Jason Chicola

Dawn Farber

Mami Karatsu

Joseph Liu

Karl Richter

John Tilly

Massachusetts Institute of Technology

Harvard Law School

December 10, 1998

Executive Summary

Paper

I. Introduction and Scope *

II. The Copyright Balance and Pressures *

A. The Existing Balance *

B. The Balance Upset *

C. Legal Attempts to Restore the Balance *

III. The Promise and Threat of Trusted Systems *

A. Trusted Systems Architectures *

1. How Do Trusted Systems Work? *

2. Major Developers of Trusted Systems *

3. Potential Problems *

B. Copyright Economics and Trusted Systems *

1. Intellectual Products as a Public Good *

2. Optimal Provision of Public Goods *

3. Optimal Provision, Current Legislation, and Trusted Systems *

C. Trusted Systems and Price Discrimination *

1. The Promise of Price Discrimination *

2. The Threat of Price Discrimination *

3. Possible Responses *

IV. Values and Architectures *

A. Fair Use *

1. Value to be Protected *

2. Likely Impact of Trusted Systems *

3. Possible Responses *

4. Tradeoffs *

B. First Sale *

1. Value to be Protected *

2. Impact of Internet Technology *

3. Likely Impact of Trusted Systems *

4. Possible Responses *

5. Conclusion *

C. Limited Duration *

1. Value to be Protected *

2. Likely Impact of Trusted Systems *

3. Possible Responses *

D. Subject Matter Coverage *

1. Value to be Protected *

2. Likely Impact of Trusted Systems *

3. Possible Responses *

E. Privacy *

1. Likely Impact of Trusted Systems *

2. Value to be Protected *

3. Proposals *

4. Recommendations *

F. Freedom of Contract *

1. Value to be Protected *

2. Likely Impact of Trusted Systems *

3. Possible Responses *

V. Conclusion *

A. Will a laissez faire approach work?…….………………………..………..…

Background

Until very recently the law has been able to balance the public and private interests surrounding intellectual property. Laws have evolved to achieve a delicate balance that promotes progress in science and the useful arts, ensures broad access to information, rewards creators with limited monopoly rights, and spurs innovation and sharing.

With the rise of the Internet and new technologies, however, this delicate balance has been suddenly disturbed. New technologies -- trusted systems -- are outweighing the law's ability to maintain a stable equilibrium. With its ever-increasing capabilities, technology is tipping the scale away from the public interest and towards private control. This fundamental shift will have many implications for the way that we innovate, share, and distribute knowledge. Soon, technology might tip the balance too far, and break the scale. It is conceivable (some might say inevitable) that the architecture of trusted systems - also called electronic copyright management systems - will replace the law in offering protection for intellectual property.

The technology of trusted systems allows computers to protect and distribute information in more secure and reliable ways. At its heart, a trusted system is a perfect access control system which uses technology as a lock. The technology of trusted systems brings many benefits. Trusted systems allow the bundle of property rights to be split -- such as the right to read, the right to prepare derivative works, the rights to copy, etc. -- and facilitate on-line payment schemes for each of these rights. However, along with the clear benefits, the technology of trusted systems enables exclusion, limitations on distribution, consumer tracking, and invasions of privacy.

An appropriate balance for the future must weigh the opportunities (and present capabilities) that trusted systems offer against the threats to the public goods of broad dissemination and incentives for innovation. As we develop proposals for the future of intellectual property, we need to guard that the dropping pan from the broken scale does not flatten the poor, the academics, and the critics, but that it serves to bring even greater knowledge to these special and vital members of our society. While we find social benefits in reporting, teaching, scholarship, parody, sharing, negotiating, and contracting, trusted architectures -- which represent a more thorough way to guard private intellectual property than law -- are increasingly threatening these social values.

We face the challenge of recommending the best form of future architectures for intellectual property to preserve these values in the future and return the scale to careful balance. A critical first step is to decide which aspects of the existing intellectual property law structure should be adapted to new technology of trusted systems, and which aspects of it are outmoded by trusted systems. In thInInI e recommendations we propose, we conclude that in most cases existing legal structure can be appropriately extended to Internet technology; but, that in some other cases the constructs of the real space law should not apply to cyberspace, and that the law should act to regulate trusted architecture.

Fair Use

We find social benefit in the "fair use" of information for criticism, comment, news, reporting, teaching, scholarship, research, and parody. Trusted systems can dictate whether or not users have access to content. That is, trusted systems enable copyright owners to choose readers. A trusted system with the power to discriminate limits criticism, comment, and parody. Copyright holders will probably not want to release protected works to those they suspect will criticize it. Trusted systems can also make copying and transferring material difficult and expensive. Unchecked, and without specific provisions built-in to protect fair use, the technology of trusted systems could eliminate fair use.

There are several potential and complementary solutions to protect fair use. The Digital Millennium Copyright Act White Paper suggests that for some fair use purposes it is legal to circumvent the technological protection of any work’s digital copy. However, in the event that technologies become so efficient as to make any unauthorized circumvention impossible, there is no legal support for fair use. One alternative is to use the law to require trusted systems to include certain provisions for fair use. For example, the government could require that all trusted systems allow free access to students, teachers, and scholars, and that any agreement that prohibits this fair use would be invalid. Tax treatment could be used to further encourage trusted systems to allow fair use, beyond minimal compliance with the law. The government could grant tax deductions to firms that significantly facilitate the fair use of their protected works.

Another architectural solution would be a fair use site, which could be maintained by either the government or a non-profit entity. If providers do not want to give the public fair use to their materials through their own site, they could provide their materials through an outside system, which scholars, teachers, and students could access with passwords. This fair use site could allow users to download materials until they used all the credits in an account that would be assigned to them. If users ran out of credit, they would need to explain their need for more credit before being allowed access to more protected material. In this system, users would be hesitant to share passwords because this would deplete their account and limit their ability to access information later. A digital library is another possible solution. Users could be allowed to "borrow" read-only materials for a limited time. After the time expired, the digital copy would be erased from their computers.

First Sale

Sharing is an important value underlying copyright law, but trusted systems could limit the sharing of information in cyberspace. In copyright law, the First Sale Doctrine limits the control that a copyright owner has over the public distribution of copyrighted work. While First Sale works well in real space, where there are physical obstacles to reproduction and distribution, it makes any transmission of data in cyberspace illegal by the letter of the law. When computers transmit data they make copies of the data which first sale prohibits. First sale, as a construct, is therefore not appropriate for extension into cyberspace.

A technical architecture that can preserve the values of sharing and fair use is more appropriate. There are many different architectures that could preserve First Sale values. For works of authorship, it might be appropriate for the trusted architecture to allow a certain degree of copying, by allowing for a certain number of copies to be made of each original work. The architecture could even allow for a certain number of copies to be made from an original, but prevent copies from being made from copies. Perhaps this architecture could even be complemented by another provision that would preclude the simultaneous use of more than one copy at a time, or which would only allow the copy to exist for a certain length of time before the trusted system denies access. Together, these provisions could come close to replicating the real world lending process that First Sale protects. The challenge to designing one of these trusted systems is ensuring that some barriers to lending remain. Otherwise, sharing in cyberspace could become so easy that it destroys the market for intellectual property.

Public Domain

Time limits on protection are another concern with trusted systems. Copyright has a limited duration of the life of the author plus 50 years or 75 years for work-for-hire. (Note: Legislation passed this year extends this to life of author plus 70 years and 95 years) The concern is that trusted systems have the potential to control access to information forever. If trusted systems were the only way to access the information, this could potentially keep expressive works and even knowledge from benefiting the public for an unreasonable length of time. The time limits in traditional intellectual property legal constructs are designed to offer incentives to create original works by offering limited-in-time rewards to innovators. Traditional copyright protection for authors is based on the authors' lifetimes and can be justified by natural property rights. The infinite protection period that trusted systems could enable would not measurably increase incentives to create; quite the opposite, infinite protection would harm society by restricting information.

One solution to this problem would be to require trusted systems to include in the architecture a tag to mark contents with an expiration date. Like a form of watermarking, this tag could release the content from protection after a certain time. This would be easy to extrapolate from existing copyright law for works-for-hire since duration is fixed, but more difficult for works of authorship since those are dependent on the life of the author. The expiration dates, therefore, should have a dynamic quality. Perhaps living authors or creators should be able to protect distribution rights up to a prescribed maximum duration, measured from the time of creation.

Privacy

Privacy is another value that trusted systems threaten. The Fourth Amendment -- which protects against unscrupulous search and seizure -- embodies the American belief in the importance of privacy. As with fair use, we need to carve into trusted system architecture provisions to ensure that we continue to protect privacy. In the specific context of access to copyrighted works, the ability to access works anonymously encourages wide-ranging consumption of intellectual material, without concerns and potential risks about centralized monitoring of access.

As a first step, providers should be required to disclose the extent to which their trusted systems protect privacy. This disclosure could be in the form of a tag that reveals how information provided by a user will be used by the provider, and, most importantly, whether the provider will grant third parties access to private information. We propose that the government require that all trusted systems protect privacy, and actively prosecute the owners of trusted systems that do not provide a minimum level of privacy protection. The government can set the standards for trusted systems by requiring companies that contract with the government to protect a greater level of privacy. In pressing criminal charges against firms for failing to adequately protect privacy, we need to guard that the government does not impede the development of potentially valuable technologies.

Another important approach to privacy protection is to let consumers actively protect their own privacy. To encourage this, the government should allow and support the use of anonymous e-cash, especially for purposes of accessing reading materials about political issues or controversial topics. There are some types of goods for which the ability to trace sales to individuals finds compelling support. However, for the domain of editorial or literary materials, the benefit from anonymous access to intellectual life outweighs the concerns about police enforcement, national defense, and terrorism that may have some validity in other domains.

Additionally, the government should play a part in informing consumers about the benefits of anonymous payment architectures. Presently there seem to be mounting trends that favor building this sort of privacy provision into trusted systems. The European Community, for example, may not allow its members to trade electronically with American companies unless the U.S. firms strengthen the privacy protection they offer to electronic consumers. As a global leader, the United States needs to strengthen its support for privacy protection and set a better example for the rest of the world.

Contracts

The right to contract is also challenged by trusted systems. The specific issue of concern regards adhesion contracts, or standard form contracts. While these sorts of contracts can reduce the costs of contract formation, the unilateral decisions favor the copyright owners, and are not subject to judicial oversight and may have negative implications for first sale and fair use. Trusted systems can require that users submit to terms of an agreement without bargaining in the exchange. Supporters of adhesion contracts point to their convenience and ease of use; opponents cite the long-held values at the base of traditional contract law: the freedoms to contract and negotiate.

Solutions to remedy these issues must at least include a clear notice requirement on all contracts. This is important in any contract, but especially so when there might be non-standard terms in a long adhesion contract. It would be appropriate for such non-standard terms to require an extra sign-off or confirmation by the user to ensure that the user is aware of their effect. New technologies, like DIVX for example, should clearly state the technological limitations they impose. Technical restrictions of the trusted system -- like a restriction on the ability to copy and paste or to print -- need to be disclosed and understood by users in advance. There are also some potential judicial remedies. The courts should invalidate certain technical limitations that trusted systems might enforce. By keeping an eye out for trusted systems that impose unreasonable terms, such as terms that conflict with fair use, the courts can be ready to ban these trusted systems.

Conclusion

While the law becomes less and less a means for intellectual property protection, it remains a crucial instrument to regulate the characteristics of the technical architectures which are replacing it. The law needs to set the requirements for trusted systems in order that the values of fair use, first sale, limited duration, contracting, and privacy are not quashed. To return a delicate balance to the public and private interests in intellectual property protection, we must ensure that provisions for important social values are incorporated into the technical architectures that control information. The law must prosecute and hold criminally liable any trusted architecture providers who fail to protect these values.

The technology of the internet is putting tremendous pressure on the preexisting balance of access and incentives struck by our existing intellectual property laws. Internet technology makes it easier than ever before to copy intellectual works and distribute them around the world at nearly zero cost. This presents the possibility that incentives for authors to create intellectual works may be greatly undermined by widespread copying. At the same time, however, the internet permits technological responses to the threat of widespread copying that have the potential for providing authors with even greater control over access to, and use of, their works than ever before. This presents the countervailing possibility that many of the rights enjoyed by users of intellectual works in the physical world may be more limited in the on-line world.

In this paper, we seek to address the implications of these technological developments in light of the values underlying existing intellectual property laws, and to propose new legal and technological architectures that may help deal with the challenges posed by new technology. Because the subject matter is potentially quite vast, it is important to define the precise scope of our project. We focus in this paper primarily upon works that fall within the scope of copyright law (and not patent or trademark), since these works (e.g. text, images, music, software, film) comprise the bulk of the intellectual property works currently available on the internet. In addition, our focus is also primarily on U.S. law. Although international issues are certainly quite important in this area, space constraints do not permit us to fully address these issues as thoroughly as they deserve. Some of the conclusions that we reach may, however, have implications for international issues as well.

