captures much of the value in "organizing" data that can be lost to free-riders. Organizing that involves selection or judgment is protectable under copyright law, even after the decision in Feist. Therefore, inclusion of that term here is not necessary to provide an incentive for such activities. On the other hand, merely mounting a database on a server is part of maintaining it, but mounting data for access does not suffer from the free-riding problem of collecting (i.e., it is an expense that must be borne by the misappropriator as well as the original publisher). For all these reasons, we think it, necessary to protect only "collecting."

b. Concern for de facto "perpetual protection"

We also believe that there is a potential "perpetual protection" problem that is more complicated. This problem is rooted in the need to provide some type of protection for revisions of databases. Legislation that provided protection to new databases but not to revisions of databases, would skew investment. There would be a disincentive to revise proven, useful databases in favor of creating new databases. Reassembling (largely) the same information in a new database would be inefficient not only for data gatherers, but for data users who-in order to use the most current data-would have to accustom themselves to the format of the new database. Therefore, any database protection legislation should offer protection for revisions of existing databases, so that new iterations of a protected database are themselves protected. But this means that eventually there may be unprotected data entries (from iterations of the database older than 15 years) intermingled with protected data entries (from more recent iterations).25

This gives rise to a potential problem. In the classic case of a copyrighted book, the text loses protection at the end of its term, although new, revised versions of the text may enjoy fresh periods of protection. This means that one can find unprotected texts of The Raven or Leaves of Grass in libraries all over the country. At the same time, new versions of these books can be under some copyright protection (including new introductions, abridgements, "notes," artwork, etc.). It is possible to compare the two versions-old, unprotected and new, protected-side-by-side.

In the digital, on-line environment, content producers may choose not to sell copies of their works; access to a database may instead be licensed to users. The advantage is that the database user can receive the most current version of the compilation. The disadvantage is that the user may lack access to an old version of the database to compare old and new entries. The question is, how can a user, accessing only the newest version of a database that has gone through many iterations, distinguish unprotected data entries from protected data entries? In Appendix A we give a simple example of how this problem would arise.

While the Administration believes that the new language of section 1408(c) helps ensure that the bill provides no de jure perpetual protection, we remain concerned that the digital environment could produce de facto perpetual protection because users would be unable to distinguish protected and unprotected data and, therefore, would be chilled in their use of unprotected data.26 Such inadvertent extension of the protection afforded by H.R. 354 could exacerbate other concerns, including the "sole source” issue and the constitutionality of the law.

25 The legislative history for H.R. 2652 in the last Congress also bore on this issue, stating: "[N]o action can be maintained more than fifteen years after the investment of resources that qualified that portion of the collection of information that is extracted or used. This language means that new investments in an existing collection, if they are substantial enough to be worthy of protection, will themselves be able to be protected, ensuring that producers have the incentive to make such investment in expanding and refreshing their collections. At the same time, however, protection cannot be perpetual; the substantial investment that is protected under the Act cannot be protected for more than fifteen years. By focusing on that investment that made the particular portion of the collection that has been extracted or eligible for protection, the provision avoids providing on-going protection to the entire collection every time there is an additional substantial investment in its scope or maintenance." (Legislative Report)

26 At the same time, we believe that this potential problem arises with particular kinds of databases. Some databases are revised extensively and constantly; for these databases, the value of the database is much shorter than 10 or 15 years. Stock exchange price listings are the most extreme example, but other lists-realtors' sale listings and used car valuations also fall in this category. Other databases will be revised rarely once a definitive version is completed, i.e. a database of Union warships in the Civil War or the passengers on the Mayflower. The databases for which the "perpetual protection" problem arises are between these extremes: they are databases that have value over many years and require substantial, but not total, revision. Examples might include a historical database of the batting statistics of all baseball players in the major leagues or pharmaceutical or toxicological databases used in the medical professions.

There have been varied proposals to address this problem. One proposal has been to "tag" data entries so that older, unprotected data can be distinguished from protected data. We are not in a position to comment on the feasibility, whether technological or economical, of this suggestion. Another proposal-which is set out in the Senate discussion draft-would be the establishment of a deposit system to ensure that older, unprotected versions of the databases would be available to the public. We believe that the storage demands of such a deposit system would exceed anything the Copyright Office or the Patent and Trademark Office now handles. It is also not clear how the costs of such a deposit system should be apportioned.

