Database Protection at the Crossroads: Recent Developments and the Long Term Perspective

II. Potential Impact of the Database Protection Laws on Science and Education

A. The User-friendly Rules of Copyright Law
B. Unbalanced Rules of the Sui Generis Model
C. Long-term Implications of the Sui Generis Model

III. Recent Developments: The Shift to an Unfair Competition Model

A. The Administration's Position
B. A Negotiated Discussion Draft in the Senate
C. Uncertain Future of the Database Legislation
IV. Conflicting Regulatory Visions of the Networked Environment

Section I: Commodification of Data in the Networked Environment: The Bigger Picture

The convergence of digital and telecommunications technologies has greatly expanded the already bright economic prospects for information goods of all kinds; but it has also unsettled the legal architecture on which the free market economies have previously been grounded. Information products behave differently from the tangible, physical products of the Industrial Revolution; and the legal paradigms that we have applied to balance incentives to create against both public good uses of information and the discipline of free competition are stretched past the breaking point. Basically, we have to rethink how best to structure competition for information goods in the emerging, worldwide information economy.

The convergence of digital and telecommunications technologies creates new markets for data and information. This is accompanied by a powerful movement to commodify data and information that had previously been treated as a public good - that is, as an inexhaustible, indivisible, and ubiquitous component of the public domain. The momentum that these forces have generated would sooner or later have faced science and education with serious new challenges even in the absence of a new intellectual property right in collections of data. The adoption of a mega property right in noncopyrightable collections of data by the European Union - in a haphazard, slapdash manner, with little serious economic or empirical investigation - merely precipitated a crisis that was already well underway.

The roots of the crisis stem from the fact that digital telecommunications networks permit publishers directly to control the uses of information goods by contract, without intervention by the state, for the first time since the advent of the Guttenberg printing press. In effect, as I have explained in an article about to appear in the University of Pennsylvania Law Review, online delivery has "restored the power of the two-party deal" with regard to information goods and diminished the dependence of publishers on artificial legal fences that copyright laws and other related rights supplied in the print environment.

This readjustment of the legal landscape is occurring from several different directions. First, new technologies give publishers the power of fencing off information goods by means of technical devices - encryption devices - and new laws make it a civil or criminal offense to disarm or tamper with these technical protection measures. ^{[1] Such measures oblige would-be users to gain access to information goods via an electronic gateway where the users must identify themselves and acknowledge the rights of the gatekeeper to the information goods, as, for example, expressed in copyright management information codes. The new laws that defend the owners encryption devices also forbid users from tampering with the owner's intellectual property identity tags.}

Second, new contract laws are being developed to permit publishers to impose non-negotiated contracts of adhesion upon all would-be users seeking to cross the electronic threshold and gain access to the information delivered online (Art. 2B of U.C.C.). These click-on, or shrinkwrap, licenses could ignore or override all the public interest exceptions that previously favored users or competitors under the existing legal structure; e.g., they could override the right to make non-infringing uses of copyrighted works or the right to reverse engineer subpatentable innovation.

At present, the power of these electronically imposed adhesion contracts is potentially limited by the existing legal architecture, and especially by the traditional intellectual property paradigms on which it rests. As we shall see, copyright law expressly permits many, if not most uses of copyrighted works that publishers would like to restrict by means of online licenses. This casts shadows on the validity of such contracts because they might disrupt the federal system (pre-emption arguments) or overstep the constitutional guarantees of free thought and expression.

So a third line of attack is to devise new legal paradigms - especially new sui generis intellectual property rights, like the European Union's sui generis right to protect the contents of databases, which give legislative approval to forms of protection that were previously unknown or questionable under traditional intellectual property law. The creation of new intellectual property rights in data thus makes it harder to resist arguments that publishers who subject online delivery of databases to technical measures and to contracts of adhesion have violated fundamental public policies derived from the copyright laws precisely because the database protection laws seem to permit acts (and foster policies) that overtly contradict or override the limits previously established by copyright and other traditional legal models.

II. Potential Impact of the Database Protection Laws on Science and Education

Let us explore the potential impact this can have on science and education if proper countermeasures are not taken. Suppose we buy a chemical handbook or a scientific article, with appended data, published in a peer-reviewed journal, in hard copy format. These works attract copyright protection, and we shall assume that they meet the eligibility criteria of that body of law.