Within the area of U.S. copyright law, we focus in particular upon the promise and threat posed by so-called "trusted systems," since such systems appear to be the leading response of copyright owners to the perceived threat of widespread copying. Numerous companies are currently developing trusted systems that enable copyright owners to exert ever-greater control over access to, and use of, their works. Moreover, recent legislation has been passed that promotes and gives added support to such trusted systems technology. Although such systems are not yet widely deployed, we believe that there is a good chance that they will be widely deployed, particularly given the strength of copyright owner concerns over copying. Indeed, such systems in many ways serve the legitimate interests that copyright owners have in preventing indiscriminate and widespread copying of their works.

At the same time, however, in providing an answer to copyright owner concerns about unchecked copying, trusted systems also raise the possibility that copyright owners will have too much control over the use of their works in the digital context. Given the above, it is not too early to consider the consequences of such systems and to begin to think about ways in which such systems, by effectively supplanting copyright law, may affect certain values that existing copyright law currently takes into account. Nothing about the technology itself guarantees that values underlying such doctrines as fair use, first sale, and the public domain will be adequately protected. We firmly believe that such values need to be taken into account when thinking about the proper regulatory and technological architecture for trusted systems. Where such values may be adversely affected by unchecked use of such systems, legal and technological responses may be necessary to preserve and protect these values.

Although a number of commentators have written about the implications of trusted systems technology, we believe that the make-up of our team presents us with an opportunity to combine both legal and technological perspectives in an effort to come up with creative ways of securing the benefits of trusted systems technology, while minimizing the potential threats posed by the technology to values underlying copyright law as a whole. Accordingly, in the proposals we advance, we seek to push the discussion beyond consideration of existing technological and legal structures and toward technological and legal possibilities that have not yet been actively considered. Although some of these proposals may seem unusual, we hope that they serve the purpose of opening up the discussion to consideration of the wider possibilities presented by technology and law. Furthermore, because the future course of technology is so difficult to predict, we do not try to fix upon a single optimal proposal, but instead seek to offer a menu of options, along with an assessment of the relative strengths and weaknesses of each.

We begin in Part II of this Paper with a brief survey of the existing copyright law balance, the pressures that internet technology places on that balance, and some of the legal responses that have been made to such pressures. In Part III, we discuss, in detail, the use of trusted systems technology as an independent response to the pressures of technology, focusing on both the promise and threat offered by the technology. In Part IV, we analyze a number of values (e.g. fair use, first sale, the public domain, privacy, and freedom of contract), and examine how, if at all, they may be affected by widespread deployment of trusted systems. For each of these values, we propose a number of potential legal and technological architectures, which would permit copyright owners to obtain the benefits of trusted systems while preserving a number of these underlying values. Finally, in Part V, we summarize our conclusions and briefly explore how such architectures might practically be implemented.

For many decades, our existing copyright laws have managed to set a certain balance, preserving some level of incentive for authors to engage in creative activity, while permitting a degree of broad access to the works by the public. The technology of the internet, however, is now disturbing that balance in a significant way. In this part, we quickly sketch out the nature of the existing balance, the way in which the internet is disturbing that balance, and a number of the legislative attempts that have been made to maintain that balance in the face of changing technology. This section provides the existing legal background against which trusted systems operate.

Copyright protection in the U.S. is granted in order "to promote the progress of science and useful arts." By granting authors certain limited rights over the use of their creative works, copyright law provides authors with incentives to engage in creative activity, resulting in the production of such works. Without such rights, incentives for the creation of works would be undermined, since others could simply copy the work without spending the time and effort invested by the original author, thereby driving the price of the work down to marginal cost and preventing the author from recouping his or her investment. Thus, in order to provide adequate incentives, copyright gives authors the exclusive right to make copies, publicly distribute copies, make derivative works, and publicly perform and display the copyrighted work.

At the same time, however, these rights are limited and do not encompass all uses of a copyrighted work, since giving authors too much protection would undercut the broad access to works that the copyright laws are designed to promote. Copyright law does not, for example, give authors a right to control reading of a copyrighted work. Nor does copyright law give authors a right to control the resale or transfer of a particular copy, once it has been sold. Copyright is also limited in time, expiring after a certain period. Finally, copyright also contains a doctrine called fair use, which permits certain types of copying for certain purposes, such as criticism, education, reporting, and scholarship. Whether a given use is "fair" depends on the particular circumstances, and will involve consideration of a number of factors, such as the impact of the use on the market for the original, the type of work, the type of use, and the amount of the original copied. Together, these limits ensure that copyright owners do not have too much control over the use of their works.

Copyright law thus represents a very basic balance. On the one hand, it seeks to provide some degree of incentive for the creation of works. On the other hand, it seeks to ensure broad dissemination of, and access to, such works. These two interests exist in some degree of tension. Providing greater protection gives authors more incentives to create works but gives users less access to the works that are created. Providing less protection gives authors fewer incentives to create works but gives users more access to the works that are produced. By providing copyright owners with certain rights, while at the same time ensuring that consumers are able to engage in certain uses, the copyright law effects a balance of access and incentives.

The law is not, however, the only factor that influences this balance of access and incentives. Another factor is technology. Much of the existing balance of access and incentives created by the law depends upon the state of existing technology. So, for example, although the first sale doctrine provides some level of free access to copies by permitting users to transfer copies to others, the particular balance of incentives and access depends upon the costs associated with transferring physical copies. If technology greatly reduces the cost of transfer, then this changes the preexisting balance. Similarly, changes in the cost and/or quality of copies may greatly affect the actual level of unauthorized copying. Photocopying a book is time-consuming and costly, and results in a vastly inferior copy, whereas copying a digital version of that book is quick, cheap, and perfect. Technology may also affect the ease of detecting copyright violations and enforcing copyrights.

Internet technology is rapidly changing the preexisting balance of copyright law. In some ways, Internet technology may actually increase the production and distribution of copyrighted works. The technology of the Internet drastically lowers the cost of publishing a work around the world. Whereas in the past, an author had to convince a publisher to invest in the work and distribute physical copies to retail outlets, now an author can potentially distribute the work worldwide by posting it on the Internet. Thus, existing authors can now reach their audiences more cheaply than ever. Moreover, cheap distribution greatly expands the universe of potential authors.

At the same time, however, Internet technology presents a potentially serious threat to copyright owner incentives. The Internet enables effectively costless copying and worldwide distribution. Whereas in the past, the costs associated with making physical copies and distributing them served as a limit on the amount of copying, now these limits are effectively removed. In addition, digital copies, unlike analog copies, are perfect substitutes. Again, in the past, the imperfection of copies had some impact in limiting the amount of copying, but in the digital context, that limit, too, is removed. Thus, absent any technological or legal response, we would expect the amount of unauthorized copying to increase significantly on the Internet.

A significant increase in the level of unauthorized copying might well result in a decrease in copyright owner incentives. If unauthorized copies of copyrighted works can be easily and costlessly distributed around the world, then this would have the effect of increasing the amount of unauthorized copying and reducing the amount authors could recoup from sales of authorized copies. Moreover, this potential impact would not be limited to copies issued in digital form, since even physical or analog copies can be easily and cheaply converted into digital form. Thus, the concern is that, without some kind of legal or technological response, the easy copying offered by the internet may result in a reduction in the number and types of copyrighted works created (though there appear to be few attempts thus far to quantify this reduction in any empirical way).

There have been a number of legal responses to this potential shift in the copyright law balance. The first response has been to devote more effort to the enforcement of existing copyright laws. Thus, for example, there have been a number of well-publicized cases in which content owners have filed copyright infringement suits against bulletin board and web-site operators. In other cases, copyright owners have filed suit against the distributors of certain technologies and devices, which facilitate the copying or playing of digital works. In still other cases, attempts have been made to use the criminal provisions of the copyright act to further deter copyright violations. These cases have very likely had some impact in reducing the amount of copyright infringement on-line, particularly large-scale infringement that might attract the attention of copyright owners. At the same time, smaller infringements remain difficult to detect and costs of enforcement may have been increased by the Internet.

In 1995, the Clinton Administration issued a White Paper on intellectual property rights on the Internet, in which it recommended changes to the copyright law in response to the perceived challenges posed by the Internet. A number of these recommendations were eventually incorporated into a treaty of the World International Property Organization (WIPO). These recommendations were then subsequently implemented in statutory form here in the U.S. in the form of the Digital Millenium Copyright Act (DMCA), which was signed into law by President Clinton several weeks ago.

The DMCA contains a number of provisions that are new to copyright law and that are designed to facilitate the deployment of, and lend additional legal support to, technological means of protecting copyrighted materials. First, it makes it an infringing act to circumvent technological measures designed to restrict access to copyrighted works. Thus, the very act of circumventing the copy-protection gives rise to a claim (even if there is no copying). Second, the DMCA bans the manufacture, import, or distribution of technologies that are primarily designed for the purpose of circumventing technological protection measures and only have limited uses other than circumvention. Third, the DMCA also makes it an infringing act to intentionally remove or alter "copyright management information," information attached to a copyrighted work that identifies the identity of the copyright owner and includes terms of use, etc. Those who violate the provisions of the DMCA are subject to suit for actual or statutory damages, as well as potentially subject to criminal liability.

The provisions of the DMCA regarding anti-circumvention do not go immediately into effect. Instead, there will be a two-year period before such provisions go into effect, during which the Copyright Office is charged with examining whether such provisions might have an adverse effect on the ability of individuals to make non-infringing uses of the work. If the Copyright Office determines that the provisions will in fact have such an effect, it can then exempt such groups and/or uses from these provisions. In addition, the DMCA contains numerous additional exemptions for specific types of activities and groups. So, for example, the DMCA exempts certain nonprofit libraries, archives, and educational institutions, permitting them to circumvent technological protection mechanisms for certain limited purposes. Specific exemptions also exist for law enforcement, reverse engineering of software for purposes of interoperability, encryption research, and security testing. The DMCA does not, however, contain any broad provision permitting circumvention for fair use purposes.

The DMCA thus sets the stage for wider deployment of systems designed to limit, through technology, access to and use of copyrighted materials. It does this by providing additional support to technological protection mechanisms through creation of an independent cause of action for acts of circumvention. The DMCA plainly envisions that such technology will provide copyright owners the means to protect their works in the on-line environment. The remaining parts of this paper will consider what types of technology are already in existence or will likely be developed, and will examine whether and to what extent legal responses, perhaps through the rulemaking procedure permitted by the DMCA or through amendments to the DMCA itself, may be necessary to maintain some of the values underlying copyright law more generally.

In this Part, we take a close look at the technology and economics of trusted systems. Currently, many companies are developing trusted systems architectures that will permit copyright owners not only to prevent unauthorized copying of their works, but also to regulate the use of such works to ever finer degrees. We describe these architectures, examine the extent to which they are currently deployed, and make estimates about the future directions of the technology, concluding that there is a good chance that such systems will be widely deployed in the future. We also analyze the economic model underlying the use of trusted systems, a model based on the idea of metered use rather than the sale of copies. Although widespread deployment of trusted systems may succeed in permitting copyright owners to capture more and more of the value of their works through metered use, the metered use model has a number of drawbacks and may not fully protect the values underlying our existing copyright law balance.

The Internet is the world’’s biggest copy machine, but only for certain goods. The most valuable forms of copyrighted goods are unavailable in on the Internet. Producers of these goods are rightfully worried that they will fail to receive compensation for their goods if they were to put them on the Internet. Mark Stefik, explains the situation well when he states, "[t]he root of the problem is that authors and publishers cannot make a living giving away their work. It now takes only a few keystrokes to copy a paragraph, an entire magazine, a book or even a life’s work…[U]nless the intellectual-property rights of publishers are respected and enforced many desirable items may never be made digitally available, free or at any price." If the producers were able to release their goods in a form that cannot be copied, they would be much more willing to publish on the Internet. Instead of the relative dearth of information on the Internet right now, there could be a wealth of copyrighted goods for sale. Since customers will pay a premium to save the hassle of going out and looking for the information, there are strong market incentives to implement a system that will proactively protect intellectual property. Many corporations are currently involved in the development of copyright management systems, including Microsoft, Xerox, AT&T, and InterTrust. The major developers of digital copyright management systems refer to their projects as "trusted systems" because the copyright holders trust the systems to follow a basic set of rules to protect their rights.

Trusted systems are a complex intermingling of state of the art technologies. The heart of trusted systems is encryption, which keeps the digital information safe during transmission through insecure channels such as the Internet. Many trusted systems will interact using a protocol such as the Digital Protection Rights Language (DPRL). The DPRL allows producers to specify the availability and price of the rights to their goods and allows different trusted systems to interact using a standard language. Trusted viewers keep users from exercising unavailable rights when accessing digital works. If a digital work should escape the clutches of the copyright management system, digital watermarking provides a method for tracking the source of the information. Repository access services will store digital works for purchase and keep transaction and use records. Digital signatures will ensure that use is restricted a person or a group, not just certain access terminals.

Although there are many ways to maintain security in a trusted system, most systems currently in development protect digital works by using encryption coupled with a challenge-response protocol. For example, Adam might want to buy an electronic book about trusted systems from an online bookseller such as Borders. Borders’ computer, B, would have to know that Adam’s computer, A, is not only a trusted system, but that it was in fact Adam’s computer. (Adam would prefer not to pay for other people’s literary purchases.) Adam’s computer, A, sends B an encrypted digital certificate confirming that A is registered as a trusted system. Borders’ computer, B, knows that the certificate is real, but since certificates can be copied, B cannot be sure that the certificate came from A. To make sure that A really did send the certificate, B encrypts a nonce, a random string a numbers, with A’s public key. (Adam’s Computer A sent his public key with the original encrypted digital certificate.) Only A can possibly understand data encrypted with A’s public key. Adam’s computer A decrypts the nonce and sends the unencrypted nonce back to B, therefore proving A is a trusted system that belongs to Adam. Borders can be confident that Adam will pay for the book and only use the book in a proscribed manner, and will continue in the transfer of the electronic book to Adams computer.

There are ways to liberate copyrighted works from copyright management systems. Digital watermarks provide a method to make copyrighted works identifiable to trusted systems independent of their source. Digital media such as audio and video consist of millions of bits. Watermarking software can change several bits without a noticeable change to the user. Rights protection software can recognize identifying information in the watermark, such as serial number or a copy control bit. With sophisticated watermarking, the watermark is traceable even if someone uses a camcorder and records video from the computer screen or records audio from the speakers. If the copy control bit is set, a trusted viewer can prevent copying. Adobe Photoshop, the most popular image manipulation software, currently employs such a copy protection scheme. In the future, entire operating systems may perform this function, virtually eliminating the illegal copying of video and audio.

The most widely used trusted systems, the ones that operate over the Internet, will use a digital rights language. A digital rights language allows different types of trusted systems to communicate and exchange data. Kahn and Wilensky describe in detail a system that uses a language similar to the DPRL. Their hypothetical system could be quite similar to a professionally developed architecture. A digital rights language enables the use and transfer of a "digital object." A digital object is a structure that includes encrypted data, an identifier, and possibly other information such as contents or rights information. Access and use of the data inside digital objects will be facilitated by a digital rights language that most trusted systems will support. The digital rights language will include commands such as ACCESS, DEPOSIT, ENCRYPT, DELETE, and COMPRESS. System architects will most likely expand the digital rights language to include mechanisms for payment, negotiation, or interaction. Only certain parties will be able to use the restricted commands.

The objects will be stored and catalogued in a repository. The repository may have many other value added services such as search or support. The repository will keep transaction records and will either outsource or process the transactions in house. The repository will have an unprecedented amount of information about users access. What is done with this information is up to the company in control of the repository. An owner (not necessarily an individual) of digital objects can decide what rights go along with his digital objects, for example, allowing printing for ten cents a page and viewing at two cents a page.

A central organization may regulate the financial and rights transactions permitted by the trusted system. When Joe pushes a button to print a document for the price of three dollars, Joe’s system will send an instruction to a financial clearinghouse to deduct three dollars from his account, or Joe may pay directly with digital cash. If Joe wants to lend his book to his friend Sue, his system will interact with a rights clearing house to ensure the transaction is allowed, and to record the transfer of rights from Joe to Sue in case of dispute. The financial and rights transaction clearinghouses could provide very useful aggregate user data to the content providers to help them market their goods.

Ultimately, discussion of trusted systems will depend on the high level interaction between producers, users, and the trusted systems. Although we are unsure of the ultimate form of trusted systems, there are many similarities between the trusted systems currently in development. Northeast Consulting gives an interesting hypothetical template for the rights involving the use of a news article:

The producer will have the capability to disallow printing, cutting, and pasting. Many companies plan instead to charge for these practices. Independent of rights-holders decision to allow cutting and pasting, the overall architecture for trusted systems should remain the same: allowing, charging for, or prohibiting a wide array of rights. The rights included in the template above include two of the four basic digital rights. It describes the permissions and payment for render and derivative-work rights. Trusted systems may also allow transport rights such as permission to copy, transfer or loan, and safety rights such as the ability to make backup copies. There is some reason to believe the trusted system that provides the broadest framework for digital transactions will emerge as market leader.

The companies that succeed financially in the sale of their trusted systems will ultimately decide the architecture of trusted systems. Consumers will have a collective say in which system will survive, because they will be the ones paying for the system. Then again, producers may have a bigger say in the financial success of trusted systems. Producers will only provide their data on the systems where they can receive the greatest financial gain.

There are many established companies in the trusted system game. Two companies are building trusted systems for specific markets. The Secure Communications Systems division of AT&T is developing Information Vending Encryption System (IVES) technology. The IVES technology allows simple transactions based on credit and gives users full access to the data after payment. This technology will probably work best for the sale of content of ephemeral value such as news broadcasts or low fidelity audio broadcasts of radio. CD-MAX, Inc. developed a technology based on accessing encrypted data on cd-rom. Users pay for the key to decrypt a CD online. CD-MAX provides the software that invisibly decrypts the information for viewing. The unencrypted data is never stored on the computer. Producers can chose from a wide range of charging options, with different rates for printing and viewing, and allowing prepayment or credit for goods. CD-MAX acts as the financial clearinghouse and acts as the intermediary for all transactions between users and producers. CD-MAX could be an excellent solution for the secure distribution of trade journals and monthly news magazines.

Xerox’s trusted system has a wider market potential. The company has developed perhaps the most secure trusted system to date. The trusted system is based on secure hardware systems interacting with a digital rights language. Users can install different levels of secure hardware. Content providers can choose the minimum level of security a system must have for the user to access their data. The system allows a wide array of pricing models. It also allows the making of backup copies and the loaning of data objects. It can automatically block access to minors to inappropriate data. Xerox’s system appears to give the most options to publishers and users, and may enjoy widespread adoption. An important part of Xerox’s system is hardware based, and the added hassle and expense of another piece of hardware may compel consumers to favor a software only solution.

InterTrust appears to be the leader in a software solution to copyright management. The heart of InterTrust’s Enterprise 1.0™ system is the DigiBox container. The DigiBox is a type of cryptographic box, with the critical data encrypted (inside the box) and pricing and use information unencrypted (outside the box). Enterprise allows a wide array of pricing options to the producer. InterTrust has formed a strategic alliance with the Copyright Clearance Center to handle the financial aspects of transactions. Enterprise includes a secure metering system so any organization could run both a content and financial clearinghouse. InterTrust is also currently in development of the next generation standard for storage and playback of video and audio, MPEG-4. The main difference between the MPEG-4 standard and its predecessors is that the MPEG-4 standard includes copyright management structures.

Two other major companies have voiced their intent be in market for trusted systems. The Copyright Clearance Center is in a strong position to handle the financial clearinghouse operations of many trusted systems. The Copyright Clearance Center currently licenses and collects royalties for the photocopying of a majority of published work. Their business plan is to perform similar functions for digital publishers. The most important player in the software game, Microsoft, has hinted at its intentions to make future generations of Windows serve as copyright management systems. Microsoft has provided no further information.

If users want to have easy access to a variety of copyrighted works, they will have to use a trusted system. Richard Stallman wrote an interesting science fiction piece about why consumers should be wary of the widespread adoption of trusted systems. In his futuristic world, everybody’s computer is a digital spy. The computer checks to make sure all copyrighted material on the hard drive legally obtained. If the computer finds illegal material, it alerts the central authority and the offender is arrested. Everyone is required to pay for what they read, so only the very wealthy are able to read freely. Poor people can afford to read only a few books a year. Hacking around the copyright management system is illegal. (Welcome to the future. The Digital Millenium Copyright Act makes hacking a copyright management system a severe crime.) Everyone uses a computer with a copyright management system, because the other operating systems are illegal.

Trusted systems in the future will have the ability to violate rights currently enjoyed by Americans today. The corporations designing trusted systems today will only protect rights as far as it helps them market their systems. Technology moves faster than law in most cases, so legislators should tread lightly in cyberspace. Then again, "[t]o argue that real space should leave cyberspace alone one needs a normative argument – an argument about why it is good or right to leave cyberspace alone." The rest of this paper describes the many cases where the market solution may not be the best for society and suggests remedies for these market failures.

As discussed previously, trusted systems are strongly supported by the copyright industry and the Congress. The reason for the copyright industry’s support is obvious; trusted systems could give them more control than ever over their products. The reason for governmental support, however, is less obvious. If we give the Clinton Administration and Congress the benefit of the doubt and ignore the possibility that campaign contributions have played a role in their decisions, we must assume that the government deems it appropriate to extend copyright protection beyond what is currently accepted. To evaluate this position, we should consider the major arguments for and against extending copyright. As we shall see, the arguments for extending copyright are based on a limited view of the purpose of copyright, since they support copyright owners’ interests at the expense of society. It seems, then, that recent laws, notably the Digital Millenium Copyright Act, defend a position that fails to support the original purpose of copyright.

The notion that copyright exists to reward copyright owners for their work is known as the labor-desert theory. Lobbyists for the copyright industry invariably argue for this approach. If the idea that copyright is for producers is considered exclusively, this position is essentially a defense of natural rights. This theory suggests that individuals have the right to the fruit of their labor, and the state has an obligation to ensure that they receive it. In its extreme form, this says that copyright owners should receive all of the social benefits that accrue from their work. As Stephen Breyer points out, this is usually neither possible nor desirable, since it would require the copyright owners to reap all the benefits from their work, while the consumers reap none. To accomplish this, copyright owners would have to use perfect price discrimination to capture the entire consumer surplus. While this goal is identical to the private firm’s goal of profit maximization, it runs counter to our social preference for consumer protection. Firms naturally want to extract as much consumer surplus as possible, since more consumer surplus equates to more profit. Society, however, is usually in favor of laws that protect consumers from exactly that sort of loss. By focusing exclusively on copyright owner rights, the adherents of the labor-desert theory only support one of the goals of copyright – namely, the preservation of incentives to create intellectual works.

Copyright exists to support the creation and dissemination of intellectual products. To more fully understand how copyright can achieve this goal, we can turn to the economic theory of copyright, considering intellectual products as a public good.

As mentioned earlier, intellectual products are a public good. This was the motivation behind the original purpose of copyright, "to promote the advancement of science and the useful arts." The economics of public goods have been studied extensively, as have the potential methods to promote their efficient allocation. This section will first explain what it means to say that intellectual products are a public good, and why there would be a market failure in the absence of government intervention. The section will then describe how the government can promote information’s efficient allocation, and explain how this relates to copyright law. Finally, this section will describe the differences between the optimal approach and the labor-desert approach in light of the growing use of trusted systems.

Public goods pose interesting problems for governments. Everyone wants better education, less crime, and cleaner air, but it is not always clear what the government’s role should be in promoting the optimal amounts of each - or in economic jargon, efficient allocation. Before applying the theory of public goods to information, a few definitions are needed:

A public good is one that satisfies two criteria; it must be non-rival in consumption and non-excludable. A good is non-rival in consumption if one person’s use of it does not restrict another’s. A good is non-excludable if people cannot be easily prevented from consuming it. Two goods with these qualities are safe streets and clean air. They can be enjoyed by all, so they are non-rival, and no one can be prevented from consuming them, so they are non-excludable. Intellectual products also satisfy these criteria; they are non-rival since one person’s enjoyment of a book or movie does not at all detract from another’s ability to enjoy it. And since intellectual work is often easy to copy and distribute, it is non-excludable.

Public goods are problematic because it is often difficult for a society to create enough of them. A society that hopes for optimal provision of public goods faces the "free-rider" problem; everyone hopes that other people will provide the good, so not enough people provide it. For this reason, the market will not always provide the goods in efficient quantities without some sort of intervention.

There are several ways that a society might create public goods. Essentially, either the government or the market can provide the goods. Government provision of public goods is often accused of inefficiency and inflexibility. Private provision generally requires some sort of intervention to give authors incentives to produce works; recall that public goods are characterized by a market failure, so producers will not create works without the proper incentives. Both forms of provision would increase the amount of intellectual work compared with a situation with no government involvement. The challenge is to find which method could achieve the proper amount at the lowest cost.

Government provision of intellectual products would be one way to promote the market for intellectual works. Unfortunately, it would probably be extremely difficult to design and administer. If the government wanted to provide society with books and art, it would have a few options. It could hire authors and artists to work for the government, and give away everything they make. Or the government could pay artists who create works that society values. Designers of such a system would face many difficult questions. They would have decide who to pay, and how to pay them. They would have to decide whether payments should be determined based on popularity, production costs, or other factors. Another approach would be to reimburse works that are perceived to have high intellectual value more than works that are simply popular. While such a system would probably not be an appropriate replacement for copyright because of the large administrative costs, it could be used to supplement copyright law. This could explain why the government funds the arts.

In certain circumstances, private provision is far less costly than public provision. This idea was formally explained by Ronald Coase (Coase theorem) in the Coase theorem, which says that assigning property rights will lead to an efficient outcome if transaction costs are zero. He suggests that private markets are often best able to solve externality problems. But applying that result is difficult for two reasons. First, the theorem says that perfect efficiency can only be reached when transaction costs are zero, and they are not, although they may be headed that way with digital technologies. Most importantly, however, it is not possible to perfectly propertize intellectual products. To perfectly propertize a good, the owner must bear its full cost and benefit. While firms with copyrighted works bear the full cost of development, they cannot possible receive all the benefits from the works use. A publisher can not, for example, receive the benefits from a new novel that was loosely inspired by an old novel. This is why intellectual products create a positive externality problem. Since the products provide benefits that the producers cannot receive, the producers will underproduce. What is necessary, then, is some policy beyond simple copyright that will ensure that more works are created.

Copyright owners seek to maximize the benefit that they receive from their products. Unfortunately, this does not necessarily maximize social benefit. For example, a teacher might give copies of a short story to students in her class. If the benefit to the class was greater than the cost of the loss of producer incentives to create additional works, it is best for society to allow the educational use. Given complete control over usage, however, the publisher will seek to prevent any use that would decrease its profits. So the firm would prevent the use if it could, even though the use is good for society. The difference between private and social benefit is a crucial underpinning of copyright law. The balance of copyright law reflects the ideas that the private benefit to producers must be maintained to ensure incentives to create, and social benefit will be greatest when intellectual works are widely available.

The distinction between public and private benefit, while crucial to the purpose of copyright, is all but absent from such recent legislation as the Digital Millenium Copyright Act. The act provides legal force to the restrictions on consumer rights imposed by trusted systems. The law makes it illegal to circumvent security measures that protect information. Trusted systems, as described earlier in this paper, are probably perfectly capable of preventing unauthorized access of all works under protection. This makes the provisions that would allow limited circumvention for certain "fair uses" largely irrelevant.

This discussion of the purpose of copyright suggests that the DMCA, and the copyright lobbying industry in general, take a limited view of the purpose of copyright by supporting private interests at the expense of public interests. These interests, which include fair use, first sale, and limited duration, are an essential part of copyright, just as the preservation of producer incentives to create is another essential part. The rest of this paper will explain how and why these public values should be protected.

While it is clear that trusted systems could allow copyright owners to control the use of their products to an unprecedented extent, it is less clear how this change will affect the market for copyrighted works. This section will describe how trusted systems could revolutionize the market for intellectual works, and compare the promise of this new market with the threat of breaking from the status quo.

Of all the powers that trusted systems could grant to copyright owners, price discrimination stands out as the one that is uniquely positioned to change the way that markets allocate information. While trusted systems could enable publishers to charge different prices for different access rights to identical products, this is not very different from selling different products in normal markets. Trusted systems do, however, have the unusual ability to allow publishers to use price discrimination to charge different prices to consumers with different abilities to pay. This practice could have significant effects on the market’s ability to allocate goods efficiently and equitably.

Trusted systems could facilitate price discrimination by allowing publishers to prevent consumers from selling their work. This would not necessarily be illegal, since the Robinson-Patman Act, which makes price discrimination illegal, applies to sales but not leases. Furthermore, the Robinson-Patman Act applies to "physical commodities," not "intangibles." A publisher could charge a consumer a fee for viewing rights to a document, but the consumer would not necessarily own the document. And since the transaction would not be a sale, there would probably be nothing to prevent the copyright owner from using price discrimination.

As suggested before, trusted systems offer both a promise and a threat. The promise of price discrimination is that markets could become more efficient by selling to consumers who were previously unable to afford the particular works. But the promise carries a threat; if publishers can sell to the very poor at lower prices, they can sell to everyone else at higher prices. If trusted systems allow copyright owners to charge consumers a price close to the maximum that each person would be willing to pay, most consumers would be worse off. As will be discussed, appropriate legislation might make it possible to enjoy the benefits of the promise while correcting the threat of widening the income gap.

An unfortunate feature of markets for intellectual works is that there is a persistent deadweight loss; that is, there are usually people who are willing to pay more than a work’s marginal cost, but are not able to pay its full price. This problem is especially relevant to digital intellectual products, since the only costs that a producer incurs from selling a copy are transactional and distributional. But with electronic sales and distribution, those costs will approach zero. Since there is a substantial gap between a work’s price and marginal cost, there are bound to be consumers who are willing to pay a price that falls into that range. If a firm sells a video for $20, but its marginal cost is $1, it will lose the opportunity to sell to those consumers who value the product at less than $20 but more than $1. Price discrimination could allow firms to sell to those people, and, subsequently, earn more profit for each copyrighted work. The consumers, who gain the ability to buy because of price discrimination, would also be better off.

In addition to promoting efficiency, price discrimination could promote distributional justice. Price discrimination could help the poor enjoy a higher standard of living than is now possible, since they would be able to buy works that only wealthier consumers can now afford. As a result, intellectual works could be disseminated more broadly than ever, fulfilling one of the original goals of copyright, to a greater degree than is possible today.

And while this would certainly benefit the poor, they are surely not the only group that would benefit. Scholars might be allowed to buy expensive reports at reduced prices. Students and senior citizens might be able to identify themselves to receive discounts on certain goods. Although there are many groups that would be better off with wider access to copyrighted works, there are others that would suffer from higher prices for goods that they already buy.

The previous discussion explained how price discrimination would increase efficiency. It is important to note, however, that the most efficient market isn’t necessarily the one that consumers want to be in. A market is efficient when all mutually beneficial transactions are made; but that does not mean that the market is "fair" to the consumers. If copyright owners become able to price discriminate effectively, there will be a transfer of wealth from consumers to firms as firms charge consumers close to their reservation price. If publishing firms become much more profitable, the owners of those firms, who are likely to be wealthy, would benefit from the consumers’ loss. So price discrimination could result in a net transfer of wealth from the middle class to the rich. This is ironic since the previous section explained that price discrimination would probably make the poor better off.

Another threat from trusted systems is that they could reduce the information available to consumers about the value of various intellectual works. For markets to be efficient, consumers must know how much various goods are worth to them; otherwise, they would not know whether or not to buy them at the market price. Since trusted systems could eliminate the after-market for intellectual works, people who are dissatisfied with their purchases might not be able to resell them. This would hurt the consumer, since it would be harder to figure out if they value a particular good more than its price.

While this decrease in information would reduce the market’s efficiency, it might be offset by the large-scale increase in information available on the internet. As more sites like Amazon.com offer users extensive product reviews, markets for intellectual works will become more efficient. In other words, the decrease in information due to the elimination of the after-market could be made irrelevant by the increase in product information available through the internet.

So, it seems, we are forced to choose between efficiency and equity on the one hand, and consumers’ interests on the other. A close examination of the two choices, however, suggests that it might be possible to preserve both. To achieve the optimal balance, the government might use redistributive taxation or limitations on price discrimination criteria.

A redistributive tax might correct price discrimination’s widening effect on income distribution. Although the tax might be difficult to design effectively, it is at least theoretically possible. Designed properly, the tax would redistribute income toward a socially acceptable distribution without diminishing the efficiency that resulted from

price discrimination. Unfortunately, changing the tax code inevitably creates unintended complications; for example, the tax would need to redistribute income from producers to consumers, but the tax might instead redistribute income from rich to poor. If it were possible to transfer money directly from producers and to consumers without changing incentives, the distribution problem would be solved. This section suggested that producers and consumers were equivalent to rich and poor, respectively. This is just a simplification. However, embedding the simplification in the tax code by taxing the wrong group would cause market distortions, which might outweigh the benefits from increased efficiency. The distortions would occur when one of the groups was hurt or helped by the tax accidentally. The rich who do not benefit from copyright profits would be taxed, and therefore hurt, while the poor who do not buy intellectual works would be subsidized, and therefore helped. The challenge in designing the tax would be to ensure that its redistributive value would not be outweighed by efficiency losses.

The government might also want to limit copyright owners’ ability to price discriminate by limiting the criteria that firms can use to set prices. There is a spectrum of criteria that firms might use, from giving senior citizens discounts on one end, to setting prices as a function of income on the other. Modest forms of price discrimination, like senior citizen discounts, would increase firms’ profits modestly. Extensive price discrimination, which could allow firms choose prices for each individual based on his income and preferences, would allow firms to capture nearly all of consumer surplus.

Although government might want to promote efficiency through price discrimination, it would probably still be concerned about protecting consumer surplus. It could probably find a compromise solution in limited price discrimination. First, the government would have to choose which criteria would give firms too much power. It might decide to disallow firms to use salary information to set a different price for each customer. A less invasive method would probably be better for consumes, such as allowing poor individuals to identify themselves and receive lower prices. For example, everyone below the poverty line might be given digital ID’s that identify them as poor to a copyright owner’s trusted system. The firms could then set two prices; one for the poor, and one for everyone else. That way, these people would be better off, but firms would not be allowed to price as effectively as if they were able to identify each customer’s preferences and ability to buy.

By granting copyright owners unprecedented power to control their work, trusted systems could cause drastic changes in the markets for intellectual property. The following sections will describe the types of change that trusted systems might cause, and suggest ways to protect against those changes’ negative consequences.

Fair use is a privilege that allows someone other than the copyright owner to use the copyrighted material in a reasonable manner without the copyright owner’s prior consent, notwithstanding the monopoly granted to the copyright owner. Although originally a judge-made rule, the fair use privilege was subsequently codified in the Copyright Act of 1976. Under the Act, fair use may be made "for purposes such as criticism, comment, news, reporting, teaching (including multiple copies for classroom use), scholarship, or research." To put these purposes concretely, The House Report of the Act describes some specific examples:

Section 107 does not, however, attempt to define the full meaning of "fair use" but instead lists the factors to be considered in determining if a particular activity falls in the category of fair use. In deciding whether a given use is "fair," a court engages in a careful balancing and weighing of the following factors: (1) the purpose and character of the use, (2) the nature of the copyrighted work, (3) the amount and substantiality of the portion used, and (3) the effect of the use upon the potential market for the copyrighted work. Whether a given use is clear is therefore not always easy to determine, and the courts often proceed on a case-by-case basis.

Fair use should not be considered a bizarre, occasionally tolerated departure from the scope of the copyright monopoly. Rather, fair use plays an essential role in the whole system of copyright, and its value should be examined in light of the purpose of Copyright Act. Unlike other countries, U.S. copyright law has not been thought to be an inevitable, divine, or natural right of authors. Rather, it is designed instrumentally to stimulate creativity for the intellectual enrichment of the public. The Constitution recognizes this by granting Congress the power to "promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries." To achieve this goal, Copyright Act gives authors the monopoly power that allows them to enjoy the rewards for their creative effort.

Although this monopoly power of creators is designed to stimulate their creative activity, the purpose of fair use is also the same as the purpose underlying copyright in general: to promote knowledge and learning. It should be noted here that accessibility plays a central role in copyright theory. Since all creative activities depend on the works of prior creators to some extent, in order for the copyright system to serve its above-mentioned purpose, it is necessary to give some free public access to copyrighted works. Continuous reexamination of prior work is especially important for certain kinds of intellectual creative activities, such as philosophy, criticism, or history. Fair use is necessary since the creation of new works is dependent on the use of former works. If authors of new works or the public were not able to access existing works, it would be difficult to promote knowledge and learning.

The purposes stated in section 107 of the Copyright Act involve uses that typically possess a high social value relative to the losses suffered by the copyright proprietor. For example, multiple copying for classroom use is allowed on the grounds that the value to society of having a well-educated citizenry, when added to the private value enjoyed by the teacher and students, outweighs any loss to the publisher from the other uses it could have licensed. As to criticism and comment, the value of fair use, in addition to the social value of these activities, also comes from the fact that it is unlikely that a license would be granted. Similarly, parody has been thought to deserve substantial freedom, both as entertainment and as a form of social and literary criticism.

As discussed earlier, trusted systems enable copyright owners or publishers to distribute their works so that they can be accessed only by trusted machines. This permits copyright owners to charge for, and trace, every use of their digital materials. Although trusted systems cannot guarantee absolute control over information, especially after it escapes from the digital media, they do pose certain threats to the values underlying fair use. Although it is inherently difficult to predict precisely how cyberspace will look in the future, the following description is based on the expectation that, in the near future, much information will be distributed on the Internet only, and digitized publication will be far superior to printed publication both in speed and quantity.

With trusted systems, publishers will be able to charge users every time they access a work, whether for reading or copying, regardless of the purpose of that access (e.g. commercial or non-commercial). Teachers may have to pay to browse articles when they choose their course materials, and teachers and students may have to pay to download course materials. Scholars may also have to pay for all the research papers they read in order to write papers and for all portions they cite in their own paper. News reporters and critics may have to pay for information they collect and cite. Thus, trusted systems may have the effect of eliminating much of the free access that such individuals are entitled to in the hard-copy world under the fair use doctrine.

Some commentators argue that trusted systems, by permitting licensing of every use of a work and by increasing the total return to authors, will ultimately increase the level of creativity and secure better creative works. We should not, however, ignore the possibility that trusted systems may permit authors to obtain an excessive and unwarranted return and, ultimately, reduce the number and quality of works by eliminating much of the free access permitted under the fair use doctrine.

Consequently, we proceed on the assumption that trusted systems, if left unchecked, may threaten fair use and frustrate the purposes of copyright, namely promotion of knowledge and learning. One of the most serious concerns is a potentially harmful impact on public education. Attempts to control user access could result in excessive restriction, conflicting with rights of individual freedom and the objectives of public education. Users such as children in poor schools would be affected most severely. Scholarship and criticism might also be affected by both the economic burdens of licensed reading and copying. Even if scholars and critics have money to pay for such licenses, a pay-per-view system may dissuade them from taking frequent mental rambles through material, especially where those rambles may or may not turn up leads and ideas that could be incorporated into the user’s own work. Moreover, since trusted systems enable consumer tracking, they may allow authors and publishers to choose their readers, and it is possible that such rights holders might hesitate to give permission for reading to scholars or critics who may make negative comments. In other words, trusted systems may limit criticism, comment and parody by giving copyright owners some ability to choose readers. Thus, limits may need to be placed on the copyright owner’s ability to employ trusted systems.

In imposing such limits, however, we should keep in mind that it is still important to protect authors’ incentives to create. Thus, in the fair use context, it is important to examine whether the social benefit of the use outweighs the incentive loss to the copyright owner. In other words, the use must have a character that serves the copyright objective of stimulating productive thought and public instruction, and must not excessively diminish the incentive for creativity. Indeed, the fair use provisions of the Copyright Act require consideration of, among other factors, the impact of the use on the potential market for the work, which some courts have held to be "undoubtedly the single most important element of fair use." Thus, when imposing limits on trusted systems, it is important to consider the desirable balance between rights of copyright owners and public interests.

Indeed, as mentioned in an earlier section, trusted systems may play a very legitimate and useful role in preserving authors’ overall incentives for the creation of works in the face of the potential copying threat posed by the internet. From the view of authors and publishers, trusted systems technology is not a tool to collect undeserved money, but to recover their rightful rewards. The problem is that trusted systems also enable authors and publishers to charge fees that should not be charged in the light of fair use values. In choosing a desirable architecture, it is necessary to reserve the useful function of trusted system while keeping the value of fair use.

In considering possible responses to the threat posed by trusted systems to the values underlying fair use, an initial consideration is whether we even need to develop any standards for fair use or architectures to protect the value of fair use. The position adopted by the White Paper is that the market will adequately serve the values underlying fair use by providing free access and/or low-cost licensing. Thus, according to the White Paper, no additional provisions are necessary to deal with the values underlying the fair use doctrine.

We believe that it is unlikely, however, that the use of trusted systems in the market will adequately protect fair use values if left unchecked. Technology to protect copyrights is used mainly to protect entities that have economic power in marketplace, and not to provide benefit for society as a whole. This is a matter of course, since companies have no real incentive to develop or deploy a new technology if it will not be beneficial. Superficially, some publishers may appear to facilitate fair use values on their own. For example, Westlaw and Lexis-Nexis give law school students free access. However, in this case, their motivation is not protection of fair use values, but the expectation that they will get repayments when the students become lawyers. Moreover, as discussed in Part III above, there is no guarantee that low-cost licensing will be sufficiently low-cost, or that it will fully protect all of the interests underlying fair use. Unfortunately, we cannot expect that the marketplace will develop the appropriate standards protecting fair use. Therefore, to protect public interest to access digital information, legislative limits may need to be imposed upon the use of trusted systems.

An alternative response would be to carve out an exception to the DMCA and permit circumvention of technological protection mechanisms for the purposes of fair use. As mentioned above, the DMCA imposes additional penalties on persons who circumvent technological protection mechanisms that are designed to restrict access to copyrighted works. Under this proposal, there would be no liability under the DMCA for circumvention for fair use purposes. Courts could continue to develop fair use doctrines as exceptions to liability under the DMCA. Alternatively, the Copyright Office could take such considerations into account in deciding whether or not to exempt certain categories of uses, as it is currently instructed to do under the DMCA.

However, when trusted systems become so effective so as to make circumvention practically impossible, this does not provide any effective protection for fair use values. The fact that certain types of circumvention are not illegal is no help when circumvention is technologically impossible or impractical. Moreover, the DMCA’s ban on certain circumvention technologies effectively limits the ability of potential fair users to gain access to technology that would permit circumvention. Although the DMCA only bans technologies that are primarily intended for circumvention and have no significant non-infringing uses, it is difficult to imagine a device primarily intended and used for fair use, since the same device could easily be used for illegal purposes as well. Without access to such technology, those with legitimate fair use privileges will be practically unable to exercise those privileges. Thus, exempting fair use circumvention from liability under the DMCA would not provide much protection for fair use values.

c. Require Access for Fair Use Purposes

An alternative legal response would be to legally require trusted systems to affirmatively permit fair use. Under this view, the law would not simply make circumvention permissible, but would prevent copyright owners from using trusted systems to restrict fair use of their works. (The government could, for example, require all trusted systems to allow free access to students and scholars.) This would be more effective than the prior approach, since access would not depend on the user’s ability to overcome the technological protection mechanisms.

It may be difficult, however, to establish a concrete standard about what kind of activity falls in the category of fair use. Unlike the application of law by judges, a trusted system cannot easily be configured to account for a fuzzy concept such as fair use. The standards for fair use are quite vague, and the application of the standards by the courts occurs in a case-by-case basis. Few clear rules exist in the area of fair use. Although fuzzy standards can be applied by judges exercising discretion, it is far more difficult, if not impossible, to incorporate such fuzzy standards in a technological architecture.

Because technical architectures may not be able to adequately capture an inexact legal concept such as fair use, limits on trusted systems may have to be imposed to account for a number of clear and easily-identifiable subsets of uses within the overall concept of fair use. Although this may not preserve all of the values underlying fair use, it may go far in preserving many of the underlying interests. Moreover, some administrative oversight might be helpful in defining new categories of required access, as circumstances change over time. Such possibilities will be discussed in the following section.

Other legal measures might be used to ensure some protection of fair use values. One alternative would be to use tax treatment to further encourage copyright owners employing trusted systems to allow fair use, beyond minimal compliance with the law. The government could grant tax deductions to authors and publishers who significantly facilitate the fair use of their protected works. However, as discussed in the next section, this approach creates the risk that it might diminish the incentives of authors or publishers to publish materials in cyberspace. Another legal measure might be to discourage publishers from denying distribution to particular persons, such as critics or parodists. Since trusted systems could allow publishers to identify who is buying the material, the government should make it illegal for publishers to deny specific users access to material without a substantial reason, such as not selling pornography to minors.

d. Fair Use Site

One architectural, rather than legal, solution would be to establish a Fair Use Site. A Fair Use Site could be maintained by either the government or a non-profit organization. Copyright owners would be required to make their materials available on the Fair Use Site, instead of building fair use into their own trusted systems. Such a Fair Use Site would be designed basically for scholars, teachers and students. Each would be given a password to access the Fair Use Site based on his/her status, just as law school students currently have free access to Westlaw and Lexis-Nexis through passwords. Such a site would preserve some of the free access to works that scholars, teachers, and students currently enjoy as a result of the fair use doctrine.

Considering the balance between rights of copyright owners and the public interest, it would not be appropriate to give these individuals unlimited free access to all information through the Fair Use Site. If all students could make free copies of entire textbooks, publishers would not be able to produce enough profits to continue to publish textbooks. Consequently, a Fair Use Site with unlimited access would diminish incentives of publishers to publish textbooks. Accordingly, access to the site should be limited. For example, students could be given free access and copying worth up to $500 per year through the Fair Use Site. The actual amount could be determined and later adjusted based on statistical research. Although students would be expected to use the $500 for reading and copying materials for class, they would not be required to reveal their purpose for getting or copying materials each time they used it. If they exceeded the limited amount, they could then apply to extend the limit; however, at this point they would have to reveal the purpose of their access.

Under such a system, users will hesitate to give their password to others, since the total amount of free access is limited. Many students may use the free access to materials for purposes unrelated to fair use, and this would result in a loss to publishers. On the other hand, however, many students will not apply for additional access even after they use their initial limit, because it would take time to complete an application form. Consequently, if the initial amount is determined appropriately, publishers will not suffer unreasonable revenue losses. Such a system would thus preserve some of the free access to materials that certain individuals currently have under fair use, without unreasonably undercutting the incentives to create the works in the first place.

e. Digital Library

The public interest in freely reading books at public libraries is an important interest underlying fair use, and could be supported through establishment of Digital Libraries. From the computers in such a Digital Library, people could gain free access to materials in digital form. The Digital Libraries, however, could choose the materials they buy, just as libraries now choose books to buy in real world. It should be possible to borrow material from the Library for a limited time, in a particular digital format that is readable only for a limited time and then disappears from the memory of the user’s computer. Copying material at the Library should be also possible, for a small fee, just like we now have to pay for photocopies at the library while reading itself is free.

Considering the purpose of a Digital Library, users should not be required to show any identification to access the internet from a computer at the Digital Library. Since the Digital Library could choose the materials it offers, we would not have to worry that a Digital Library would provide opportunities for minors to enjoy pornography.

Given the options listed above, we believe that some combination of both legal and technical architectures would be appropriate to protect the values underlying fair use. From a technical perspective, we recommend the creation of some type of Fair Use Site operated by a non-profit entity, rather than trying to require copyright owners to build fair use into their trusted systems. There are two reasons to support a Fair Use Site. One is the burden of cost. As mentioned before, trusted systems play a useful role in permitting copyright owners to collect rightful rewards for their work and in maintaining incentives to create works and publish them in cyberspace. From this point of view, a burden of cost and time to build-in systems to protect fair use may not be desirable, since it may shrink the value of the trusted system itself. Second is the issue of privacy. Since both built-in systems and a Fair Use Site function based on the identity and status of users (scholars, teachers, or students), some identification of users may be difficult to avoid. To protect users from misuse of their private information, a non-profit entity would be preferable as an operator of the system rather than publishers or government.

For the reasons discussed above, we would also recommend establishing a Digital Library to protect the public interest to freely read books.

Thus, the law should require all publishers to:

1. make their material accessible through the Fair Use Site
2. sell their material to a public Digital Library for reasonable prices
3. only discriminate among users if there is reasonable cause.

The First Sale Doctrine ("first sale") is one of many limitations on the monopoly rights granted by copyright. First sale is a limitation on the copyright holder’s exclusive right to distribute to the public. First sale restricts copyright holder’s distribution rights to the initial distribution of a particular copy of the work, and in alignment with the law’s general resistance to placing restraints on alienation rights, it prohibits copyright holders from controlling copies of their work after the transfer of ownership.

As soon as someone becomes the owner of copyrighted material, she is free to do what she wants with her particular copy. According to the Copyright Act "the owner of a particular copy [of a work] … is entitled, without the authority of the copyright owner, to sell or otherwise dispose of the possession of that copy." When a copyright holder sells a particular copy of a work, he may no longer control its distribution. The purchaser (now the new owner of a copy) is free to sell the copy to someone else if she should choose to do so. Once a book publisher sells a copy of its book to a bookseller, the publisher does not have rights to require the bookseller to take any other actions, except by separate contract. The publisher cannot, for example, require the bookseller to display the book prominently in its store windows or to sell the book for a certain price. After the initial point of sale, First Sale limits the control that a copyright holder has over his work.

It is important to note that First Sale applies directly to distribution rights; it has never been intended to cover duplication rights. As technology has advanced, however, and duplication has become easier and more intertwined with distribution, First Sale has effectively extended its bounds of restrictions to issues of duplication.

At a basic level, the First Sale construct encourages the freedom of sharing. First Sale is one of the provisions built into the Copyright Act to protect the public interest of access to information. The philosophy is that members of society can best build on the knowledge that already exists, and advance the public good through "progress in science and the useful arts" if broad access to the existing knowledge base can be ensured. Copyright and First Sale encourage creators to share their knowledge by granting them certain monopoly rights as copyright holders over the distribution of their work.

First Sale is one limitation on the monopoly rights granted by copyright that aims to protect the reasonable freedoms of lawful users of copyrighted material. First Sale prevents copyright holders from restricting appropriate uses of copyrighted works after the users have received those works through the appropriate distribution channel and, if necessary, have compensated the copyright holder for access to the material. The First Sale construct attempts to preserve the freedom of customers to use copyrighted works after they have compensated the creators to make the work available to them. Limiting the copyright owner’s control over copies to the first time they are sold helps enable libraries and used bookstores to make information available to the public at minimal cost.

With modern technology distribution implicitly requires duplication. The First Sale Doctrine, which was conceived before the advent of modern computer technology, covered only distribution rights for copyrighted works; it did not address duplication rights for copyrighted materials. However, the way that computers pass information today inherently involves duplication. When a computer passes information, it creates a copy. A duplicate version of the information exists for an instant before it can be erased. In a narrow reading of the First Sale doctrine, then, this distribution by computer transmission violates the letter of the law.

If some technology were available to allow the transmission of a copy without making an unlawful reproduction -- so that no copy remains with the original owner -- the first sale doctrine would apply and the transmission would not be an infringement. Some argue that the first sale doctrine should apply to digital transmission, in the case where the transmitter destroys or deletes the original copy from which the reproduction in the receiver originated. Proponents of this view argue that at the completion of the transfer activity, only one copy would exist between the original owner who transmitted the copy and the person who received it -- the same number of copies as at the beginning. Opponents argue against this view by noting that whether or not the same number of copies exist after the transaction has finished, that the transaction when viewed as a whole still violates exclusive rights and that there can be no applicable exception from liability.

The language of the Copyright Act, the legislative history, and case law make clear that the doctrine is applicable only to those situations where the owner of a particular copy disposes of physical possession of that particular copy. In this case, without any doubt and by nature of the technology, a reproduction of the work exists in the receiving computer. The Working Group on Intellectual Property Rights interprets the law strictly to find that the application of the first sale doctrine in such a case would infringe upon the reproduction right. This would seem to indicate that all sharing of copyrighted information by means of modern data transmission technologies is infringing.

A more reasonable position is that form should not be elevated over substance and that the statute should be modified to allow a transaction in which only one copy remains at the end, regardless of how that end was achieved. An alternative to this is to narrow the scope of the First Sale Doctrine to exclude copies obtained via transmission. For instance, if a copy of a work is legally purchased on-line and downloaded onto the purchaser's disk, the disk could not be resold. The first sale doctrine should apply if the particular copy involved is in fact the copy that is further distributed, even if the copy was first obtained by transmission. Further, if the technology utilized allows the transmission of a copy without making an unlawful reproduction -- i.e., no copy remains with the original owner -- the first sale doctrine would apply and the transmission would not be an infringement.

All transmissions of information by computer should not be violations of intellectual property law. So long as the copy is valid under duplication rights or other access rights, then any legal construct (First Sale included) should not restrict the ability to share because of the medium used. We believe that if First Sale is going to be interpreted narrowly, that its scope should be limited accordingly to exclude copies made by transmission involving duplication. In these cases of transmission with duplication, where First Sale would not apply, we need some other provisions to ensure that the value of sharing can still be enjoyed.

At the same time that many are suggesting that the domain of First Sale should be limited, Copyright law as a means of protection for intellectual property is being supplanted by technical architectures. These architectures - trusted systems - are challenging the public interests of sharing. By implementing digital rights languages, trusted systems can exercise complete control over the distribution of digital works, including both the initial distribution and further distributions. In part because of these advantages of the perfect protection offered by trusted systems, many content providers are choosing to use technical architectures instead of law to protect intellectual property. The rise of access control architectures threatens to further unbalance the value of sharing that the First Sale Doctrine represents – and this public interest value is already on shaky footing in cyberspace.

Trusted architectures can completely eliminate uses that we take for granted in the real world. When we buy a record we assume that we can play it forever, or at least until it physically wears out; however, the rights to play a digital album might be restricted to last only for a finite period. In the real world, you can hand a book to your spouse and show him or her a quote. You might leave a book open on your coffee table in your living room, or in the waiting room at your office for your guests to browse. Most would agree that these seem to be reasonable uses that enhance the intellectual life of the populace. The ability to share could be very different in a cyberspace governed by trusted systems, however. Technical architectures can prevent your spouse or your friends from ever metaphorically opening your book or reading any parts of it, however small. Or if the trusted architecture did allow you to share your copy it could cost you significantly. These examples highlight the difference between our notions of what reasonable uses should be, and the potential consequences that trusted systems could cause if they remain unchecked.

Certainly the uses of copyrighted material described in these examples are not destroying the market for the book or album, or deterring the authors and artists from creating in the first place. Indeed, this type of sharing does not have any effect on the author’s decision or motivation to create the work. This sharing might even benefit creators if, for example, your guest chooses to go out and buy a copy of the book for himself because of his introduction to it through your copy.

Using trusted architectures, providers can precisely control access to information. Trusted systems have the capacity to reduce and eliminate the possibility of sharing information at all. Unless we use law to build-in provisions to protect the values of sharing that the First Sale Doctrine represents, trusted systems threaten to eliminate this important mode of learning and intellectual discovery.

There are many forms that the technical architectures could take. The key is that these architectures incorporate provisions into trusted systems which allow for reasonable opportunities for sharing to continue, but also prevent rampant piracy that could result if it is too easy to copy and share. We will next outline several of the alternatives that could be implemented. These alternatives include the following ways in which trusted architecture can allow for copying:

The trusted architecture could prevent users from copying copies. By allowing copies to be made only from the original, the system can prevent a runaway escalation in pirated editions. Under this system copies can only make a copy from an original. These schemes which allow copying are probably most appropriate for works that maintain their value even after being lent out. For these works, like books and music albums, sharing would probably not seriously affect their sales, since flipping through a book could make a consumer want to buy the book. For other works, like articles, the opposite is true: a consumer that borrows a friend’s newspaper is much less likely to buy it himself. For these types of works, rampant copying could significantly damage the market for the good.

This architecture could be complemented by a provision to preclude the simultaneous use of more than one copy at a time. Under such a system, even the original purchaser would not be able to access the material if someone he shared a copy with was already accessing the material. By including this provision, we build into the system deterrents to letting duplicates of originals spread too far, or beyond the copyright owner’s realm of control. In the real world, one would be hesitant to lend a book to another that he did not trust to bring it back. Under this architecture, users would be hesitant to distribute copies of a work in a way that would make it difficult for them to gain the access which they had rightfully purchased. Under this architecture, users would be able to lend with complete assurance that their copies would be returned to them, since the system could automatically return the copies after some pre-determined period.

The copying provision could alternatively be complemented by a provision that could allow copies to exist for a limited time before the trusted system would deny further access. This architecture would allow sharing for a limited time period, but would also be valuable in deterring piracy. This architecture shows a lot of promise, especially if it can be combined with the copying architecture that would allow duplicates to be made only from originals. Even if a copy were illicitly distributed, it would destroy itself after a specified time and not be able to be reproduced in the meantime.

In addition to allowing copying, trusted systems have the capacity to place restrictions on transfers. However, in the digital world, certain types of transfers may be valid and reasonable uses of the work. For example, a trusted system could prevent a person from emailing to himself an article that he has paid for. Provisions need to be built into trusted architectures to allow for this kind of reasonable transfer, which does not destroy the market. An appropriate transfer protocol seems to be the simultaneous use provision. So long as a work is not being used in two locations simultaneously, the transfer of the work would seem to be fair and reasonable.

One way to allow sharing and effectively prevent simultaneous use is to use a check-in system. Like the DIVX standard, trusted systems can allow users to share through a process of checking-in with a central source. Under such a system you can loan your copy of a digital movie to your friend, so long as you and your friend call into the central command center to transfer the rights of use. Clearly this process of checking-in has many negative implications for privacy, since the system could be designed to collect data on user behavior. As technology advances, some technical alternatives could allow a more reasonable and privacy-conscious way to prevent simultaneous use.

In different combinations, the provisions we have outlined could preserve some of the real world barriers to sharing and lending, obstacles that prevent the rampant spread of pirated books and video games which would leave creators without any compensation for their efforts. The challenge to designing one of these trusted systems is ensuring that some barriers to lending remain. Otherwise, sharing in cyberspace could become so easy that it destroys the market or the incentives to create in the first place.

Surely, authors and other copyright holders have valid concerns about recovering their costs and risks to innovate, and in protecting their intellectual property from piracy and theft. However, for an era where trusted systems can offer perfect control of access, we must weigh these private interests against the public interest in the reasonable sharing of knowledge. The value of sharing that the First Sale Doctrine embodies is worth preserving, so long as creators will still produce works. We must remember that the primary purpose of copyright, as it is constitutionally conceived, is not to secure a return to authors but to advance the public good. We propose to limit the scope of First Sale so that it does not cover copies produced by transmission, and to use law to require that the architects of cyberspace build-in technical protections to preserve these values.

Unlike the other limitations on the monopoly granted by copyright law, the limited duration provision was, and is no longer, a function of the traditional architectural limits of available technology. For instance, in the traditional media world, copyright owners would be hard pressed to eliminate the transfer of a work – books can easily sold or loaned to another party. In the past, policing all such behavior was almost impossible. Whether originally as a result of the architecture of books or as a value to protect in itself, the first-sale doctrine arose to allow such uses. Contrast this to the limited period of protection offered to copyrighted work. The Founders could have conceivably provided for an infinite period of copyright since no architectural limits existed which would make that impractical.

However, to make copyrights last forever would conflict with the proclaimed purpose of the Copyright Clause. As quoted previously, the "Congress shall have Power … To promote the Progress of Science and useful Arts, by securing for limited Times to Authors and Inventors the exclusive Right to their respective Writings and Discoveries." Society strikes a bargain with the author solely for society’s benefit – in exchange for producing the work, society grants a limited in time monopoly. Providing an economic incentive will encourage authors to create and as a result, society will benefit. Determining the appropriate level of economic incentives is at the heart of the balancing procedure and is inherently a utilitarian approach.

Consistent with its constitutional power, Congress has enacted a series of copyright statutes that set the standards and protections offered to copyright holders, which include a limited duration monopoly. Only a few months ago, the term of protection given to copyrighted works was generally life of the author plus fifty years. In the case of "works made for hire", the duration of protection was seventy-five years. On October 27, 1998, the Sonny Bono Copyright Term Extension Act extended the duration of protection generally to life of the author plus seventy years and for works-for-hire to ninety-five years. The justification for linking the copyright duration to the author’s lifetime relies on natural property rights. Authors should enjoy the fruit of their labor during their lifetimes and so should their immediate dependents. Though this is not in line with the constitutionally announced utilitarian purpose of copyright, the labor desert theory certainly has influenced the development of copyright law.

After the period of copyright protection ends, the work becomes a part of the public domain so it can be freely adapted, copied or otherwise used. Since future creative work depends in part on the existing work within the public domain, we want to maximize the number of works in the public domain and the creation of new works. Each of those goals positively reinforces the other. If the duration of protection could be shortened by a few years without an ensuing diminution in the number of works created, society as a whole would be better off and this would be the appropriate course of action.

The introduction of trusted systems, however, can foster the opposite trend – increasing the duration of protection possible infinitely. Though the monopoly offered by copyright law would expire, the protection of trusted systems could continue. Trusted systems could impose access restrictions forever and could become the dominant means of protection. Copyright law may be relegated to a secondary status.

The utilitarian motivation of copyright law would not favor the infinite protection implications of trusted systems. Possessing a limited monopoly forever does not seem to provide any incremental incentive to create. The cost of having fewer works within the public domain would be borne by society without any additional benefits. Investment in copyrighted works would remain essentially unchanged while individual authors’ descendants may benefit.

Because the duration of copyright protection is theoretically set by a balancing of the incentive to create and of the public interest in access to creative works, we should not allow technology such as trusted systems to disrupt the contemplated balance. Overprotection will harm our future selves by reducing the number or quality of future creative works. Authors themselves will not benefit from an infinite protection period – the future parties who control their works will benefit. If those parties control the work in their best interests, namely economic interests, society will be worse off.

If we accept that copyrighted work should not enjoy infinite protection, the next question is what period of protection should be allowable. The most logical choice is the duration of the copyright. We would not want the period to be longer the copyright duration for that would disrupt the balance created. If we set the maximum protection period of trusted systems to be less than the copyright period, then some copyright infringing acts would occur and this may increase the enforcement costs of the copyright owner. This factor would weigh in favor of not having the trusted systems protection be less than the copyright protection. On the other hand, if other values we posses are not adequately provided for, such as fair use, we may want to allow leakage to compensate for the reduction or elimination of those values. Proceeding on the assumption that other values will be protected, we should set the maximum protection period of trusted systems to the copyright period.

In order to ensure a limitation on the protection period, the architecture of trusted systems must incorporate some concept of time. If the distribution model is based on a central source in that all copies of created works reside on a single source, that central source will maintain a list of the protection expiration dates associated with the list of copyrighted works. Conceptually this architecture can be thought of as expiration tags. The tag would at least have to contain some identification of the work and the expiration date. If the distribution model were based on distributed sources so that copies existed in multiple locations, each work would have the tag attached to it. This tag would only have to contain the expiration date.

Another feature of the expiration tag is flexibility of the expiration date. The copyright owner should have the option of choosing the expiration date anytime from the creation of the work to the expiration of the copyright period. Though most of the copyright owners may choose to longest possible period, some may want their work in the public domain and this should be encouraged. For instance, if an author wanted to cultivate a market for his/her work, he/she may want to distribute the work for free so that demand will exist for later work. This author would set an immediate expiration date on the tag so that the trusted system would allow all types of access to the work.

Once the copyright owner has initially set the expiration tag, he/she should have a limited ability to change the date. The copyright owner could choose to bring the expiration date forward, and, of course, this should be permitted. However, the owner should not be able to extend the expiration date. In terms of incentive theory, once the work has been created and distributed, there is no additional benefit to society of providing additional protection to the work. Furthermore, consumers of the work may have purchased the work with the expiration date in mind. Their expectations should be protected so the copyright owner should be bound by the initial decision.

In order to prevent a conflict between the trusted system distribution scheme and copyright law, the term of protection for a work should be the lessor of the protection offered by the trusted system and that offered by copyright law. One should not be liable for copyright infringement if access is freely allowed by the trusted system. The owner of the work should be estopped from claiming copyright infringement if he/she chose a shorter period of trusted system protection. Congress can make slight modifications to copyright law to effect this change. For instance, suppose the copyright owner set the expiration date to be five years after creation. Near the end of the five-year period, the revenues garnered from the work suddenly increase so the copyright owner’s best interest would be to extend the expiration date. The architecture of trusted systems would not allow it, but copyright law would be a good substitute mechanism for extending protection. To resolve this issue, if the copyright owner sets the tag expiration date earlier than the copyright expiration date, the copyright owner should not be permitted to use copyright law to prohibit copying or other use during the intermediate period. A trusted system is not necessarily a copyright system.

One small implementation issue that can be exploited for the public interest is the indeterminacy of the author’s lifetime in cases where that is a relevant date. The provisional tag expiration date could be set anytime from the creation date to one-hundred years following it. Upon the author’s death, the copyright owner could set a new expiration date up to the copyright expiration date, that is, seventy-years after the life of the author. If the copyright owner does not set a permanent tag expiration date before the provisional tag expiration date occurs, then the provisional tag expiration date becomes the permanent tag expiration date. Another resolution may be to only allow the permanent tag expiration date to be set during a fixed number of years following the author’s death. The one-hundred years is a somewhat arbitrary figure but is set with the author’s life expectancy in mind. Ideally, the provisional tag expiration date should never occur before the author’s death. The one-hundred year term is chosen to ensure that. In the very unlikely case that one-hundred years was not sufficient, we should allow an updated provisional expiration date.

A further refinement of the tag expiration architecture is to allow multiple tags for each work. Each tag would correspond to one "stick" associated with the work. Since trusted systems can discriminate among different uses such as reading, printing, copying, etc., each of these "sticks" can have a separate tag. For instance, a copyright owner may choose to allow consumers to read freely, but charge them to print. Therefore, the tag expiration for reading would be immediate, but the tag expiration for printing would be some time later.

At the end of the day we should keep the possibility of infinite protection in perspective. Putting aside any limitations thrust on trusted systems for the moment, the infinite protection danger heavily depends on the exclusiveness of trusted systems as the distribution source. If the work were available through other means, then trusted systems would be unable to effectively lock-up the work. The work would no longer be protected by copyright so if it were otherwise accessible, then the trusted systems danger would be thwarted.

To sum up,

Furthermore, in the case when the copyright period depends on the author’s lifetime:

The scope of copyright protection covers only original works. A fact by itself is not eligible for copyright protection. The Constitution provides protection for authors and inventors, not for discoverers or publishers. Facts exist in the world whether or not anyone is aware of them. Hence, a party who discovers and/or publishes a fact is neither an author nor an inventor.

The Court affirmed this view in Feist Publications, Inc. v. Rural Tel. Serv. Co. in denying copyright protection to a telephone directory. Though the Court acknowledged that a compilation of facts may be eligible for copyright protection, the organization or arrangement must exhibit a modicum of originality. Alphabetical order does not pass this threshold – copyright protection for a compilation of facts is "thin."

The Court’s pronouncement was particularly worrisome to database producers. A database is essentially a compilation of data, usually facts. Database producers often invest considerable amounts in the creation and maintenance of their products. Databases are typically stored electronically so that distribution through trusted systems is readily available. Consequently, trusted systems can counter the thin Feist protection.

Currently, database owners typically rely on contractual provisions to protect their investment. Contracts are an inadequate solution, for the provisions only apply to the contracting parties – third-parties are not bound by those terms, so they could conceivably copy the entire database many times over once they obtained access. If the third-party were a competitor, it could offer a competing product at a lower price since it would not need to recoup investment costs. If this became the norm, few, if any, companies would be willing to invest in database products, since adequate protections would not be readily available. Society would be worse off for not having these products available.

In response to the vulnerability of databases to copying, sui generis proposals have been introduced in Congress addressing database protection. H.R. 2652 based database protection on an inappropriate theory that prohibits acts that harm the actual or potential market for the product. H.R. 3531 offered protection to databases for renewable twenty-five-year periods. The renewal turns on whether the database undergoes "any change in commercial significance." In effect, the bill gives an enormous incentive to database producers to update their products; if they do, they can enjoy infinite protection by the law. Both of these bills were defeated in the most recent term, but supporters of the bills have indicated that they intend to reintroduce them in the next session of Congress.

Nevertheless, trusted systems offer a more efficient mechanism for protecting databases. The proposed bills suffer from the same enforcement problems as copyright law does – ex post enforcement as opposed to ex ante prevention. As described in previous sections, trusted systems can police even minor violations, but can eliminate what we consider to be permissible uses. Although a trusted system architecture is an answer to the concerns of database producers, does it properly take into account the concerns of society?

The same reservations we have regarding trusted systems protecting copyrighted works arise in the context of trusted systems protecting databases. These concerns may in fact be more pronounced in this case for we value the free dissemination of facts.

To analyze the effect of trusted systems on the dissemination of facts, we need to begin with a framework. The proposed framework maps the possible uses of a work on a continuum. The continuum is separated into two regions: permissible and impermissible uses. The permissible use region is further divided into free and charged uses. With respect to databases, we need to draw the first boundary between permissible and impermissible uses. As an example, consumers should not be permitted to copy the entire work and sell the copies but should be permitted to extract single data points. As to the second boundary, we need to distinguish between permissible uses that should be free and those that need not be. For instance, a user may be charged for extracting a data point from the database but should not have to pay to use the raw data. These lines between different categories need to be delineated as shown in Figure 1..

Though we cannot rely on copyright law for it is mostly inapplicable to databases, we can draw upon some of its doctrines. Infringement and the limits on the copyright monopoly can shed some light on these boundaries – as analogues to impermissible and permissible gratis uses, respectively.

The difficulty in translating these concepts to a trusted system architecture is due in part to the uncertain boundaries and to the distinctions between qualitative terms to quantitative ones. Specifically, ex ante it is impossible to determine all the uses of the data which would constitute "fair use." Even if this could be determined, could it be reduced to quantitative terms? Since the quantitative rules will not be perfect substitutes for qualitative measures and optimal protection cannot be achieved, we will need to decide whether to err on the side of overprotection or underprotection.

To set the boundary between impermissible and permissible use, we should examine the purpose of protecting databases – protecting the investment in producing the databases. The market effect of database uses is an appropriate distinction to consider. If the current use will directly impact the future market for the database, the use should be prohibited. Otherwise, it should be allowed. Uses that could inevitably harm the database producer consist of substantial extractions of data. Extracting fifty-percent of the data from the database with a single query would constitute a substantial extraction. An extraction of a single entry would not. Somewhere in the middle is the boundary.

We have to divine from the pattern of use, both in terms of quantity and frequency, what constitutes a substantial extraction. In fact, this inquiry largely depends on the character of the database and of the average user. Therefore, the database producer should have flexibility in setting this standard. If the producer overprotects the database, the revenues accordingly decrease. The database producer has the incentive to set it at an appropriate level.

The more significant issue relates to the permissible free uses. The database producer has little incentive to allow free use when it can charge for use. In this area, we must ensure that adequate limitations are placed on the producers in the interest of society as a whole. We should lean more towards underprotection for we want to allow and encourage these uses. These uses correspond to the limits of the monopoly granted by copyright.

First of all, we should recognize the difference between use of the database and use of the data entries. Limitations may be imposed on the use of the database, while the use of the data entries should be without limitations. Restrictions on the use of data entries could amount to the ownership of facts. In the interest of promoting knowledge and learning, society should disallow this. The first sale doctrine becomes largely inapplicable once there is unrestricted use of the raw data.

The uses protected by the fair use concept of copyright – education, scientific progress, parodies, critiques – deserve protection in the database world as well. While parodies and critiques are less relevant in this context, education and scientific progress hold primary importance. The obstacle in protecting these uses in the database context arises in distinguishing them from other uses. Based solely on the pattern of use, educational use may be indistinguishable from commercial use. After all, a database query is a database query. The remaining alternative relies on the status of the user. The approach taken to protect fair use with respect to copyrighted works should be extended to databases. Users associated with institutions such as schools and libraries should be privileged. Students, educators, scientists, and the like should have at least limited free access. This may be accomplished through special user accounts or possibly through a debit card scheme.

Finally, the duration of protection for the database must be examined. As with copyrighted work, trusted systems can allow an infinite protection period. It is even more important that databases not be locked up for that would hinder the free transfer of knowledge. At some point in time, access should be free. Fortunately, we have flexibility in setting the term of protection because copyright law is not operating in the background. The proposed twenty-five year period of protection offered in H.R. 3531 is too generous. Ten years seems a reasonable period to recoup investment costs and make a profit. We should allow database producers to obtain new ten-year protection periods with each new revision of the product. Each revision would be protected by its own ten-year period. Database producers would have tremendous incentives to substantially update their products. If the original version and the version available ten-years later are more-or-less the same, the market demand for the subsequent version could drop to zero. If they are differentiated enough, the effect may be negligible.

We have already concluded that trusted systems are both necessary to the advancement of knowledge and learning, and feasible to develop. If we conclude that information property holders will refuse to make their media digitally available except through trusted systems, we should begin to think about how trusted systems would change our rights when we purchase information. There is a fundamental change in how corporations are legally bound to respect privacy when a transaction occurs. Although content providers must keep communications confidential, they are free to sell almost any information they gain from transaction records. The trusted system could keep records of everything we purchase the rights to read, listen to, or view. "The new copyright management technologies force us to examine anew the sources of [freedom to form … thoughts and opinions in privacy.]"

The exercising of various digital rights could lead to novel invasions of privacy. The system could record whom we exchange intellectual property with when granting transport rights – permissions to copy transfer or loan. With each record the system could add to an international database of everyone’s friends and family. Marketing divisions of major corporations have long participated in the practice similar to logging render rights – permissions to view or print. Corporations formally call it profiling: the process of learning as much about a consumer as possible. They will now be able to record the exact time, date, and extent of every render right granted. Another use of trusted systems will be to give discounts or limited access to members of a club, age group or county. Membership to these groups could be logged by the system and sold to the highest bidder. Northeast Consulting’s primer on Digital Rights Management Technologies lists several corporations’ plans to develop trusted systems. None of the corporations describes outright their protections for privacy. It would be surprising if corporations were touting their technologies in terms of privacy. Market forces encourage content providers to provide the minimum level of privacy protection that consumers will tolerate. "[I]n the market, an individual confronted with electronic contracts of adhesion will be forced to give up control of personal information, even though she would rather not." Worse yet, most users will be unaware of exactly what information they are giving away. Corporations will be unwilling to disclose exactly how personal information will be used. For the same reason corporations will be against "providing a menu of privacy options with the necessary detail to comprehend them, [because this] would draw attention to unsavory privacy practices that the collector may not want to highlight." Under the status quo, Americans will stand to lose significant amounts of personal privacy if they want to access intellectual property through cyberspace.

The belief that privacy is an important right is embodied in the Fourth Amendment’s protection against illegal search and seizure. Most modern constitutions describe privacy as a fundamental right. Justice Brandeis remarked that privacy is "the most comprehensive of rights and the right most valued by civilized men." Americans value privacy because privacy enables them to avoid embarrassment, develop intimacy, and prevent misuse. A classic example of embarrassment occurs when video cameras are put in bathrooms. Although the cameras would be a boon to the adult film industry, Americans would be outraged because of the embarrassing invasion of privacy. The intimacy argument goes as follows: We become more intimate with our friends if we share secrets with them; if there are fewer secrets, there will be fewer intimate relationships. Misuse of publicly available information already occurs. "[M]urderers have tracked down their targets by consulting government maintained address records." Filling in the registration card for a new camcorder could increase the chances of getting an audit. Cyberspace enables the easy collection of far greater amounts of information and therefore puts users at far greater risk of abuse. Unfortunately, legal protections on privacy could stifle the growth of information markets. The accumulation, organization, and sale of personal information are big businesses in America. Although personal information markets may bolster the economy, they could make the general populace worse off. "Individuals may not participate in the National Information Infrastructure for fear that the costs to their privacy will outweigh the benefits."

Even for the larger part of the population that will participate with or without privacy, there are still reasons to respect their privacy. In an online poll 89% of Internet users said they would like to have more privacy. Systematic unwanted observation is in tension with an individual’s dignity. If there is a publicly available record of everything read, there could be a de facto censorship of ideas. Nobody is going to want to read about revolutionary political ideas if they know they could lose their reputation. A lack of privacy in trusted systems could put freedom of association at risk. Consider NAACP v. Alabama where Alabama was denied access to the membership lists of the local NAACP members. With the advance of technology, the oppressive state government could buy the same list from a corporation that collects such information. Limitations on privacy could equate to limitations on freedom of association. Since a "digital copyright system could perform the essential continuing control functions while still preserving reader anonymity" there are few technological reasons to violate user privacy.

There are some cases where the populace is better off with less privacy. Police agencies at all levels could perform their duties more effectively with lower standards on privacy. One interesting solution to this problem is key-escrowed anonymous e-cash. The law enforcement agents could hold a private key to break the anonymity of the e-cash in case of criminal violations. The e-cash would be effectively anonymous for the private sector. Statistical research always benefits from better and greater amounts of information. To maintain a free and vigilant press there needs to be certain restrictions on privacy. The Supreme Court held in, Florida Star v. B.J.F, that if personal information is truthful, lawfully acquired, and of public interest, it may be disclosed absent a state interest of the highest order.

The communications industry and the W3C (World Wide Web Consortium) are spearheading ideas for industry self-regulation of privacy. The W3C claims "products using P3P [Platform for Privacy Protocols] will allow users to be informed of site practices, to delegate decisions to their computer when possible, and allow users to tailor their relationship to specific sites." The P3P encourages web sites to inform users regarding what site operators plan to do with the personal information they are collecting and allows users to make an informed choice of what information they are giving away. This will ultimately "create a platform that is advantageous to both privacy and commerce." These same ideas could apply to trusted systems. Content providers could have a tag that lets users (or their computers) know to what level providers protect privacy.

Content providers could also adopt the other P3P proposals, although it there may not be sufficient market incentives for them to do so. Since content providers have monopoly power and sell to many uncoordinated buyers, providers have total power when contracting for privacy. The government could provide the market incentive for companies to make privacy protection an option. The government can demand, and may even have a compelling security interest to demand, that any companies that contract with the government must protect privacy. The government could go a step further and criminally prosecute firms that fail to protect privacy. Unfortunately, this action might have a chilling effect on the development of new technologies. If the government supports the use of anonymous cash, users could pay for goods anonymously and actively protect their own privacy.

The government does currently protect some aspects of privacy, and may push for further privacy protections. There are many privacy bills currently floating around congress including the Medical Privacy in the Age of New Technologies Act and the Federal Internet Privacy Protection Act. The act regarding medical privacy recommends that Congress "recognize that there is a right to privacy with respect to health information." The act relating to the Internet makes sure no federal agency can make personal information available on the net. It is interesting to note that Congress passed the Video Privacy Act after a reporter searched a Supreme Court Justice Nominee’s video rental records. Many congressmen knew their video rental record would not turn out as "clean" as Bork’s record was, if their own records were released to the public. Once corporations and reporters begin to pry into Congressmen’s personal information, expect stronger privacy laws.

Even if the government does nothing, there could still be some strong market forces to protect privacy. Under Article 25 of the European Data Directive, a personal information "transfer may take place only if … the third country in question ensures an adequate level of protection." The European Community may not allow e-commerce with American companies unless they raise their privacy protections to European standards. The deadline for adhering to the EDD was October 25^th, and fortunately the Europeans have not cut the transatlantic cables yet. Many European countries are not up to EDD standards; when all of Europe is, expect the EU to be less tolerant of America’s poor privacy protections.

We think that the government should foster the same privacy principles outlined by the European Data Directive, but not criminally enforce them as the European Union plans to do. Instead, the government should require that any company that sells the government intellectual property to follow a few basic guidelines. They would be follows:

~ Users should have the right to remain anonymous whenever possible.

At this point in time we think that the government should use only market forces to protect this high standard of privacy. The government should continue to maintain minimum standards of privacy by prosecuting individuals who divulge confidential personal health and financial information. Any governmental action to protect more than the proscribed minimum level of privacy through prosecutions could impede the widespread development of copyright management technologies and should only be considered if there is a compelling government interest.

Among the values that might warrant protection in a world of trusted systems are values associated with freedom of contract. In the physical world, consumers enter into certain types of contracts (the most basic one being the contract for a purchase and sale) with copyright owners or their agents in order to obtain access to copyrighted works. Contract law, however, polices the terms of the contract, and enforces certain types of contracts and clauses, but not others. In particular, adhesion contracts are subject to greater scrutiny by the courts. Trusted systems potentially change the existing environment by permitting the copyright owner to exert greater bargaining leverage over the consumer and by permitting the copyright owner to bypass contract law altogether and enforce certain terms unilaterally, through the technology. Thus, steps may need to be taken to preserve some of the underlying values of contract law in a world of ubiquitous trusted systems.

The basic value underlying contract law is freedom of contract -- that is, the freedom to negotiate for the terms of a contract. Under standard contract law, a necessary condition for a contract to be enforceable is a meeting of the minds, an agreement between the two parties over the terms of the contract. Contract law enforces and encourages such agreements in order to facilitate mutually-beneficial arrangements and private ordering. Where two parties agree to be bound by the terms of a contract, the government will step in and enforce the terms of that agreement, absent certain special circumstances.

Standard-form or adhesion contracts present a special situation. Such contracts are typically rather lengthy and detailed, and are often presented in a take-it-or-leave-it fashion, with no real opportunity to bargain over the terms. An example is a car-rental agreement -- no one really reads all of the fine print and there is no realistic opportunity to bargain over the terms contained in the agreement. As a result, there is a very real risk that in many cases, the consumer might not actually have read the particular terms or had any opportunity to bargain for specific terms. Such contracts would thus appear to undercut freedom-of-contract values mentioned above.

At the same time, however, actually requiring that such contracts always be read and that every term be negotiated would greatly increase the costs associated with entering into the contracts. For example, if you were actually required to read every clause of a car-rental contract and bargain with the car rental company over each term, you would likely go insane. Standard-form contracts may well be quite efficient, in many cases, particularly where they reflect what most consumers would reasonably want to see in the contract. The Uniform Commercial Code is an example of a bunch of standard or default terms that operate in the commercial context, based on commercial practice and designed to reduce bargaining costs.

The courts generally balance these competing concerns by finding standard-form or adhesion contracts generally enforceable, but striking out terms that are "unreasonable" or against public policy. So, for example, a clause in a car-rental agreement imposing liability on the renter for accidents would be reasonable; a clause stating that the renter must hand over his or her first-born child if he or she does not pay would be unreasonable. In this way, the courts permit low-cost contracting through the use of standard-form contracts, but police the terms of such contracts to make sure that they are not unreasonable, unexpected, or against general public policy.

With respect to copyrighted materials, most contracts in the physical world involve a simple purchase and sale, and therefore do not raise many of the concerns mentioned above. You go to a bookstore or a music store and buy the book or compact disc. No complicated terms are involved. Once you acquire possession of the book or compact disc, you can listen to it as many times as you want, lend it to a friend, sell it, etc. The concerns about unreasonable terms or adhesion contracts are not implicated.

In a world of ubiquitous trusted systems, however, there may be reason to be more concerned about contract terms. In a world of trusted systems, copies of copyrighted materials will not be "sold" to consumers outright, the way they are sold in the physical world. Rather, they will be licensed or metered out to consumers in smaller increments. A consumer will pay to play a particular song once, or to print out a copy of a certain article, or to access a particular movie. The basic transaction between copyright owner and consumer will no longer be a purchase and sale. Instead, the relationship will be more complicated and ongoing.

This has three potential effects on concerns about contract law values. First, ubiquitous trusted systems would allow copyright owners to bypass contract law entirely and enforce certain restrictions on use unilaterally, through technology. There would be no need for recourse to the courts. On the one hand, this might raise some concerns, insofar as, unlike with contract law, the courts cannot police the particular terms imposed by the technology. On the other hand, however, the types of restrictions that can be imposed by technology are more limited than those available through contract law, since the technology only restricts access and use. Thus, it is relatively unlikely that trusted systems could be used to unilaterally impose unreasonable terms. So, for example, a trusted system cannot prevent the reader from writing a critical review of a movie, whereas a contract term can attempt to do so. Thus, although trusted systems give copyright owners more control over certain types of use and access, the scope of uses subject to technological control is more limited.

Second, trusted systems may increase the ability of copyright owners to impose restrictions on use through contract by reducing the costs of licensing. In the physical world, it would be unreasonably costly for the copyright owner to negotiate for restrictions on your use of a compact disc when you go to the cash register to purchase it. On-line, however, it is a much simpler thing to impose such conditions in a "click-wrap" agreement that you must pass through in order to gain access to the work. Although such arrangements already exist on the internet, they would be strengthened in a world of trusted systems. Thus, to the extent that trusted systems make such licenses more ubiquitous and attractive to copyright owners, courts may have to be more attentive to possibly unreasonable terms.

At the same time, the reduction in licensing costs may undercut some of the justifications for permitting standard-form contracts in the first place. That is, if standard-form contracts are justified in part because it would be too costly to require actual negotiation of all of the terms of a complicated contract, what happens when the cost is greatly reduced by technology? For example, we could envision systems in which a menu of terms could be embedded in the code of a digital copy. A user’s software could then have certain default settings that would automatically agree to certain terms, reject others, and bring still others, such as non-standard terms, to the user’s attention for specific approval. If such systems in fact develop, then standard-form contracts may in fact become less common in the on-line environment, and the justification for upholding them might be undercut.

Third, the wide deployment of trusted systems may make less obvious to consumers the particular terms of use being purchased. Whereas in the physical context, the terms of use associated with the purchase of a book are quite clear, such terms may be less clear where the use of a particular work is licensed in small increments, particularly where terms of use are imbedded in the code of a digital copy (as the Digital Millenium Copyright Act appears to encourage). Thus, trusted systems may have some impact on the knowledge that consumers have when entering into particular usage agreements.

In light of the above discussion, some measures might be necessary to preserve contract law values in a world of ubiquitous trusted systems. First, although trusted systems do not appear to be able to unilaterally impose the types of unreasonable restrictions that contract law routinely strikes down, the courts should keep an eye open, in case such restrictions do somehow develop. If trusted systems are used unilaterally to impose restrictions that would be found unreasonable under standard contract law, then the courts should forbid such uses. Although it is hard to see, at this point, what such restrictions would look like, the courts should be vigilant. Also, if copyright owners use their control over access to provide selective access in order to affect behavior (e.g. by restricting access to those who write favorable reviews), this should also be forbidden.

Second, attention should be paid generally to ensuring that consumers get accurate information about precisely what they are purchasing. Particularly where trusted systems technology limits the uses of a work invisibly through code, consumers should be aware of such limits prior to making the particular purchase. It may be a good idea to require that particularly unusual restrictions on use be made even more prominent, such as through an extra pop-up window signoff. Where consumers purchase certain uses of copyrighted materials without clear notice of restrictions, courts should be willing to invalidate the terms of the contract.

Third, one possible response might be to attempt to facilitate non-standard bargaining, perhaps by requiring digital copies to contain contact information, such as e-mail, so that consumers could contact the copyright owner and seek non-standard use terms. At the very least, if a consumer has a non-standard use in mind, he or she would at least be able to contact the copyright owner. This measure would turn the labeling provisions of the Digital Millenium Copyright Act into a labeling requirement. It is unclear, however, precisely how helpful this requirement would actually be, since the copyright owner could always refuse to engage in bargaining, and it would be difficult to ensure that good faith bargaining was taking place (i.e. the copyright owner could always set an unreasonably high price).

Fourth, it is probably too early in the day to require imposition of default or compulsory terms on trusted systems, at least for the purposes of contract law. At this point, the concerns raised by adhesion contracts generally do not appear to warrant this kind of intervention. However, over time, it may well be that certain abusive practices crop up, and if so, it may then be appropriate to impose such terms. Moreover, other substantive areas of concern such as copyright law may well result in imposition of such terms in all trusted systems arrangements. Such provisions are discussed in earlier portions of this paper.

This paper has discussed in detail a number of values that could be threatened by the widespread use of trusted systems. Throughout this paper, we argue that the government should proactively seek to preserve these values. Critics, however, will be skeptical of government involvement in the development of such systems, and argue that the market will lead to the development of systems that both producers and consumers want. This claim challenges the relevance of this paper to the future of copyright, so we will address it by considering a few current developments related to trusted systems.

Although the public is not generally aware of the existence of trusted systems, several major companies are both building and using them. A few recent developments in this growing industry offer lessons to consider: 1) copyright technology is becoming more "trusted," and 2) consumers have the opportunity to affect the evolution of the technology.

Technology is becoming more "trusted." This should not be a surprise, since copyright owners are concerned about losing business from piracy. The RIAA recently sued Diamond Multimedia for selling a portable music player called the Rio. The Rio is unique because it is the first portable player to play a certain file format, MP3. The RIAA sued Diamond because most MP3’s are pirated songs, and the Rio encourages people to pirate more. Diamond won the lawsuit, and has now partnered with Liquid Audio, a trusted system for music. This leads us to believe, for music at least, that different standards are converging in a way that gives more control to copyright owners.

Diamond and Liquid Audio are not the only companies that see a profitable future in trusted systems; it appears that Microsoft is also getting involved. In an internal memo Microsoft has announced that the next version of Windows will be a trusted system. Given its power in the market, Microsoft is uniquely positioned to create the standard in trusted systems.

Although trusted systems will probably be used widely, the public interest may not be trampled. Consumers have tremendous power to influence the development of trusted systems. Two recent examples demonstrate consumer’s power; 1) the success of DVD over DIVX and 2) the recent failure of DigiCash.

DVD and DIVX are standards for digital video. DIVX is essentially an extension of DVD that is a simple trusted system. DIVX encourages users to share videos, but to view the disk, the lendee has to pay a few dollars. To watch a DIVX movie, users have to plug a telephone line into their DIVX player, and get permission before watching. DIVX offers consumers the ability to rent a video and not have to return it, since it would be unusable to the consumer after the rental period expired. But DIVX also involved a hassle, and a degree of centralized control and monitoring that consumers objected to. Most consumers preferred DVD over DIVX, and DIVX use has grown slowly as a result.

Electronic cash has also been unpopular, as DigiCash’s bankruptcy in November 1998 illustrates. DigiCash’s problems were not technical adequacies, but rather, a lack of demand. The Economist characterized it as "a solution in search of a problem," and suggests that consumers feel that way also. DigiCash’s failure suggests that consumers are not quick to accept new technologies. This delayed response time could be critical to the preservation of the values outlined in this paper. If it takes a long time before trusted systems are widely accepted, then legislators will have more time to decide on an appropriate response to them. Alternatively, the delayed response time could help consumers learn more about trusted systems, and develop positions on such issues as privacy and fair use.

Finally, we present a summary of our strategy for dealing with potential threats from trusted systems to public interests: