the secret becomes generally known for a reason other than theft. The bottom line is that, to the extent that database protection has significant breadth, the duration of the protection should be limited. This leads me to a conclusion that is my variant of the KISS rule, which is “keep it short, stupid”; we don't want long, high, broad protection. To the extent that database protection imposes significant economic costs, the duration of those costs should be minimized, conditional of course, on providing adequate incentives for creators of databases. For example, should the protection last 15 years or 25 years? That seems to be a fairly easy decision. Think about a 10 percent interest rate; one dollar 25 years from now is worth only nine cents today. Extending protection from 15 years to 25 years does almost nothing in terms of the financial value of the protection, assuming a flat revenue stream. At a 15 percent interest rate, if you go from a 15-year term to 25 years, the profit flow is worth something like 10 percent more, even though the term is something like two-thirds longer. There should be a “height” requirement that substantial effort has been directed to a new or improved database to qualify for protection. This sweat-of-the-brow requirement is an unusual standard. The grant of a patent or a copyright is conditioned on the creative output of the inventor or the author, not on the input that was required to make a new product or to a new literary work. It is, however, consistent with providing investment incentives for large and complex databases. There is also a suggestion that I want to credit Hal Varian for emphasizing, which is the requirement to register a database to obtain the protection. Registration is important for a couple of reasons. One is that it helps to define what is actually being protected. Two, it reduces the transaction costs of the compliance and enforcement. There can be very high transaction costs in trying to find out whether or not you are actually going to violate some sort of intellectual property right or database right. In terms of the breadth of the database protection, you should extend protection only to new or improved databases. The legislation should not “grandfather” protection for existing databases. Protection should only be for new effort. There is no need to reward databases that have already been created. In my view, protection should not extend to information that is already in the public domain, and there should be no barrier to independent creation of a database. Upgrades should not extend the term of protection for older databases. How do you actually make that happen? Registration makes it feasible to limit database protection to newer databases while allowing protection for older databases to phase out. Let me run through some conclusions. There is not a lot of evidence of need for change in this area, but I caution that it is awfully hard to prove a negative. There is evidence that strong intellectual property rights and other forms of legal assistance have not been critical to innovation in most industries, although certainly patent protection is important in some markets,
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS such as pharmaceutical products. There is evidence that strong intellectual property rights can be a barrier to innovation in some industries. I think it is clear that if database protection is significant—that is, if it matters—it should be short. Protection, if any, should focus on high-effort databases; that is, the height standard for databases should be the effort required to produce the database. It should not include information that is already in the public domain. Protection, if any, should not bar independent creation. Registration of the database facilitates appropriate protection and reduces transaction costs. Finally, there can be fair-use provisions that are designed to reduce transaction costs, and also fair-use provisions to permit access to, or copying of, a database to create new or expanded databases, provided they are not used for commercial gain. This is the scientific fair-use exemption. More effort should be directed to defining a protectable database. Another specific recommendation is the proposal to have a registry for databases, so at least you know what is protected. More work needs to be done to study the effects of stronger versus weaker intellectual property protection. Does stronger protection generate more investment? It has been sort of an assumption for many decades, but it is not necessarily true. In fact, there is evidence to the contrary. For example, computer fonts are not protected in the United States, although the programs that create them are protected. Fonts are protected in the European Union. Funny thing; most fonts actually get developed in the United States, not in Europe. Finally, have we learned anything from the year of experience with database protection in the European Union? General Discussion DR. BENSON: Dennis Benson, National Library of Medicine. You mentioned the difficulty in defining what a database is. The high-effort database might be equally hard to define. Could you elaborate on that? DR. GILBERT: Yes, I don't know how you make a quantitative definition here. I think there has to be some analogous standard. We have a standard of creativity for copyright, the Feist standard, which is not a quantitative standard. None of the standards for patent protection are truly quantified. Nobody really knows what “novel” is; nobody really knows what “nonobvious” is. To say that we have not quantified this wouldn't exactly put it on unprecedented grounds. I do think that there should be some at least qualitative standard that says if you merely assemble things in one person-day of work, you are not. really contributing. There is a social bargain; the bargain is that you are going to get some protection. It is not clear exactly how much protection you will get given the other forms of protection that are already available. In some instances, this would generate substantial protection. There should be some bargaining, something in return. DR. BENTLEY: Most of the databases we discussed this morning are continuously updated, daily, hourly. So, every time you have to put in a new entry you have to update the registration? DR. GILBERT: That is a very good point. Some implementation questions here clearly need to be thought out.
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS It strikes me as not actually inappropriate to do the following, which is to say that there is a particular version of the database that is protected, and it is registered as a protected database. You get a certain number of years, 15 years or whatever it is, on that version of the database. You are talking about rights with respect to the improvements on it. If you own the rights to something, 90 percent of which is protected by copyright, and 10 percent of which is new and is unregistered, then, if someone copies that 100 percent, that is still a copyright violation and it may still be a database violation, but with respect to the 90 percent, not with respect to the other 10 percent. If you haven't registered the 10 percent of the upgrades, those upgrades might not be protected. You might decide on some intervals at which to register another version of that database. That risks putting the old version into the public domain after a certain time period. It creates problems, but I don't think insurmountable problems. The other side of that is the risk that the database protection becomes of indefinite term because you always have a new database every day. MS. LIPPINCOTT: My question refers to your reference to the case of Magill, the European Union, and the television listings. What would compel the television stations to set a reasonable price, if they just had to say, yes, they would negotiate and make them available. What if they said they were available for $1 million a day? DR. GILBERT: Would that have been an abuse of dominant position? Is that your question? I don't know. That particular issue was not litigated. In all of these cases, there is a presumption that there is a price that becomes equivalent to a denial. The concern expressed in court was that this was an exercise of a dominant position because they had a monopoly on these listings and because there is a view that, in effect, they were afraid that these listings in the TV guide format might compete with a product that they were thinking about producing, perhaps at some future time. If they had charged $1 million a day for the listings, I expect certainly that the economic reasoning would be the same. MR. REICHMAN: Two questions. The social safeguard copyright law is not the standard for activity that you are using. It is the ease with which independent creation occurs. So, one question is, Do you have similar ease of independent creation with regard to databases? The European Union assumed that it is, and I wondered if you considered that. Secondly, you seemed to consider only exclusive copyright regimes, and then you kind of threw trade secret law in there. You don't get an injunction against trade secret law. It is a liability regime. If you exclude that, deducing from your own analysis, because there is no appreciable creative contribution, would it not be better, I wonder, to move toward a liability regime, some form of unfair competition? DR. GILBERT: Those are both good questions. I will try to answer them in finite time, at least respond to them. First, with respect to copyright and the ease of creation, one other point is that copyright isn't always something that gives very narrow protection. One case that I think is particularly instructive is West Publishing, which had a copyright and produced a compilation of law documents in a particular format, called the Star Pagination format. Even though the database from which the legal opinions came eventually became widely available through electronic distribution, the Star Pagination system had become a sort of library standard used by the legal system. Therefore, no one could really come out with another product without violating the West copyright on its method. So, it acquired some very significant protection. There is even an example that I have heard of in which a particular access code on a software program was written in the form of a copyrighted poem. The argument was that you
OCR for page 101
PROCEEDINGS OF THE WORKSHOP ON PROMOTING ACCESS TO SCIENTIFIC AND TECHNICAL DATA FOR THE PUBLIC INTEREST: AN ASSESSMENT OF POLICY OPTIONS could not reverse engineer the software code to get into this program because it would be a violation of copyright to reverse copy the poem. So, copyright protection can sometimes have some very extraordinary consequences. Would database protection involve the same problem? I think it depends on how the law is written. It might, or it might not. We have certain features of copyright that lend to instances of considerably strong protection, more so than most patents. Number two, with regard to the appropriate regime, whether it should be a liability regime along the line of trade secrets, I really haven 't thought a lot about that. I would merely say that I know a comment was made earlier about a database protection regime along the lines of unfair competition and trademark. There are some instances in which trademarks have been used in a very restrictive way. It can work that way as well.