U.S. Copyright Office
Report on Legal Protection for Databases

Excerpts of the Executive Summary, August 1997

I. Copyright Protection for Databases in the United States

A. Before Feist

Databases, or "compilations," have been protected by copyright since 1790, when the first U.S. Copyright Act was enacted. As courts applied copyright law to compilations, two distinct rationales for protection emerged. One, known as "sweat of the brow," viewed the compiler’s effort and investment as the basis for copyright protection. The other, which drew on late 19th and early 20th century case law exploring the nature of authorship in works generally, viewed the creativity and judgment of the compiler in selecting and arranging materials as the basis for protection.

Regardless of the rationale, courts gave compilations a broad scope of protection from unauthorized copying. Any copying from a protected compilation was generally held to be an infringement. Courts required a would-be competitor to go to the original sources and compile the material independently. Many cases contain statements condemning the conduct of copiers who "reap where they have not sown," and essentially use copyright law as a means of policing unfair competition.

The 1976 Copyright Act for the first time incorporated a definition of "compilation" that requires original selection, coordination or arrangement. Courts applying the new statute, however, remained divided in their treatment of compilations. Some circuits abandoned "sweat of the brow" in favor of a unified requirement of creative originality for all works. Others continued to apply "sweat of the brow" to compilations generally, or to directories specifically.

B. The Feist Decision

In 1991, the Supreme Court in Feist Publications v. Rural Telephone Service Co. rejected the "sweat of the brow" doctrine, holding that creative originality was required by the constitutional provision empowering Congress to enact copyright laws. To be copyrightable, a compilation must evince a modicum of creativity in its selection, coordination or arrangement. The Court held that the work at issue, a white pages telephone directory, was uncopyrightable because it lacked even this modicum of creativity. The Court also made clear that the scope of protection for compilations is "thin" because it covers only the original elements of a compilation's selection, coordination or arrangement.

C. Subsequent Judicial Interpretation of Feist

Most cases subsequent to Feist have found the compilations at issue to be copyrightable. The scope of protection afforded, however, has been narrow. A number of post-Feist cases have held that substantial takings from copyrightable compilations did not constitute infringement, either because the defendant’s compilation differed in more than a trivial degree from the plaintiff’s, or because any elements of selection, coordination or arrangement that were copied were found not to constitute creative authorship.

II. Database Industry Practices

Database producers, motivated by the limitations in the coverage of copyright law, have adopted three main strategies to protect against unauthorized use of their products: (1) enhancing copyright protection by altering the structure or content of their databases to incorporate greater creativity; (2) increasing reliance on contracts; and (3) employing technological safeguards to prevent unauthorized access and use.

Enhancing copyright protection may entail either adding copyrightable text or altering the selection and arrangement of the database to make it more creative, while contracts are a major source of protection for database producers.

Technological safeguards, while offering great promise to producers in supplementing legal protection for electronic databases, are still in the early stages of development and not yet in common use.

III. Copyright Office Registration Practices

The Copyright Office, as the government agency responsible for registering copyright claims in works of authorship, makes determinations of copyrightability on a daily basis, including for databases. In examining an application to register a database, the Copyright Office must determine whether it is a copyrightable "compilation" as defined in the Copyright Act.

The 1991 Feist decision gave the Copyright Office clear authority to reject works for which protection was claimed solely on the basis of "sweat of the brow." The Office revisited its examining practices for compilations and issued new guidelines in 1991. It concluded that most compilations would meet the standard of originality required by Feist. The guidelines did, however, require extra scrutiny for certain categories of compilations, including telephone directories. The Office continues to register most claims submitted for compilations.

IV. The International Context

A. International Treaties

Databases have also been the subject of attention in the international arena. Their status as copyrightable subject matter is guaranteed by the two major multilateral treaties relating to copyright: the Berne Convention for the Protection of Literary and Artistic Works and the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPs). The TRIPs Agreement requires World Trade Organization member countries to protect "compilations of data or other material, whether in machine readable or other form, which by reason of the selection or arrangement of their contents constitute intellectual creations." A similar obligation is contained in the World Intellectual Property Organization (WIPO) Copyright Treaty concluded in Geneva in December 1996, which has not yet become effective.

B. European Database Directive

In March 1996, the European Union adopted a directive to the member states on the legal protection of databases. The directive must be implemented by member states in their national legislation by January 1, 1998.

The database directive covers compilations of data in any form, including hard-copy as well as electronic databases. It establishes a dual structure for database protection. One component is copyright protection for the structure of the database. The other is a new "sui generis" property right in the contents of the database.

The copyright portion of the directive seeks to harmonize the scope of copyright protection for databases throughout the European Union. It does so in two major respects. First, it sets a uniform standard of originality, protecting databases that, by reason of the selection or arrangement of their contents, constitute an author’s own intellectual creation. This is a higher standard of originality than is currently employed in the United Kingdom, Ireland and the Netherlands, but lower than is generally found in Continental "author’s right" systems.

Second, the directive establishes a uniform list of rights and exceptions to rights. These rights are reproduction, adaptation, distribution, and communication, display or performance to the public. Lawful users are permitted to engage in acts necessary for purposes of access to and normal use of a database’s contents. Member states may also provide certain exceptions in their national laws for private copying, illustration for teaching or scientific research, use for public security or administrative or judicial procedures, or other traditional copyright exceptions. The exceptions may not, however, "unreasonably prejudice the rightholder’s legitimate interests or conflict with normal exploitation of the database."

The sui generis right under the directive protects against unauthorized extraction or re-utilization of all or a substantial part of the contents of a database. It is similar to the Nordic countries’ "catalogue rule," which provides a short term of protection against unauthorized reproduction of catalogues of information. Insubstantial parts are excluded from protection. The directive contains a list of exceptions that member states may adopt, similar to those permitted for copyright, and a general prohibition against users engaging in acts that "conflict with normal exploitation of the database or unreasonably prejudice the legitimate interests of the maker of the database." The term of protection is 15 years. Any qualitatively or quantitatively "substantial change," including one resulting from an accumulation of small changes, "which would result in the database being considered to be a substantial new investment," qualifies the resulting database for its own 15-year term of protection. The right is available only to database makers who are EU nationals or habitual residents. The EU has indicated that it will agree to extend protection to nationals of third countries only where the third country offers comparable protection to EU databases.

An early version of the directive included a compulsory license for sole source databases. As part of an overall compromise, this provision was deleted. The directive notes the important role of competition policy in the database area, and establishes a review procedure to determine the effect of the sui generis right on free competition.

One member state has already enacted implementing legislation; most others expect to meet the January 1, 1998 deadline.

C. Proposed WIPO Treaty

In 1996, proposals for a treaty on the protection of databases were submitted to the World Intellectual Property Organization (WIPO) by the EU and the United States. A draft treaty based on these proposals was prepared for consideration at the WIPO Diplomatic Conference held in Geneva in December.

The draft treaty proved controversial within the United States. It was never reached or discussed in substance at the December 1996 Diplomatic Conference. The general subject of database protection will be discussed at an informational meeting at WIPO in September 1997. The draft treaty itself is not scheduled to be the topic of debate.

V. Prior Congressional Consideration

In May 1996, Congressman Carlos Moorhead, then Chairman of the House Subcommittee on Courts and Intellectual Property, introduced H.R.3531, the "Database Investment and Intellectual Property Antipiracy Act of 1996." No hearings were held on the bill, and no corresponding bill was introduced in the Senate.

Similar concerns were expressed about H.R.3531 to those expressed about the proposed WIPO treaty. Many of the same groups urged that all interested parties be given an opportunity to provide input and that a thorough analysis of the issues be undertaken.

VI. Copyright Office Meetings

In order to obtain a better understanding of the issues, and to provide Congress with complete and balanced information, the Copyright Office scheduled meetings with a broad spectrum of interests. From March through June 1997, the Office held 16 meetings with groups or individuals with an interest in the subject of database protection, including representatives of the library community, science agencies and organizations, educational groups and database producers. The meetings were led by Marybeth Peters, Register of Copyrights, with the assistance of the Office of Policy and International Affairs. The Copyright Office made clear that it was working from a clean slate, rather than assuming that any of last year’s proposals would be the starting point for congressional consideration.

The meetings indicated general agreement as to certain basic principles: (1) databases are vulnerable to copying, and adequate incentives are needed to ensure their continued creation; (2) individual facts should not be the subject of private ownership; (3) anyone should be free to obtain facts independently from original sources, even after they have been incorporated in a database; (4) government databases should not be protected; (5) it is important not to harm science, research, education and news reporting; and (6) "free riding" in the form of substantial copying for commercial, competitive purposes should not be permitted.

Some participants held strong views either in favor of new legislation or in opposition. In general, many members of the library and science communities, as well as some educational groups, telephone companies and Internet-related businesses, expressed opposition, while a majority of database producers advocated legislation. However, positions were not uniform within all of these groups. A large number of the participants in the meetings were undecided, or took neutral or intermediate positions.

VII. Issues

Most of the substance of the discussions at the Copyright Office meetings can be grouped under one of six topics: (1) as a threshold question (determining whether it is necessary to reach the other issues), whether additional legal protection for databases is needed; (2) what form it should take, whether a new form of property right or a tort concept closer to unfair competition; (3) how critical concepts such as "database," "substantial investment" and "substantial" or "insubstantial part" should be defined; (4) how it can be ensured that public interest uses of information are not harmed by new protection; (5) what should be the duration of any such protection; and (6) how "sole source" data should be handled. One additional issue would also need to be examined: what constitutional constraints may limit Congress’ ability to legislate in this area.

A. Need for Protection

All agree that the proponents of a new form of statutory protection have the burden of establishing the need for such protection. Views differ sharply, however, as to the type and degree of proof required to satisfy this burden. The options propose d range from a threat of future harm to empirical data generated through broad-scale studies.

Much of the discussion in the meetings focused on whether the combination of existing sources of protection – copyright law, contracts, state misappropriation doctrine, trade secrecy, trademark law and technological safeguards – is sufficient to provide incentives to produce a wide variety of databases.

Proponents assert that it is not sufficient, based on the following analysis: Since the Feist decision, copyright protection for databases has been limited, and in a number of cases database producers have been unable to obtain relief from the courts against substantial, competitive copying. Trade secrecy is largely unavailable for databases made available to the public or exploited commercially. Trademark law protects only "brand name" databases and only against uses that involve the trademark and confuse the consumer as to the origin of the database. Contracts, while an effective means of protection, are not binding on third parties. Remedies for breach of contract may not be satisfactory, and contract law may differ in different jurisdictions. In addition, the scope of enforceability of database contracts is not well settled and has generated controversy. Misappropriation is not a well-defined doctrine and varies from state to state. As set out in recent case law, the doctrine has various shortcomings from a producer’s perspective. Finally, technological measures can be a useful adjunct to legal protection for electronic databases, but they are still in early stages of development, are vulnerable to technological circumvention and do not prevent reuse of a database once it is available in accessible form.

Opponents believe that current copyright law is adequate and appropriate, as supplemented by other existing forms of protection. Some stress the fact that few databases have been held unprotectible, and view the scope of protection provided by the courts to be appropriate in light of the public interest in access to information. They see contract protection as highly effective, question the seriousness of the privity problem given the importance of timely updates for many databases, and note that Congress can consider the issue anew if the law on enforceability of contract terms develops in a different direction. They view the misappropriation doctrine as a strong means of protection, targeting the type of conduct most likely to cause meaningful commercial harm, while avoiding an impact on beneficial, public interest types of uses. They urge that Congress not step in prematurely, since the courts are so far interpreting that doctrine reasonably. Finally, they assert that technological protection has the potential to be extremely effective, easier and more economical to rely on than legal rights, and could make additional legal protection unnecessary as a practical matter.

One final aspect of the asserted need for new legislation relates to the international context. Proponents point to the recent EU database directive, which effectively conditions protection for non-EU databases on the availability of comparable protection in the given database’s country of origin. They assert that as of 1998, when the directive’s requirements take effect, their lack of sui generis protection will leave American database producers at a competitive disadvantage in Europe, one of their biggest markets. Opponents state that the United States should not follow Europe but should take the lead in establishing appropriate intellectual property policy, particularly in areas relating to the use of government data, where the U.S. approach has historically differed from that of many European nations. They assert that implementation of the database directive will leave U.S. producers no worse off in Europe than they are today and question predictions of a meaningful competitive disadvantage.

B. Form of Any New Protection

Two basic models have been proposed for the form of any new protection: (1) an exclusive property right; or (2) some form of unfair competition law, focusing on the nature of the conduct prohibited.

The choice between the two models has many ramifications. Depending on how it is drafted, an unfair competition model could obviate the need for definitions, for exceptions or for a defined term of protection. As to international consequences, an approach that differed significantly from the model of the European directive might not trigger reciprocal protection. Finally, the two models may have differing constitutional implications. Much would turn, however, on the precise delineation of either approach – how the scope of the rights is defined or what conduct is proscribed.

Proponents prefer the property rights model, in part because of the predictability and transferability of rights and the potential for serious damage to the market for a database even without the elements of competition or profit.

Among the opposing groups, and some neutral groups with specific concerns, there was a strong preference for the unfair competition model. A few participants sought as much specificity as possible and wished to avoid the uncertainty inherent in a general prohibition against conduct determined to be unfair.

C. Definitions

During the meetings, there was extensive discussion of the definitions used in the draft WIPO treaty and in last year’s bill, particularly the definitions of "database," "substantial investment" and "substantial part," or its converse, "insubstantial part." In several of the meetings, concern was expressed about the vagueness of these terms and about the possible impact on news gathering and educational activities if protection is provided against the accumulation of a number of insubstantial parts. A few participants in the meetings suggested that some or all of the definitional questions could be avoided if an unfair competition model were chosen rather than a property rights model.

D. Public Interest Uses

Many participants identified certain activities with public interest elements that they urged should be allowed to continue without new restrictions on the ability to use data or new costs in doing so – primarily scientific, research and educational activities and news reporting.

Analytically, there are various ways in which this concern could be addressed. One possibility relates to the form of protection chosen; depending on how it is articulated, a statute based on unfair competition is likely not to cover many such activities. If a property rights model is chosen instead, the scope of the rights granted could be drafted so as to exclude such activities as appropriate.

The exclusion from protection of insubstantial portions of a database helps but does not fully resolve the problem. In some circumstances users such as scientists or reporters need to extract substantial portions or all of a database in order to analyze its contents and draw conclusions.

Another possibility would be to provide an explicit exception or exceptions to cover those activities that Congress decides should be permitted without the need to obtain authorization. This could be accomplished through a broad, general exception similar to the fair use defense in copyright law.

Finally, some have suggested the possibility of compulsory licenses for certain socially favored types of uses in order to ensure the availability of data at a reasonable price. This leads to the question of whether the marketplace can appropriately handle non-profit scientific and educational uses. Some databases are produced specifically for this market; others have both commercial and non-profit uses. Many database producers today engage in differential pricing. It is unclear whether enacting a new form of protection would alter this practice or tend to raise prices.

E. Duration

In theory, protection should last just long enough to provide adequate incentives by allowing a fair return on investment. The difficulty lies in determining how long that period is, devising a term that works across the board to encourage the production of all types of databases. The more difficult question is the extent to which changes made in a database, for example in the process of updating or verifying its contents, should result in a new term of protection. Proponents of legislation argue that the same policy justifications for providing incentives to create databases also apply with respect to incentives to keep existing databases comprehensive, timely and accurate. Concerns have been expressed, however, that the result could be perpetual protection; as long as a database continued to be updated, new terms of protection could attach ad infinitum.

F. Sole Source Data

In theory, the answer to many of the concerns that have been expressed about restricting the availability of data is that, regardless of what model of protection is chosen, the database producer would not own the data in itself. Anyone would remain free to obtain all of the same data from other sources.

Nevertheless, there are circumstances in which this answer alone may be unsatisfactory. When the data is not available elsewhere, the ability to prevent its extraction from the database may in effect amount to ownership of the data itself. The two prototypical examples of "sole source" data are (1) government data provided to a private producer on an exclusive basis; and (2) data generated by the database maker itself, such as telephone subscriber information. Unless the producer chooses to make such data freely available, it will not be possible for anyone else to obtain it independently.

A variety of mechanisms have been proposed to deal with sole source databases. Broadly categorized, they are exclusions from protection; compulsory licenses; and regulation through other bodies of law such as antitrust or industry-specific government oversight. A combination of these approaches could also be considered, allowing greater fine-tuning to the nature of the database and its market.

Finally, arguments have been made for special treatment of databases which are not literally sole sources, but may be the only economically feasible sources of particular data. This may be the case where the data requires substantial time and effort to obtain, or the database has a narrow niche market and no other producer has the resources or ability meaningfully to compete with a first comer. Such databases appear to present somewhat different policy questions than literally sole source databases. It is unclear whether granting new legal protection would change the nature of existing markets, either exacerbating a lack of competition or encouraging more.

G. Constitutionality

Two primary constitutional issues also require consideration: (1) possible constraints imposed on congressional power to legislate in this area by the language of the Copyright Clause and (2) First Amendment limitations.

1. Copyright Clause

The Copyright Clause imposes certain restrictions on Congress’ ability to enact copyright legislation. Copyrights cannot be of indefinite duration, but can only be granted "for limited times." In addition, the Supreme Court in Feist held that Congress could not constitutionally provide copyright protection based on "sweat of the brow," but could only protect works of authorship embodying a modicum of creativity. The questions are then whether Congress can provide protection for "sweat" or investment without creativity under a different Article I power, most likely under the Commerce Clause, and whether any such protection must incorporate a limited term. The answers to these questions may depend on the form of protection that is chosen. If database legislation appears to be the equivalent of copyright under another name, but providing protection to uncopyrightable subject matter for unlimited times, the use of a different label and the recitation of a different constitutional basis will not alone be sufficient to save it. To the extent that the legislation promotes different policies from copyright, and does so in a different manner, it is similar to trademark law, and therefore seems likely to survive a constitutional challenge. The more the statute differs from copyright, the more likely it is to be constitutional.

2. First Amendment

The First Amendment must also be kept in mind in considering any new database protection legislation that may restrict the communication of facts. The courts have held that copyright law accommodates First Amendment values through the idea/expression dichotomy and the fair use doctrine. Depending on the model chosen and the formulation adopted for any database legislation, an explicit statutory provision clarifying that individual facts are not protected might assist in resolving First Amendment concerns. If individual facts remain free to be used for purposes of expression, there may be little need to copy a substantial portion of an entire database. To the extent that making a statement requires the use of more than a few facts, the form of protection and the nature and scope of the statutory exceptions would be highly relevant.

Note: This report and much other valuable background material on intellectual property issues may be found at

Printed copies of the report are available from the Government Printing Office at 202/512-1800.

At the request of Senator Orrin Hatch, Chairman of the Senate Committee on the Judiciary, the Copyright Office has prepared a report on legal protection for databases. The report does not make recommendations on either the advisability or the form of any database protection legislation; rather, it presents the issues to be addressed and some options for dealing with specific concerns.