At different junctures in this statement, we have recommended establishing express statutory defenses to remedy possible problems in a database protection; we make the same type of suggestion here. Where the database that is the subject of a litigation is the descendant of a now unprotected database and has substantial elements in common with that unprotected database, the defendant should be able to raise, as a defense, that the most recent unprotected iteration of the database is not reasonably publicly available.

In other words, if Smith Industries has been issuing the "Smith Industrial Database" annually since 1980, and then in 1999 if Smith Industries sues someone for unauthorized distribution of the "1999 Smith Industrial Database" the defendant can raise as a defense that the 1983 Smith Industrial Database is no longer reasonably publicly available. If the 1983 database is reasonably publicly available, there is no such defense.

The virtue in this approach in comparison to mandatory "tagging" or deposit systems is that it allows each private enterprise to determine how to make its now unprotected database available to the public. Moreover, the database producer does not have to make this final decision until the term of protection is over. Some concern has been expressed about this proposal by database producers who produce continuously updated databases; their situation in relation to this proposed defense merits examination. But, as we said above, we propose the defense when the protected database "is the descendant of a now unprotected database and has substantial elements in common with that unprotected database." We believe that for many continuously updated databases, the most recent database would have almost no elements in common with their 15-year ancestor.27

5. The "sole source" problem

There has been much discussion of what is called the sole-source problem: that many markets for data will be supplied by only one database provider. The solesource problem arises most acutely when one entity controls access to a unique, unreplicable collection of information, such as weather data that occurs once and cannot be replicated. This control may arise either purposefully, as with an exclusive contract with the data's original generator, or incidentally, when the data's original generator ceases to maintain it. Other practical sole-source situations can arise when an existing database operates as a natural monopoly; that is, it is possible, but not economically efficient, for someone else to build the dataset independently.

Even now, a sole-source may use contracts to preserve its market position against free riding by would-be competitors. Any form of database protection carries with it the possibility that it could further insulate a sole-source database provider against potential competition. Consequently, it will be important that any database protection legislation incorporate provisions that guard against the possibility that sole-source database providers will employ their new rights to the detriment of competition in related markets.

A partial answer to the sole-source problem is a savings clause such as the one in H.R. 354, providing that nothing in the bill operates to the detriment of federal antitrust law. Thus, for example, database owners would be as subject as any other economic actors to the application of the essential facilities doctrine, which prohibits owners of assets that are essential to the ability to compete in a market, and are not feasible to replicate, from refusing to deal with firms that need that access. On the other hand, this doctrine has been invoked relatively rarely, and understandably so: part of the incentive for the development of any valuable product or service is the hope that the product or service will be so attractive to consumers that it will

27 We recognize that this might still leave the problem of an old, but still protected iteration of the frequently refreshed database having a defense raised against it because the most recent unprotected database is not reasonably publicly available. At the same time, we are not convinced that producers of frequently refreshed databases cannot find means to ensure that at least intermittent, historic versions of their databases are reasonably publicly available.

become dominant. Regularly compelling access to valuable products and services could diminish their developers' incentives to invest in them in the first place.

At the same time, in markets such as data collection and dissemination, where natural-monopoly characteristics suggest that consumer choice among competing database products and services will not be common, some safety valve over and above the rarely used essential-facility doctrine may be necessary to ensure that database providers are not able to deny access to firms that require it in order to compete in downstream markets. Additional possibilities include the development of doctrines comparable to the misuse doctrine in patent and copyright law or, in extreme cases, the idea/expression merger doctrine in copyright law. 28

As with some other problems we have identified above, however, much of the concern arising from the sole-source problem can be eased by defining both the protected activity and the prohibited conduct narrowly. If the bill protects only data collection and generation, it will be covering value-adding conduct that enhances welfare, even though only one firm may find it worthwhile to engage in collecting and disseminating a particular type of database. Similarly, to the extent that the bill prohibits only distribution and extraction for the purpose of distribution, while conversely permitting transformative uses of data, it would leave data providers free to add value and enter markets that the original data collector's work alone was incapable of serving.

E. Fifth Principle-Balance protection with permitted uses

Any database misappropriation regime should provide exceptions analogous to fair use principles of copyright law; in particular, any effects on non-commercial research should be de minimis.