A. The User-Friendly Rules of Copyright Law

The rules of copyright law constitute a balanced regime of public and private interests. In retrospect, we are struck by the friendly treatment this body of law gives to users and competitors alike.

1) For example, we could immediately use all the data and all the ideas disclosed in the book or article because copyright law does not protect idea or data and it does not protect against use of expression as such (only certain uses).

2) We could independently rewrite our own version of the same article (and sell it) because copyright law allows independent creation, and all the unprotected data is spread out before our eyes (we must not pass off and should attribute sources).

3) We could combine the data and the ideas into a follow-on article or book that borrowed the originator's unprotected ideas and data (but not his stylistic expression).

4) We could combine the published data and ideas with other data and ideas into a multiple or complex interdisciplinary database without permission or additional payment to the originators, in part because these are unprotected matter and in part because copyright law does not prohibit use as such, only certain uses, such as reproduction or adaptation of expression only;

5) Even if we made classroom or scientific use of the protected expression for nonprofit purposes, these uses might well be fair or privileged uses under U.S. copyright law and would possibly be subject to compulsory licensing under E.U. copyright laws.

6) Finally, having once purchased the book or the article, we could lend it to other scientists or give it to them (first sale doctrine), borrow it from a library, use it as often as we like for virtually any purpose, and make photocopies of it for scientific purposes under the fair use doctrine (U.S.) or the private use doctrine (E.U.). ^[2]

B. Unbalanced Rules of the Sui Generis Model

Now, suppose that the contents of the chemical handbook or of the scientific article were disseminated online and surrounded by technical fences as previously described. Suppose further that they are protected by the E.U.'s sui generis exclusive property right in noncopyrightable collections of data or by the U.S. version of that right, as set out in H.R. 2652 (March 1998). The House of Representatives adopted H.R. 2652, and the Subcommittee on Intellectual Property then attached it to the Digital Millennium Copyright Bill, which became H.R. 2281, as sent to the Senate. The database portion, however, was dropped prior to its enactment.

The E.U. Directive protects any noncopyrightable database that is the product of substantial investment against extraction or reutilization of the whole or of any substantial part (evaluated quantitatively or qualitatively). Hence, this law could protect the noncopyrightable data appended to the article in question or collected in the handbook, which might now be disseminated online. This protection lasts as long as new investment is made in updates or maintenance; hence perpetual protection is a likely result, despite a nominal 15-year term. There are no exceptions whatsoever for "reutilization" by scientific and educational bodies, and there are no mandatory exceptions for "extraction" for scientific and educational purposes (although states may adopt this exception for noncommercial purposes). There is an exception for use of an insubstantial part, but the risks of invoking it are high.

U.S. Bill H.R. 2652 (105th Congress, 2d sess., May 12, 1998), later H.R. 2281, was built on essentially the same model. It protected against use or extraction in commerce of all or a substantial part of a protected collection of information that is the product of substantial investment if such use or extraction would "cause harm to the actual or potential market" for a product or service that incorporates the collection. Publishers claim that a single lost sale would fit within this definition of harm to the market. Any substantial new investment in updates or maintenance would prolong protection beyond fifteen years, with no limit. There were no exceptions for science and education but, at the last minute, a provision tacked onto the bill stated that scientists would be liable only for harm to "actual markets," but not for harm to "potential markets."

If we apply this proposed regime to noncopyrightable databases delivered online and surrounded by technical protection devices, we obtain the following results:

1) We cannot make any uses of the data or ideas that are not permitted by the form-contract site licenses that regulate access to the online database from which the data or information are extracted.

2) The site license can charge one price for reading or consulting the database, a second price for downloading it, and a third price for using it or reusing it in other contexts.

3) Even though we have paid to access the ideas and data (and they are not copyrightable by definition), we cannot use them in any ways that are not permitted by the terms and conditions of that site license.

4) Therefore, we cannot independently generate a similar article or study, based on the same data or ideas, without permission, even though these data and ideas are now revealed to the public. Because they no longer enter the public domain, we need to obtain a new grant or substitute funding to repeat the collection process, in which case scarce funds will have been used to duplicate the creation of knowledge already in existence. This, of course, contradicts the norms of science, which favor building on previous discoveries and research results.

5) In many instances, the data will be based on one-time events that cannot physically be regenerated. Even when regeneration is possible, the cost in relation to the niche market of likely users is normally so high that no second comer is usually willing to regenerate the data. Hence, sole-source providers are a dominant feature of the database landscape, and real competition is the exception.

6) No one can combine the data or ideas into a more efficient follow-on product without a license; the licensor is under no duty to grant such a license: and the sole-source provider wants no competition from follow-on products. This also suggests that the price will not be set so high as to encourage independent creation of the same data, when otherwise feasible. So potential producers of follow-on products will tend to invest in other activities, thus reinforcing the sole-source structure of the existing market.

7) We cannot combine data legitimately accessed from other databases to make a complex new database for addressing new scientific problems without obtaining additional licenses and permissions because publishers claim we violate their redistribution rights. (Notice that there is never a sale that exhausts the publisher's rights, only a license, which the proposed Article 2B of the U.C.C. would make perpetual.)

8) The prices to be paid will be as high as the market will bear in the absence of competition, and the natural and artificial barriers to entry remain extremely high (sole-source the norm).

9) We cannot lend, borrow, or transfer our data to other scientists (although we paid for access) because there is no first sale, only a license; we need permission (and the payment of fees) to accomplish this, and we would harm the market in so doing.

10) The data will not enter the public domain for at least fifteen years, and possibly never, if the private party continues to invest in maintenance or updates of a dynamic database.

11) If government data are privatized (U.S.) in ways that unduly restrict access or are sold directly on onerous conditions, taxpayer-financed data will be sold back to science and education at monopoly prices (unless subsidized by the state again). This will routinely occur in the E.U. (example of Italian professor).

12) No limits on licensing - the data provider can override even some of the exceptions contained in the bill (in the E.U., not the right to use insubstantial parts) and any scientific and educational exemptions (in U.S., if not E.U. law) that are eventually adopted. Alternatively, the publisher can charge more for access and reuse if he knows that the state will require him to charge little or nothing for extractions by scientific and educational bodies. The only limit in most cases is what a monopoly market will bear. ^[3]

C. Long-term Implications of the Sui Generis Model

We believe that the long-term implications for science and education are very damaging. Consider the following observations.

1) "All science operates on databases. The near-complete digitization of data collection, manipulation, and dissemination over the past 30 years has ushered in what many regard as the transparency revolution. Every aspect of the natural world, from the nano-scale to the macro-scale, all human activities, and indeed every life form, can now be observed and captured as an electronic database. There is barely a sector of the economy that is not significantly engaged in the creation and exploitation of digital databases, and there are many - such as insurance, banking, or direct marketing - that are completely database-driven." ^[4]

2) "Science builds on science. In all areas of research, the collection of data sets is not an end in itself, but rather a means to an end, the first step in the creation of new information knowledge, and understanding. As part of that process, the original databases are continually refined and recombined to create new databases and new insights. Typically, each level of processing adds value to an original (raw) dataset by summarizing the original product, synthesizing a new product, or providing an interpretation of the original data [NRC 1995].

3) The processing of data leads to an inherent paradox that may not be readily apparent. The original unprocessed, or minimally processed, data are usually the most difficult to understand or use by anyone other than the expert primary user. With every successive level of processing, the data tend to become more understandable and frequently are better documented for the nonexpert user. As the data become more highly processed, documented, and formatted for easier use, they also are more likely to become copyrightable. Yet it is the raw, noncopyrightable data that are typically of greatest use and value to researchers, who can manipulate and experiment with the original measurements in pursuit of their own research goals. Thus, strong intellectual property protection of noncopyrightable data sets, which generally also have the least commercial marketability, could disproportionately affect the availability of data most commonly used in basic research and higher education.

4) U.S. science is premised on easy access to a robust public domain, easy reuse and recompilation of existing data, and easy access for value-adding purposes.

5) We expect generally higher prices for all research; higher administrative and transaction costs; a blocking-off of access to upstream data now freely available. We also predict that universities will begin to set limits on what investigators can do with data; there will be a gradual capturing of government-generated data; insuperable transaction costs will arise in the creation of complex databases from multiple sources - public and private - to address new problems; disincentives to reuse data in existing collections so as to avoid the "harm to markets test" will be created; there will be diminished scientific cooperation and exchange; lost opportunity costs will spread." ^[5]

In a worst case scenario, this type of database protection law would disrupt the system of cheap access to upstream data for purposes of basic research, which would then gradually reduce downstream applications of scientific breakthroughs subject to exclusive property rights. This scenario could undermine a key comparative advantage of the U.S. economy. In sum, putting a strong property right too far upstream too soon could have a disastrous effect on the long-term competitiveness of the U.S.

III. Recent Developments: The Shift to an Unfair Competition Model

Concern about these issues led the Presidents of the National Academy of Sciences, the National Academy of Engineering, and the Institute of Medicine to send several letters to the Administration and to leading members of Congress responsible for this legislation. The International Council for Science (ICSU) likewise intervened at relevant meetings of the World Intellectual Property Organization (WIPO), and documents submitted by ICSU have played a prominent role in successful efforts to block rapid or premature efforts to launch an international treaty regulating databases that might be built around the unbalanced E.U. model. ICSU has also begun direct consultation with publishers' representatives, with a view to working out some common understanding applicable to database protection issues affecting the scientific community.

Initial efforts to slow the process down in the U.S. were not successful. The concerns of the scientific community were at first ignored by the Patent, Trademark, and Copyright Office (PTO), given little weight by the Copyright Office, and virtually ignored again by the House Subcommittee on Intellectual Property, which passed the Database Bill as an attachment to the Digital Millennium Copyright Bill, H.R. 2281.

A. The Administration's Position

More recent developments, however, have been favorable to the interests of the scientific and educational communities. To begin with, interagency review within the Administration has produced a series of important position papers that support the theses already put forward by the scientific community and that reject the one-sided views initially expressed by the PTO. On August 4, 1998, the General Counsel of the U.S. Department of Commerce wrote Senator Patrick Leahy, Ranking Minority Member of the Senate Committee on the Judiciary, to advise that any new database legislation must avoid capture by private parties of government data and that "any effects [it may have] on non-commercial research should be de minimis." This letter was sent as a consensus position of all departments and agencies of the Administration.

Consistent with these views, the Administration expressed concerns about possible "increase[d] transaction costs in data use, particularly where larger collections integrate datasets originating from different parties or where different parties have added value to a collection through separate contributions. . . This is especially important for large-scale data management activities, where public investment has leveraged contributions from the private and non-profit sectors."

The letter went on to express further concerns "that the . . . exception for noncommercial research and educational uses does not ensure that legitimate non-commercial research and educational activities are not disrupted by the prohibition against commercial misappropriation" and that sole-source providers might unduly burden "issues of access and use." Finally, the Administration's letter referenced the Department of Justice's "serious constitutional concerns that the First Amendment restricts Congress's ability to enact legislation" of this kind, and that other constitutional obstacles would have to be overcome.

These constitutional impediments were elaborated in a 26-page study by the Legal Counsel of the U.S. Department of Justice, dated July 28, 1998, which constitutes a serious indictment of the sui generis model then pending before Congress. On September 28, 1998, moreover, the Chair of the United States Federal Trade Commission wrote the Chair of the House Committee on Commerce to express additional concerns about the pending database legislation. In particular, the FTC found that "certain provisions within the proposed legislation raise concerns about possible unintended, deleterious effects on competition and innovation" (at 7), and that "the potential for anti-competitive use of a 'collection of information' is substantially increased when there is only a single source for the data" (at 14).

Finally, it should be noted that the USPTO held a conference on April 28, 1998, to reexamine database protection and access issues. In July 1998, this Office issued a Report that, while endorsing sui generis database protection, expressed support for some of the concerns that the scientific and educational communities had been voicing. ^[6]

B. A Negotiated Discussion Draft in the Senate

In late July, the Chair of the Senate Committee on the Judiciary, Senator Orrin Hatch, invited the stakeholders to participate in strenuous negotiations, which lasted from August through early October, under the leadership of Senator Hatch's Counsel for Intellectual Property, Edward Damich, and with the participation of the counterpart staffer in Senator Leahy's office, Marla Grossman. The U.S. Academies took the unusual step of using internal lobbying funds to participate directly in these negotiations. They submitted a series of alternative proposals aimed at providing a balanced piece of legislation that would protect publishers against free-riding conduct and preserve the incentive to invest (so-called unfair competition approach) without creating a strong exclusive property right in the building blocks of knowledge. ^[7]

Although the direct negotiations produced no major breakthroughs or compromise solutions, they did succeed in clarifying the different positions. It seems fair to say that, when exposed to direct interrogation, the publishers' detailed demands were so exorbitant that the scientific and educational communities' initial concerns appeared modest by comparison. Indeed, in response to one hypothetical situation after another, the publishers made it clear that the exclusive property right they championed, linked to the online delivery systems I have described, would inflict much greater disruption on the basic scientific enterprise than critics had previously estimated.

Perhaps because the publishers' actual demands amply confirmed the concerns that the Administration's own position papers had expressed, the final phases of the negotiations, as mediated by the Senate staffers, produced far-reaching modifications to H.R. 2281. These changes substantially implemented the Academies' own position.

1) First, the exclusive property right approach was ultimately abandoned in favor of a true "misappropriation" (unfair competition) approach. This was accomplished by conditioning liability on acts that "cause substantial harm to the actual or neighboring market" of database proprietors, ^{[8] and by inviting courts, in the legislative history, to determine "substantial harm" in light of "whether the harm is such as to significantly diminish the incentive to invest in gathering, organizing, or maintaining the database."} [9]

2) Second, a full carve out that would immunize customary scientific activities was adopted by the Senate staffers, in place of the limited and unacceptable "fair use" approach that the Patent Office had recommended. A "fair use" approach, modeled on copyright law, is inadequate because other basic copyright immunities and exceptions, including the idea-expression dichotomy, are not carried over into the database protection environment. On the contrary, because a database law protects collections of facts and data that are ineligible under copyright laws, most scientific activities that were previously permissible would become infringing acts under such a law. The burden would then be on scientists to show that a vague fair use exception should excuse some of these infringing acts from whatever test of harm was adopted.

In contrast, the Academies successfully argued that traditional scientific activities should remain untouched and unhampered by any new database protection law, exactly as the government's own position paper maintained. ^{[10] To this end, §1304 of the final version of the Hatch Discussion Draft stated that "nothing in this chapter shall prohibit or otherwise restrict the extraction or use of a database protected under this chapter for the following purposes:}

1) for illustration, explanation, or example, comment or criticism, internal verification, or scientific or statistical analysis of the portion used or extracted; and

2) in the case of nonprofit scientific, educational, or research activities by nonprofit organizations, for similar customary or transformative purposes.

Only if a scientist caused substantial harm to the database-maker by using unreasonable and non-customary amounts of the collection for a given purpose, or if he in fact made a substitute for the original, or otherwise sought to avoid paying for the use of research tools devised as such, would liability kick in. On this approach, the burden would be on publishers to show that scientists had crossed the line of permitted, traditional, or customary uses, which are immunized. The principle that science should be left no worse off after enactment than it was before, as enunciated by the National Research Council, ^{[11] would thus have been implemented.}

3) Additional immunities and exceptions favoring certain instructional and library uses of databases were also defined, ^{[12] although more thought needs to be given educational users generally in this context.}

4) Efforts were made to reduce the likelihood that private interests might permanently capture government-generated data, although more remains to be done on this entire subject.

5) A clearly-worded duration clause ended protection at fifteen years, which largely overcame the risk of perpetual protection. A rudimentary database deposit scheme was also established, which increased the payoffs for the public domain, although more incentives are still needed to ensure that deposits are actually made.

6) The need for regulation of licensing terms and conditions was expressly recognized in a series of provisions requiring periodic studies of the misuse doctrine in connection with licensing agreements or the use of technological measures that might frustrate the "permitted acts" clause of the Bill; with sole-source provider contracts that imposed unreasonable terms or conditions; with tying or other practices traditionally recognized as abusive; and with practices shown to have "prevented access to valuable information for research, competition, or innovation purposes." ^{[13] The legislative history then clarified that courts were free to apply a judge-made doctrine of abuse to licensing agreements from the time of enactment and need not "refrain from applying the doctrine of misuse until the study is completed."} [14] There was some possibility that these restraints on licensing could have ultimately been codified in the operative clauses of the act itself.

7) The legislative history also clarified the definition of databases in ways that tended to exclude ordinary literary works, ^{[15] and it denied protection "to any ideas, facts, procedure, process, system, method of operation, concept, principle, or discovery, as distinct from the collection that is the product of investment protected by this Act."} [16] It would, of course, be better to codify these provisions expressly in the Act itself.

C. Uncertain Future of the Database Protection Legislation

1) The foregoing presentation reveals the extent to which the Hatch Discussion Draft evolved away from the strong exclusive property right approach, adopted in the E.U. Directive, toward a minimalist unfair competition approach that protected publishers against piracy while consciously avoiding harm to science and other public-good users of data. Of course, not all the issues of concern to science were addressed in a fully satisfactory manner; but given the need for compromise and consensus, the ability of the staff to produce a relatively balanced bill from such unpromising material as the House Bill deserves commendation.

Perhaps the biggest unaddressed issue was that of value-adding uses. The Discussion Draft does not resolve the tensions between database publishers, who seek to control value-adding uses of a protected collection, and a dissident group of publishers and allies, who believe value-adding uses should be free. On this point, the Academies proposed a scheme favoring easy use of data for commercial value-adding purposes in exchange for the payment of reasonable royalties under an automatic licensing scheme; but neither side would accept this approach. Nevertheless, under the misappropriation approach to "substantial harm," as adopted in the Draft, courts could work out the criteria for balancing incentives to invest against incentives to compete for the short run, and these case-by-case solutions could be legislatively evaluated later on.

(2) In the end, the Hatch Discussion Draft was not adopted mainly because time ran out in which to remove the last remaining wrinkles that prevented an agreed compromise. As a result, the database component of H.R. 2281 was stripped from the Digital Millennium Copyright Act enacted at the end of the legislative year, and work on databases will begin all over again under the aegis of the new Congress, which will convene in January 1999.

(3) It is not clear that the Hatch Discussion Draft will even constitute the basis for the next round of discussion. For one thing, the U.S. government is expected to develop an administration bill, which could differ significantly from the bills that were heretofore put on the table. It remains to be seen if the government's bill will ultimately treat science as favorably as the final version of the Hatch Discussion Draft.

At the same time, some members of the coalition that opposed H.R. 2281 may now feel inclined to present a bill that implements unfair competition principles more aggressively than was contemplated in the final Hatch Bill. While there is room for improvement in this regard, there are also new dangers to be avoided. For example, one of the biggest publishers backing H.R. 2281 is reportedly unhappy with the provisions that clearly limit protection to the fifteen year period, as embodied in the Discussion Draft. This unhappiness has increased because of a recent decision by the U.S. Court of Appeals for the Second Circuit, which denies strong copyright protection for the dissemination of legal decisions under existing methods. ^{[17] There is, accordingly, some growing pressure for a so-called unfair competition approach that would become unlimited in duration, even though this could easily degenerate into a de facto exclusive property right conferring perpetual protection by the back door.}

(4) Whatever happens next, I believe that the final version of the Hatch Discussion Draft constitutes a milestone along the route towards a balanced model of database protection, and that its lessons should inform the next round of legislative deliberations. There is evidence that the Japanese government has also decided to embark upon an unfair competition approach, which presents a unique opportunity for the U.S. and Japan to present a united front to the rest of the world. In that event, it seems likely indeed that the rest of the world would move in the direction of a minimalist unfair competition regime, which might leave the E.U. alone to continue its experiment with a strong property right or to modify its Directive so as to obtain a more balanced system of protection with fewer social costs.

Moreover, the scientific community will experience serious challenges to the unrestricted flow of data that has been its goal, regardless of the approach to database protection that ultimately emerges from Congress and the legislation of other countries. This follows because, as pointed out at the beginning of this paper, publishers can control the dissemination of data by combining technical protection measures with adhesion contracts in the online environment, even without the adoption of specific database legislation. While the presence of an intellectual property right would strengthen the publishers' position and put the scientific community under grave legal disadvantages, the absence of an exclusive property right would not free science from the need to rethink its whole approach to maintaining the unrestricted flow of scientific and technical data in an emerging information economy.

(5) A new study by the National Research Council, currently underway, will examine these issues, beginning next year, and it is hoped, will shed further light on the options for science in this regard. Meanwhile, it seems clear that science will have to consider ways and means of reconciling a greater degree of commercialization for databases generated within the academic community with the need to maintain privileged access to the same databases for scientific and other public interest objectives. Universities and scientific institutions that generate data will thus have to consider measures for disciplining grants, and uses of data obtained from grants. Separate channels for the distribution of scientific data may have to be created, with particular rules for participating scientific organizations. It may also prove desirable to develop an extended licensing authority for scientific data that could administer these resources with low transaction costs and uniform rules for commercial and non-commercial users.

(6) In general, efforts must be made to preserve the sharing ethos with respect to publicly-generated scientific data, to encourage those who invest in the production of privately-generated data to provide price discrimination in favor of science and education, and to develop differentiated products for the scientific and educational communities. These objectives will, in turn, require close collaboration with governments, with a view to ensuring that data generated at the taxpayers' expense remains available for scientific and educational purposes, and that efforts to stimulate greater investment in the development of new databases do not end by creating barriers to entry or otherwise discouraging follow-on innovation and public good uses of the building blocks of knowledge.

IV. Conflicting Regulatory Visions of the Networked Environment

A. The convergence of digital and telecommunications technologies has restored the power of publishers to control online delivery of information by combining encryption devices with intellectual property rights and standard-form adhesion contracts. The possibilities inherent in this phenomenon have given rise to two conflicting philosophies. One wants strong intellectual property rights and unfettered contractual power to work things out in a Utopian, perfect market setting. The other, fearing market imperfections, wants strong regulatory measures to translate pre-existing constraints on the exercise of intellectual property rights into the digital environment, with a view to favoring specified public good users by legislative prescriptions.

B. I am suspicious of both philosophies and wary of the risk that, under either approach, premature legislative action will be taken before we understand the empirical realities that need regulating. In this regard, we do not know enough to know what kind of property rights are needed at what levels of competition, and we are likely to be blinded by existing legal models, which are in fact unsuited to vast segments of the information economy. Patents and copyrights, for example, work well to stimulate major creative achievements - large grain-size intellectual productions, known as inventions, and works of authorship - but they are totally unsuited to small grain-size innovation that depends on mere investment and sweat-of-the-brow labor, without any corresponding intellectual achievement. ^[18]

At the level of mere investment and routine innovation, we do not need strong intellectual property rights to overcome market failure because there is no shortage of investment once the causes of that market failure are removed. We should not give entrepreneurs legal monopolies to undertake investments they would make anyway, in their own business interests, because the social costs of such monopolies' in lessened competition and other negative collateral effects are almost certain to outweigh the benefits.

C. I am told that strong property rights, together with private contracts, can solve all problems in a perfectly free market. But standard form adhesion contracts - click-on licenses - are not contracts in the usual sense; they are dictated, not negotiated, and the market for databases is characterized by an abundance of natural monopolies. If we adopt unbalanced intellectual property laws that unnecessarily provide incentives for publishers to invest, but we discourage follow-on innovation and public good uses of the databases we generate, the end result may be bad for the database industry and dreadful for our whole technological super-structure, which depends on the unrestricted flow of upstream data. Instead, we need a regime that loosely preserves a balanced relationship between public and private interests, which courts can develop gradually in response to the empirical conditions of the evolving information economy. ^[19]

D. To me, the fallacy of most proposals concerning database protection is that they ignore the dual nature of data and information as such. On one level, data are raw material of the information economy, a basic ingredient of the public domain, from which scientists and entrepreneurs both draw to fashion their respective products. On a second level, data and information are bundled into downstream products that rightly attract strong intellectual property rights and related contractual agreements. The mistake is to think that intellectual property rights that have empirically been well-suited to downstream applications - mainly derived from the patent and copyright models - are equally well-suited to upstream regulation of the data as an input into the process of innovation.

The opposite is true. If we put strong intellectual property rights too far upstream too soon, we will balkanize the public domain and make the transaction costs of recreating it by contracts prohibitively expensive and complex. ^{[20] This, in turn, will impede the cumulative and sequential development of technical paradigms} [21] by depriving routine innovators of access to the building blocks of knowledge.

The truth is that we have always left small grain-size innovation to weaker forms of entitlement, that is, to liability principles rooted in unfair competition law, and this is a basic principle on which the competitive economy of the industrial revolution was constructed. ^{[22] I suspect that these are the lessons we need to carry over to the information economy - lessons sounding in reverse engineering and the reuse of ideas, rather than in legal monopolies on products of routine innovation. Until convincing evidence to the contrary is presented, we should address the risk of market failure in the information economy by erring on the side of underprotection rather than overprotection. This is logical because there is no real or potential shortage of investment once the causes of market failure are controlled; and it is sound public policy, because we do not wish needlessly to monopolize the building blocks of knowledge, slow down value-adding innovators, or retard the progress of science.}

The inclination to place strong intellectual property rights in upstream fragments of knowledge is contrary to our entire intellectual property tradition and to our basic constitutional heritage. ^{[23] We have always favored the free use of facts and ideas, and we have always sought unbridled competition with respect to the products of mere investment. I do not doubt that we shall have sub-optimal investment in the production of databases if we fail to protect publishers against certain forms of piratical conduct that threaten to deprive them of the fruits of their investment.} [24] But I also fear that, if we combat this risk of market failure with strong monopolies in the collections of data, we shall balkanize the information economy and end by recreating the medieval economic quandary in which products could not flow across countries or continents because too many feudal monopolists demanded payments every few miles down the road. ^[25]

In approaching something as radical as protecting noncopyrightable collections of data that have been free of protection for hundreds of years, I believe we should err on the side of caution and not succumb to various pretexts for introducing strong legal monopolies into the emerging information economy. We do not want public scientific data to be colonized in the way that some companies have managed to colonize the distribution of judicial opinions in this country: the West Publishing Co. model is not the proper model for science, for databases, or for the information economy in general. We want information to flow as freely as possible upstream, consistent with minimalist legal measures to deter free riders and to encourage entrepreneurs to make the investments their own business sense tells them to make.

To achieve this result, I believe we should fashion weak entitlements, rooted in the liability principles of unfair competition law, without risking the creation of exclusive property rights in the raw materials of the information economy. We should not imagine that these liability regimes must necessarily be rudimentary or incapable of technical and conceptual refinement. New models and proposals are emerging, and digital technology makes it possible to implement these more refined liability regimes without loss of efficiency and without sacrificing the public good uses of data and information to the monopolists' power. In the Information Age, as in the Industrial Revolution, we should continue to believe that competition is the lifeblood of commerce, and we should accordingly structure all legal entitlements so as to produce a high degree of competition and maximum upstream use of data and information.

Should we err on the side of caution, and underprotect the building blocks of knowledge, we can always adjust the level of protection upwards later on, in the face of compelling empirical evidence. But the opposite is not true; acquired rights and legal monopolies cannot so easily be eradicated once they are legislatively established. The wrong decisions today could compromise our future technological superiority, which rests on an appropriate division between upstream and downstream uses of data and information.

Notes:

^{[1] J. H. Reichman and Jonathan Franklin, Privately Legislated Intellectual Property Rights: The Limits of Article 2B of the U.C.C., forthcoming in U. PENN. L. R. (1999).}

^{[2] See Reichman & Samuelson (1998). Return to Text}

^{[3] See Reichman & Samuelson (1998). Return to Text}

^{[4] Uhlir (1998). Return to Text}

^{[5] Uhlir (1998).}

^{[6] See U.S. Patent and Trademark Office Report on (and Recommendations from) April 1998 Conference on Database Protection and Access Issues, USPTO, Department of Commerce, Washington, DC, July 1998. Return to Text}

^{[7] See, e.g., NAS, NAE, and IOM Proposed Amendments to H.R. 2281: Explanatory Memorandum (Part I), August 13, 1998; NAS, NAE, & IOM, Opponents' Revised Amendments to H.R. 2281, Sept. 4, 1998 (concerning (1) permitted acts for scientific, educational, and research purposes; (2) exclusions; (3) definition of "collections of information"; (4) licensing.) Return to Text}

^{[8] Hatch Database Discussion Draft (HDDD), Oct. 5, 1998, §13021. Return to Text}

^{[9] HDDD, Proposed Conference Report Language, §1302, at 33. Return to Text}

^{[10] See Letter from the General Counsel of the Department of Commerce, supra note _____. Return to Text}

^{[11] NRC, BITS OF POWER:________ (NAS Press, 1998). Return to Text}

^{[12] See HDBDD, §1307. Return to Text}

^{[13] See HDBDD, § 1305, 4. Return to Text}

^{[14] HDBDD, Proposed Conference Report Language, at 36-37. Return to Text}

^{[15] Id., at 31. Return to Text}

^{[16] Id. Return to Text}

^{[17] [West Publishing Decision, 1998]. Return to Text}

^{[18] See J. H. Reichman, Solving the Green Tulip Problem (YU Conference, July 1998). Return to Text}

^{[19] See id. at ______. Return to Text}

^{[20] Reichman & Franklin, U. PA. L. REV. (1999). Return to Text}

^{[21] See id. at _____. Return to Text}

^{[22] See Nelson; Scotchmer; Reichman, Legal Hybrids. Return to Text}

^{[23] See Reichman, Legal Hybrids; Eisenberg. Return to Text}

^{[24] Tyson; Hunsucker. Return to Text}

^{[25] Drahos, Into Feudalism}