Weinberg

Stanford Law Review
Copyright (c) 2000 The Board of Trustees of Leland Stanford Junior University

May, 2000; 52 Stan. L. Rev. 1251

Hardware-Based ID, Rights Management, and Trusted Systems

Jonathan Weinberg

(Professor of Law, Wayne State University)

I owe thanks to Phil Agre, Karl Auerbach, Lorrie Cranor, Jessica Litman, Neil Netanel, Paul Resnick, and Joel Reidenberg. Earlier versions of this paper were presented at a Haifa University Conference on the Commodification of Information and at the Telecommunications Policy Research Conference. All Internet citations were current as of May 22, 2000. Copyright © 2000 by Jonathan Weinberg and the Board of Trustees of the Leland Stanford Junior University.

IV. Trusted Systems and Common Identifiers

A. Introduction

I want to explore some of the social implications of implementing, on a widespread basis, trusted systems based on identifiers such as the PSN. In that connection, it seems to me that two characteristics of the PSN are notable. First, the PSN is keyed to the holder's identity, rather than his characteristics. It enforces a particular model of trust, in which to learn the characteristics of a particular would-be information recipient, a publisher first ascertains that person's identity and then looks up the characteristics associated with that identity.
This model stands in contrast to a more privacy-protective approach, in which a person can present credentials verifying certain characteristics, such as country of residence, without necessarily disclosing his identity at all. n74 For an example, consider an idea floated by Ira Magaziner in 1998. n75 Magaziner was looking for an answer to one problem presented by Internet anonymity: It undercuts the ability of geographic jurisdictions to tax, because it may not be clear to the merchant and interested governments whether taxes are due and to whom. Magaziner suggested that consumers could make purchases online through the use of "electronic "resident cards'" encoding their country of residence, so that escrow agents could collect taxes associated with that jurisdiction. n76 The proposal was unworkable, but was in one respect privacy-friendly: It contemplated that people would reveal, and carry with them online, a single personal characteristic (their country of residence), without having to reveal any other characteristics. The merchant could learn a buyer's residence, but that information transaction would not reveal his name. The PSN, by contrast, eschews this approach. For the PSN to be used as the basis for a trusted system, the content provider must correlate the PSN with its other data relating to the individual owning that computer, by tying all of that data to the single identifier that the PSN represents.
The second notable characteristic of the PSN is that it is a common identifier. That is, it is well-suited to use by different information collectors in [*1269] unrelated transactions, increasing the ease with which a wide range of information about a person can be aggregated into a single overall dossier. n77 The greatest obstacle to efficient aggregation and manipulation of data today is the need to reconcile inconsistent formats and identifiers; n78 a standard, common GUID can eliminate that obstacle. n79 To the extent that a variety of content providers and other merchants have each collected information tied to individual PSNs, it is a simple matter to compile those files into larger databases.

B. Privacy

Widespread deployment of trusted systems based on such global identifiers will have social consequences. Such systems - in which the user's computer identifies itself during every transaction, to anybody who asks - are pernicious from a privacy perspective. They allow the user to be tracked through cyberspace more easily and thoroughly than is possible under current technology. They have the potential to make the Internet a forum in which database proprietors have what Phil Agre has referred to as a "God's-eye view of the world" - a perspective in which all things have their true names and our Internet representations can straightforwardly be traced back to our real-world identities. n80 Under such an architecture, a much greater [*1270] proportion of ordinary transactions would require consumers to present unique identification numbers that would in turn be digitally linked to a much wider range of personal information. To the extent that collecting identification and assembling dossiers is easy, content providers may do so even when they have no compelling use for the information.
Advertisers and others already see great value in compiling dossiers of personally identifiable information for each of us. Consider, in this regard, the recent Abacus-Doubleclick merger. The combined company announced plans to cross-reference Abacus's database of consumer buying habits, containing real names and addresses and detailed buying information, with Doubleclick's database of consumer Internet surfing and buying habits. n81 Doubleclick targets ads to users, based on "dozens of characteristics, including geographical region, language, and business." n82 It backed off plans to associate its online information about individual consumers with Abacus's personally identifiable offline information only in the face of Federal Trade Commission and state investigations, private lawsuits, and a consumer boycott. n83 The adoption of common identifiers would facilitate the correlation of individual data profiles across databases without public relations headaches.
Systems facilitating the close tracking of content - of what people read, view, or listen to - seem particularly problematic. All of these are the constitutents of human thought. In the analog world, information or entertainment goods are commonly sold on a cash basis, leaving no paper or electronic trail. The copies themselves have no surveillance capabilities, and cannot report back to their makers. The copyright owner, indeed, collects no information about the user at all. n84 Trusted systems threaten to abandon those rules, facilitating the monitoring of individual thought. They raise the specter of the Panopticon, and of subtle and not-so-subtle pressures on individuals [*1271] to eschew socially or governmentally disfavored information goods. n85

C. Communications Policy

A second set of consequences relates to the effects of this technology on the economics and politics of content markets. To begin with, all technological measures protecting digital content (of which trusted systems are a subset) raise an important set of issues typically associated with intellectual property law. Such measures allow sellers of entertainment or information to assert effective control over uses that are privileged by intellectual property law, and over subject-matter that is assigned by intellectual property law to the public domain. That is, they enable sellers to exercise control notwithstanding intellectual property law's judgment that society in those circumstances is best served by free use of the material by the public at large. n86 Others have written cogently about these points, and I will not linger long on them here. n87 Instead, I want to raise broader concerns, which I believe resonate less with those of intellectual property law, and more with those of communications policy generally. n88
[*1272] It is useful here to review two capabilities that trusted systems based on common identifiers give to sellers of entertainment or information. First, the reliable identification of the would-be consumer, together with any other information the seller can collect describing that would-be consumer, helps give the seller individualized information about each member of its target audience. Second, the access and copy protection capabilities of the system mean that the seller has much stronger control over which consumers get access to the work; the seller therefore can sharply limit any secondary market in the work.
In the Old Way of Doing Things, technical inefficiencies made it difficult to disseminate speech to a dispersed but tightly controlled group of folks. There was always some leakage: If you wanted to disseminate speech, you had to give up some control over its dissemination. For example, once a content owner distributed a copy of a work, it had no technological means of preventing the owner of that copy from selling, loaning, privately displaying, or giving away the copy as he chose. n89 And the copyright law's "first sale doctrine" denied content owners the ability to impose such restrictions within the four corners of the copyright law. n90 These limitations on content owners' effective rights helped democratize access to content. They allowed gratis redistribution of, and secondary markets in, copies of the works. Though content owners have complained that the Internet threatens to divest them of any control over the distribution of their works, trusted systems threaten to eliminate even the small avenues for royalty-free redistribution that exist in the nondigital world. n91
It is worth lingering on the role of sharing, and other forms of redistribution, in the nontrusted system world. "Small-scale, decentralized reproduction of intellectual property" has long been a fact of life in markets for information, entertainment, and computer software. n92 People copy music tapes and CDs for themselves, family, and friends; they photocopy magazine articles; they allow family and friends to use, and copy, their computer software. They persist in doing so, notwithstanding the best efforts of the copyright industries to convince them that it is illegal, largely because they find it [*1273] hard to believe that this is something the law does or should proscribe. n93 And that system seems to work - at least, it has not obviously injured in any palpable way producer incentives to create intellectual property.
The nontrusted-system world is also characterized by a lot of sharing, in the vernacular sense, that does not implicate copyright law. As I noted above, people lend each other analog copies of protected works, and read, watch, or listen to works they have borrowed, all without implicating the copyright laws at all. At least in a static analysis, n94 both of these sorts of sharing are good, since they increase the distribution of information and thus social benefit without any social cost. n95 Put another way, sharing allows the distribution of information at the optimal demand price, because that price is equal to the marginal cost of distribution, which in this case is close to zero. n96 The most successful institutions in American life today that are based on such sharing are public libraries, which were established precisely to enable large-scale sharing of analog works.
The trusted-system world could involve a number of changes to this status quo. The content provider's enhanced control over access to the work could constrain both types of sharing I have described above. It would allow content providers to sharply limit the small-scale copying of intellectual property that has become both accepted and commonplace among consumers today, but that producers contend violates their copyrights. It could also greatly limit the small-and large-scale lending and borrowing of intellectual property that takes place today and is unimpeachably consistent with the copyright laws. After all, as noted above, the control given content providers by trusted systems does not rest on whether the content provider can assert intellectual property rights in the work, or whether a particular use of the work by a consumer would violate those rights.
The trusted-system world also seems well-suited to facilitate discrimination on the part of the content provider, a change with more ambiguous results. Most obviously, trusted systems will facilitate price discrimination - that is, the content provider can ask different consumers to pay different prices, unrelated to the provider's own costs. That is not the way markets typically work: Most commonly, a producer sets a uniform price, which each consumer chooses to pay or not pay. Sellers cannot effectively engage [*1274] in price discrimination unless three conditions are met. n97 First, the seller must be able to prevent (or limit) arbitrage - it must ensure that buyers who paid a low price do not turn around and resell the information or entertainment to someone who would otherwise be willing to pay the content provider the higher price. Otherwise, any attempt by the producer to partition the market would be unavailing. Trusted systems make this possible by greatly enhancing the content provider's ability to control any redistribution of the work.
Second, the seller must have market power. All copyright owners have some degree of market power because of the legal control that intellectual property law gives them; that is one reason they are able to charge prices in excess of marginal cost. Some, naturally, have more market power than others, based on the demand for the work and the availability of near-substitutes.
Finally, the seller must be able to set prices in a way that in fact reflects consumers' willingness to pay. Trusted-system technology can make this possible in two ways. First, as noted above, a trusted online architecture based on global user or platform identifiers will allow content providers to tie each consumer to a wide range of personally identifiable information. For example, when the consumer presents her PSN to gain access to a digital work, the content provider will be able to pull up other information associated with that PSN in order to make a judgment about the particular consumer's willingness to pay. n98 Alternatively, the content provider can shift its payment model from the "sale" model prevalent today, in which the consumer buys a copy of the work and can then read, listen to, or watch that copy an unlimited number of times without further payment, to a "pay-per-read" system in which the customer pays a smaller amount on each occasion that she reads, listens to, or watches the work. This allows the content provider to collect more money from those customers who want to use the work many times and presumably are willing to pay more for that ability, and less from those who want to view the work only once. The difference between those prices is largely unrelated to the content provider's own costs. n99
Is this price discrimination a good thing or a bad one? Some have argued that it is beneficial. n100 Price discrimination in information goods is socially [*1275] useful, the argument runs, because it increases the distribution of information. Without price discrimination, the content provider must charge a single market price, and people unwilling to pay that price will be shut out of the market entirely. If the content provider can engage in price discrimination, by contrast, it can charge every consumer the exact price that she is willing to pay, thus simultaneously maximizing profits and maximizing the number of people who will be exposed to the information and entertainment in question. "We can say with confidence that many more consumers [will benefit] from the author's creation." n101
The matter, though, is not nearly so straightforward. Price discrimination is unquestionably good for producers since it converts consumer surplus into producer profits. But, as a general matter, whether price discrimination increases overall welfare is a more difficult question, resting on the facts of each case. n102 An increase in the distribution of the good is a necessary, but not a sufficient, condition for increasing total welfare. n103
More to the point, in thinking about whether the price discrimination enabled by trusted systems would be a good thing, we need to ask the question, "Compared to what"? One of the key reasons that trusted systems enable price discrimination is that they sharply decrease sharing; they are designed to eliminate any redistribution of the information good beyond the control of the content provider. That is, price discrimination allows the sale of information to consumers willing to pay less, but at the expense of cutting off existing means, through sharing and secondary markets, of getting the information or entertainment at low or no cost to some of those same consumers. Indeed, from a static perspective sharing is a more efficient way of allowing the market to reach those consumers, since it makes the good available to them at a price more nearly approaching the zero marginal cost of supplying it to them.
Secondary markets, involving redistribution of information goods after their first sale and outside the control of the initial seller, can do the same job as price discrimination of getting information goods at lower prices to lower-valuation users. That is what used bookstores are all about. The price discrimination that trusted systems may facilitate therefore may not increase the number of consumers getting the good at all; it may simply ensure that the low-valuation consumers receive the good from the initial seller rather than someone else. n104 Further, it does so at the distributional cost of shifting all surplus away from consumers.
[*1276] The points I have made so far are open to a variety of counterarguments. First, it might be argued that price discrimination will do a better job of getting the information or entertainment to low-valuation users, many of whom may not have the opportunity to gain access to the work through resale or sharing. The copyright-law implications of issues surrounding small-scale redistribution, even when no new copies are created (aside from the RAM copies that are associated with any invocation of a digital work), are hotly debated within the legal community. Large-scale copying and redistribution of digital works is illegal even when no price is charged. n105 Yet public libraries, at least, are set up precisely for the purpose of getting free information works to users unwilling to pay the price set by the market. Consumers interested in viewing a work only once and willing to wait until it is available can borrow from the library; those interested in viewing the work multiple times are more inclined to buy it. This is precisely the sort of result price discrimination is supposed to achieve.
By contrast, it is unclear to what extent price discrimination in practice can achieve the advantages theory promises for it. It is difficult to gauge consumer preferences precisely, and publishers are unlikely to drop prices too far based on guesses about a particular class of consumers' willingness to pay. While theoretical perfect price discrimination promises perfectly efficient markets, real-world third-degree discrimination will fall short of that ideal, as publishers group consumers by second-best proxy characteristics and attempt to set prices for each group. Nor will consumers easily accept second-degree discrimination: The splashy failure of DIVX, a pay-per-view movie format, should give rise to some doubt about the enthusiasm with which ordinary folks will embrace usage-based prices for digital works. n106
Next, one might argue that if sharing were technologically disallowed then market prices would fall. Without the possibility of sharing, the argument runs, information goods are not as valuable to purchasers. Yet when the content provider must set a single market price, it cannot easily raise that price to take into account the benefits of sharing, because different buyers will place significantly different values on the ability to share (and will in [*1277] fact share with markedly different numbers of people). n107 Moreover, the existence of leakage also acts to constrain prices, by providing a near-substitute for the purchased good. n108
Finally and most obviously, one might argue that this analysis overlooks the dynamic impact of sharing and the nature of secondary markets in digital works: Sharing and resale do not generate revenues to the content provider, so they do not provide incentives to stimulate production. More baldly, one might argue that my discussion is in essence an argument for piracy - which will certainly lower prices to the consumer, but at the cost of diminishing incentives to produce. Secondary markets in the digital world, the argument runs, may involve large numbers of illegal perfect copies. Sale of those copies cuts directly into the profits, and thus the incentives, of the initial producer.
I do not contest that producer incentives are necessary. Publishers must be able to sell information goods at a price sufficiently above marginal cost, and for a sufficiently long period of time, to recover their fixed (first-copy) costs. Otherwise, they would lose money. To that end, there must be sufficient entry barriers limiting other folks' ability to sell those works as cheaply. We do not know, though, how much in the way of incentives producers need. n109 Ordinary economic theory suggests that publishers will invest so long as they expect profits, taking into account normal rates of return. If publishers have adequate incentives even without the extra rents that price discrimination gives them, then we may get a better social result by reaching lower-valuation users through secondary markets, sharing, or even some degree of piracy than through the increased control that trusted systems bring. n110
Further, there is a connection between media concentration and the power of information providers to identify consumers and to thus discriminate. To the extent that sellers' ability to price discriminate will rest on their access to personally identifiable information about buyers, publishers with access to those databases will have a competitive advantage over those who [*1278] do not. This may have two negative effects. First, it will tend to concentrate media markets - and, to the extent those markets are characterized by winner-take-all dynamics, n111 will help determine who those winners are. Second, it will increase the value of the dossiers, and thus increase the commercial pressure on privacy.
The control facilitated by trusted systems and common identifiers may allow other sorts of discrimination as well. Most generally, it will increase producers' ability to pick and choose who will be allowed to view or read particular works. Given the power of a common identifier such as Intel's PSN to facilitate the association of a wide range of information with a given personal identifier, producers could in theory use these tools to allow access to a speech work only by persons who live in preferred zip codes, have certain levels of family income, or are white. There may be only limited circumstances in which a mass marketer of entertainment and information would have an incentive to do so: Most obviously, perhaps, a publisher might discriminate because of ideological motivations, or if particular content gained cachet from only limited distribution. From a free speech and communications policy standpoint, though, it seems disturbing to see extensive social investment in a technology built around the ability to prevent the movement of speech and information to the public at large.
For the most part, today, content producers and consumers share control over the uses and dissemination of speech works. Content producers have extensive control by virtue of their ability to produce and license the technological artifacts (such as film reels) embodying those works, reinforced by the rights granted them by the copyright law. Consumers have some control as well, by virtue of their own abilities to use, copy, and manipulate such works in ways that the copyright law either does not forbid or expressly privileges, n112 or in ways that have been effectively immune from copyright enforcement. And because these are speech works, that distribution of control has political consequences. It shapes the overall movement of information and expression within society. The rise of trusted systems based on common identifiers would shift that control. n113

FOOTNOTES:

n74. See Agre, supra note 7, PP 18, 35; David Chaum, Security Without Identification: Transaction Systems to Make Big Brother Obsolete, 28 Comm. of the ACM 1030, 1030 (1985); Lessig & Resnick, supra note 1, at 412-13. Part V infra discusses this in greater detail.
n75. Ira Magaziner, at the time, was the President's Senior Internet Advisor.
n76. See Internet Taxation System is Mulled by White House, Wall St. J., Sept. 11, 1998, at B4.
n77. See David Chaum, Achieving Electronic Privacy, Sci. Am., Aug. 1992, at 96 <http:// ganges.cs.tcd.ie/mepeirce/Project/Chaum/sciam.html>.
n78. For an example of an attempt to gain the benefits of such standardization, consider the RosettaNet specification, under development by a consortium of information technology companies. See RosettaNet, Executive Overview <www.rosettanet.org/general/overview.html>. The point of the RosettaNet project is to build an XML-based language of common data descriptors and business processes to streamline electronic business-to-business transactions. See, e.g., Ellis Booker, XML Greases Supply Chain, Internet Week, Aug. 23, 1999, at 1 <http://www.techweb.com/ se/directlink.cgi?INW19990823S0001> (describing the current status of XML-driven supply chains). Yet an initial draft of the RosettaNet specification raised controversy in part because of the ease with which such standardization allows the sharing of personally identifiable information: The specification directed vendors to provide the purchaser's name and address to every company involved in the item's production. See James Glave, RosettaNet: Nothing Personal?, Wired News (Sept. 10, 1999) <http://wired.lycos.com/news/news/technology/story/21699.html>; James Glave, The Killer Consumer Gossip App, Wired News (Sept. 10, 1999) <http://wired.lycos.com/news/ news/technology/story/21668.html>.
n79. See Graham Greenleaf, "IP, Phone Home": ECMS, (c)-tech, and Protecting Privacy Against Surveillance by Digital Works 7-8 (1999) (paper presented to the 21[su'st'] International Conference on Privacy and Personal Data Protection held September 13-15, 1999, in Hong Kong) <http:// www2.austlii.edu.au/<diff>graham/publications/ip<uscore>privacy> ("The success, importance and danger of ECMS [electronic copyright management systems] is likely to depend in large part on the extent to which they achieve interoperability between multiple publishers (within one ECMS), and ultimately, between different ECMS and different media types.").
n80. Agre, supra note 7, PP 23, 26. Agre, I should note, disclaims authorship. See Email from Phil Agre to the author (May 10, 1999) (on file with the author). He cites Donna Haraway's reference to the "god-trick" in connection with the postulated ability to see like a God from a position transcendent and outside of lived experience, see Donna J. Haraway, Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective, in Simians, Cyborgs, and Women: The Reinvention of Nature, 183, 193 (1991), and Edwin Burtt's much earlier reference to a similar perspective in The Metaphysical Foundations of Modern Physical Science: A Historical and Critical Essay (1925). Neither author developed the idea, though, in relation to computer representations of identity, much less privacy policy.
n81. See Courtney Macavinta, DoubleClick, Abacus Merge in $ 1.7 Billion Deal, CNET News.com (Nov. 24, 1999) <http://news.cnet.com/news/0-1005-200-1463444.html>; Courtney Macavinta, Privacy Advocates Target Abacus Shareholders, CNET News.com (June 29, 1999) <http://www.news.cnet.com/news/0.1005-200-344244.html>; Courtney Macavinta, Privacy Fears Raised by Doubleclick Database Plans, CNET News.com (Jan. 25, 2000) <http://news.cnet.com/ news/0-1005-200-1531929.html>.
n82. See Doubleclick, Annual Report-Overview: Making Internet Advertising Work <http:// www.doubleclick.net/annualreport/overview.htm>.
n83. See Diane Anderson & Keith Perine, Privacy Issue Makes Doubleclick a Target, Standard (Feb. 3, 2000) <http://www.thestandard.com/article/display/0,1151,9480,00.html>; Jeri Clausing, Michigan Moves Against Doubleclick, Cybertimes (Feb. 19, 2000) <http://www. nytimes.com/library/tech/00/02/cyber/articles/18doubleclick.html>; Bob Tedeschi, In a Shift, Doubleclick Puts Off Its Plan for Wider Use of the Personal Data of Internet Consumers, N.Y. Times, Mar. 3, 2000, at C5.
n84. See Greenleaf, supra note 79, P 8.
n85. See Julie E. Cohen, Some Reflections on Copyright Management Systems and Laws Designed to Protect Them, 12 Berkeley Tech. L.J. 161, 184-85 (1997).
n86. Given its enactment of the Digital Copyright Millennium Act of 1998, Pub. L. No. 105-304, 112 Stat. 2860 (1998), which criminalizes the act of "circumventing" such technological protections, Congress appears not to share that concern. See generally Pamela Samuelson, Intellectual Property and the Digital Economy: Why the Anticircumvention Regulations Need to be Revised <http://www.sims.berkeley.edu/<diff>pam/papers/Samuelson<uscore>IP<uscore>dig& usc ore;eco<uscore>htm.htm> (criticizing the anticircumvention rules).
n87. See generally, e.g., Symposium, Intellectual Property and Contract Law in the Information Age: The Impact of Article 2B of the Uniform Commercial Code on the Future of Transactions in Information and Electronic Commerce, 13 Berkeley Tech. L.J. 809 (1998) (collection of articles critically examining proposed Article 2B of the Uniform Commercial Code); James Boyle, Foucault in Cyberspace: Surveillance, Sovereignty, and Hardwired Censors, 66 U. Cin. L. Rev. 177 (1997) (explaining that legal rules barring circumvention of technological protection measures clothe both the state and content providers with power they would not otherwise have); Julie E. Cohen, Copyright and the Jurisprudence of Self-Help, 13 Berkeley Tech. L.J. 1089 (1998) (arguing that self-enforcing digital contracts grant publishers absolute control inconsistent with copyright and First Amendment principles); Mark A. Lemley, Dealing with Overlapping Copyrights on the Internet, 22 U. Dayton L. Rev. 547 (1997) (emphasizing the broad scope of rights granted to copyright holders by a rote application of the existing statute to Internet technology); Lawrence Lessig, Intellectual Property and Code, 11 St. John's J. Legal Comment. 635 (1996) (suggesting that technological protection of intellectual property can make legal protections irrelevant, and thus privatize law); Jessica Litman, Reforming Information Law in Copyright's Image, 22 U. Dayton. L. Rev. 587 (1997) (pointing out the dangers to important information-policy concerns presented by the intersection of copyright law and Internet technology); Greenleaf, supra note 79 (observing that technical protections of intellectual property may obviate public interest protections in intellectual property law).
n88. Cf. Kenneth W. Dam, Self-Help in the Digital Jungle, 28 J. Legal Stud. 393, 395-97 (1999) (suggesting that copyright scholars, precisely because they tend to approach technological protection systems from an intellectual property standpoint, have been insufficiently appreciative of the virtues of technological protections).
n89. See Litman, supra note 87, at 600-01.
n90. See id.
n91. See note 12 supra and accompanying text.
n92. Stanley M. Besen & Sheila Nataraj Kirby, Private Copying, Appropriability, and Optimal Copying Royalties, 32 J.L. & Econ. 255, 255 (1989); see also Yannis Bakos, Erik Brynjolfsson & Douglas Lichtman, Shared Information Goods, 42 J.L. & Econ. 117 (1999) (investigating how consumer sharing of information goods affects seller profits); Litman, Copyright Noncompliance (Or Why We Can't "Just Say Yes" to Licensing), supra note 17; Michael J. Meurer, Price Discrimination, Personal Use and Piracy: Copyright Protection of Digital Works, 45 Buff. L. Rev. 845, 852-56 (1997) (discussing the economic impacts of unauthorized sharing).
n93. See Litman, Copyright Noncompliance (Or Why We Can't "Just Say Yes" to Licensing), supra note 17, at 252-53; Jessica Litman, Revising Copyright Law for the Information Age, 75 Or. L. Rev. 19, 40-41 (1996).
n94. But see text following note 108 infra (discussing the dynamic impact of sharing).
n95. See Kenneth J. Arrow, Economic Welfare and the Allocation of Resources for Invention, in The Rate and Direction of Inventive Activity: Economic and Social Factors 609 (1962); Yochai Benkler, Free as the Air to Common Use: First Amendment Constraints on Enclosure of the Public Domain, 74 N.Y.U. L. Rev. 354, 424 (1999).
n96. See Benkler, supra note 95, at 424 & n.273.
n97. See Meurer, supra note 92, at 870.
n98. Economists refer to this as third-degree price discrimination. See W. Kip Viscusi, John M. Vernon & Joseph E. Harrington, Jr., Economics of Regulation and Antitrust 290-91 (2d ed. 1995).
n99. This falls within a category that economists refer to as second-degree price discrimination. See id. at 249-55, 290-91.
n100. See, e.g., Fisher, supra note 5 (arguing that price discrimination makes information products available to a wider range of consumers). But see Wendy J. Gordon, Intellectual Property as Price Discrimination: Implications for Contract, 73 Chi.-Kent L. Rev. 1367 (1998) (querying Fisher's premises).
n101. Fisher, supra note 5, at 1239.
n102. See Viscusi et al., supra note 98, at 290-95; Meurer, supra note 92, at 896-98.
n103. See Viscusi et al., supra note 98, at 293-95; Meurer, supra note 92, at 898.
n104. See Gordon, supra note 100, at 1378-89.
n105. See No Electronic Theft (NET) Act of 1997, Pub. L. No. 105-147, 111 Stat. 2678 (codified at, inter alia, 17 U.S.C. 506-07 & 18 U.S.C. 2319-2320); William McCall, College Student Convicted of Piracy, AP Online, Aug. 21, 1999, available in Westlaw, WL 22036202 (describing the first conviction under the 1997 law).
n106. The DIVX plan was that a user would purchase a videodisk for $ 4-5, and have free access for 48 hours after the first play. After that time, the user would pay a fee for every subsequent viewing; those viewings would be purchased through a central server connected to the DIVX player by telephone line. See David Dranove & Neil Gandel, The DVD vs. DIVX Standard War: Network Effects and Empirical Evidence of Vaporware 8 (Tel Aviv Univ. Eltan Berlgas School of Economics Working Paper No. 14-99, 1999). DIVX was discontinued, for lack of consumer interest, in June 1999. See Carl Laron, Of Edsels and DIVX, Electronics Now, Sept. 1, 1999, at 2.
n107. Bakos et al., supra note 92, engage in a much more sophisticated analysis of this phenomenon. The authors conclude that sharing decreases producer profit when the diversity in team size, defined as the number of consumers sharing any particular copy of the good, exceeds the diversity in individual consumer valuations. They note that seller profit is enhanced if high-valuing consumers tend to share with low-valuing consumers rather than with each other, and if low-valuing consumers tend to share with a greater number of people than do high-valuing consumers. Both of these contingencies will tend to "even out" the value that each team places on the good, and thus will allow the producer to reflect that value more nearly in its selling price. Id. at 122-27.
n108. See Benkler, supra note 95, at 433 n.302.
n109. See Litman, The Exclusive Right to Read, supra note 17, at 44-46.
n110. See Bakos et. al., supra note 92, at 148 ("Profitability and social efficiency need not go hand in hand: sharing can be profitable [for content providers] in situations where it is not efficient, and efficient in situations where it is not profitable.").
n111. See note 68 supra.
n112. See, e.g., 17 U.S.C. 1008 (privileging consumers' noncommercial use of digital audio recording devices, digital audio recording media, analog recording devices, and analog recording media for making musical recordings); Recording Indus. Ass'n v. Diamond Multimedia Sys., 180 F.3d 1072 (9th Cir. 1999) (declining to enjoin the manufacture and sale of the Rio portable music player, which plays downloaded MP3 files).
n113. See generally Lawrence Lessig, Code and Other Laws of Cyberspace 154-56 (1999) (raising equity-based concerns).