Given the difficulty of foreseeing how "substantiality," "extraction" and other legislative terms will play out in a complex and rapidly changing environment, we expressed concern last summer that H.R. 2652 lacked a balancing mechanism analogous to the fair use doctrine in copyright sufficient to address the wide range of circumstances in which information is aggregated, used, and reused. We were especially concerned that the section 1203(d) exception for non-commercial research and educational uses did not ensure against disruption of legitimate non-commercial research, and that educational activities were not disrupted by the prohibition against commercial misappropriation. Last year, we also were concerned with equitable issues of access and use that may be especially important in markets exclusively served by a single data producer.

In reviewing the permitted acts provisions of H.R. 354 (section 1403), we would like to suggest, as an initial matter, that the Subcommittee rearrange the various "permitted acts" to move more clearly from absolutely shielded activities for all persons (such as use of insubstantial parts (1403(b)) to the more limited shields on activities set out in 1403(a)(2). We propose that the Subcommittee reorder section 1403 as shown in Appendix B. We believe that this reordering would provide legislation that is easier to understand and a clearer platform for full discussions on whether the permitted activities adequately address policy and constitutional concerns. This proposed reordering is separate from any substantive recommendations. As to the substantive elements of the permitted acts section, the Administration is pleased that H.R. 354 limits the liability for nonprofit educational, scientific, and research purposes to uses that harm directly the actual market and that the legislation now includes as section 1403(a)(2)(A) provisions for "additional reasonable uses" similar to the fair use provisions of section 107 of the Copyright Act. However, we are concerned that the last sentence in section 1403(a)(2)(A) could be interpreted as overriding the criteria in section 1403(a)(2)(A) with a standard that differs in form but not in substance from the basic operating provisions of section 1402. The Administration would not agree with any intent to override a "fair use"-like balancing test; on the other hand, if the last sentence of 1403(a)(2)(A) is intended only to restate the basic prohibition without disturbing the balancing test, it is extraneous language. We therefore recommend its deletion.

We recognize the desire to avoid the precise fair use terminology of the Copyright Act in order to make clear that the legislation is grounded in misappropriation rather than intellectual property. However, in the interests of transparency and predictability, we believe that the fair use principles of copyright are a sound platform on which to build. Providing the safeguard of familiar fair use criteria can help mini

28 Further, a requirement that database providers notify users of their intent to assert rights against misappropriation can mitigate against the possibility of some sole-source situations ever developing; if users are on notice that they may be liable for their conduct involving data from a particular database, they may have reason to seek out alternative sources of the data, so that they will not be locked into a single, dominant source down the road.

mize any unintended consequences of the untested basic operating provisions of section 1402. We believe that this would give courts the tools they need to do justice in particular situations.

The fair use factors may need to be framed or supplemented to allow courts to take into account that the subject matter is industrious collection rather than original expression, that the protected interest is purely economic, and that the proscribed behavior is a form of unfair competition. The provision would also have to be recrafted to focus on distribution, rather than use, if the basic prohibition were amended as we have suggested above. Courts might also be called on to recognize the unique conditions of some database markets. But we believe that the vast experience of courts in using the judicially-crafted principles of fair use should be built into database protection legislation. It is worth noting that in the 23 years since Congress codified the fair use factors, it has neither narrowed nor expanded these factors. While it may be appropriate to diverge from copyright fair use in creating the permitted uses regime for database legislation, the differences between the two should be clearly understood and recognized by concerned parties.

Finally, we would reiterate a point made earlier: the scope of the basic prohibition will determine the weight that the permitted uses section must bear in judging both the policy and constitutionality of any database protection legislation.

F. Sixth Principle-Ensure protection for U.S. companies abroad and promote harmonization

Consistent with the goals of the World Trade Organization (WTO) and U.S. trade policy, legislation should aim to ensure that U.S. companies enjoy available protection for their database products in other countries on the same terms as enjoyed by nationals of those countries.

There has been some discussion in the United States about the effects of the European Union's 1996 Database Directive (EU Directive) on American database producers. The EU Directive requires European Union Member States to provide sui generis protection for databases, but denies this protection to nationals of any foreign country unless that country offers "comparable protection to databases produced" by EU nationals.29

The Administration opposes such "reciprocity" requirements, both domestically and internationally. We believe that commercial laws (including intellectual property and unfair business practices laws) should be administered on national treatment terms, that is, a country's domestic laws should treat a foreign national like one of the country's citizens. This principle is embodied in Article 3 of the Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPS Agreement) as well as more generally in the Paris Convention for the Protection of Industrial Property and the Berne Convention for the Protection of Literary and Artistic Works.

The Administration believes that Congress should craft U.S. database protection legislation to meet the needs of the American economy. A database protection law properly balanced for the robust digital economy of the United States will serve as a model for other countries that hope to build businesses, employment, and economic activity in the new millennium.

At the same time, we believe that a law along the lines of H.R. 354 (with proper attention to the concerns we have identified) will amply provide protection "comparable" to that provided by national laws implementing the EU Directive. From the perspective of a private database producer, a misappropriation law as discussed in both the last and current Congress would, we believe, provide a cause of action and meaningful remedies in the same range of situations in which the laws implementing the EU Directive provide a cause of action and meaningful remedies.30 Although we believe that a law along the lines of H.R. 354 would provide American database makers with protection under the EU Directive's reciprocity provision, the Administration would, for the reasons stated above, oppose any effort to put automatic reciprocity provisions into American law in this area. United States Trade

29 This is established in Recital 56 of the EU Directive. Recital 56 also provides that a foreign national will enjoy database protection when those "persons have their habitual residence in the territory of the Community." This may provide protection to American database producers who have substantial business operations in EU Member States. Pursuant to Article 11/3 of the EU Directive, a determination whether a foreign state offers "comparable" protection must be made by the European Council based on recommendations from the European Commission.

30 The EU Directive is not a national law. It "directs" the Member States of the EU to implement a legal framework. H.R. 354 would have to be compared, for example, to German, Dutch, and/or Italian law to make the proper comparison of national law to national law. Such a comparison is well beyond the scope of this statement.

Representative Charlene Barshefsky cited the reciprocity provision of the EU Directive as a subject of concern in announcing the Administration's 1998 Special 301 Review.

While we believe that a United States database protection law should adhere to a national treatment model, the Administration would support an appropriately crafted provision that would allow the President to affirmatively deny database protection to foreign nationals on the appropriate finding by Executive Branch agencies such as the USTR and/or the Department of Commerce. This could, for example, be achieved by statutory language or legislative history making database protection for foreign nationals subject to USTR's Special 301 process.

G. Additional Issues

1. Gradations of Criminal Liability

While we agree with Chairman Coble's decision to shield non-profit researchers and educators from any criminal liability under section 1407, we believe that the existing criminal provisions should be further refined, particularly by drawing a distinction between misdemeanor and felony conduct and requiring minimum amounts of damage under each. This will expand the range of charging options available to prosecutors. We have attached our recommendation for statutory language as Appendix C.

2. Data-Gathering Activities of Law Enforcement Agencies

We believe it is important to make clear that the legitimate data-gathering activities of law enforcement and intelligence agencies will not be affected by the bill. While we believe that intelligence gathering and national security activities are already shielded from liability by section 1402 in that these activities will not cause "harm to the actual or potential" market of the product, we propose an additional statutory provision and legislative history as shown in Appendix D to confirm that these activities fall outside the bill's reach.

3. Administration Study

The Administration believes that, given our limited understanding of the future digital environment and the evolving markets for information, it would be desirable to conduct an interagency review of the law's impact at periodic intervals following implementation of the law. Such a government study might be conducted jointly by the Department of Commerce, the Office of Science and Technology Policy, and the Department of Justice in consultation with the Register of Copyrights and other parties. We believe that such a study should not be limited to any one set of issues or concepts; rather, it should explore issues including: database pricing before and after enactment of the law; database development before and after enactment of the law; international protection for American database producers; the impact of the law on scientific research and education; access issues; and "sole source" databases.

I thank the Subcommittee for the opportunity to appear before you today and look forward to working with you during the legislative process. I would be pleased to answer any questions that you may have at this time.

APPENDIX A

Imagine that in 2000, a database producer makes a database; we will designate the first twelve entries alphabetically:

In 2003, it "expands and refreshes" the database, so that the first fifteen entries are as follows: