Stopping Digital Copyright Infringement Without Stopping Innovation[1]

Mark A. Lemley[2] & R. Anthony Reese[3]

 

            Suing actual infringers is passé in copyright law.  In the digital environment, the real stakes so far have been in suing those who facilitate infringement by others.  Copyright owners tend not to sue those who trade software, video, or music files over the Internet.  Indeed, such suits are so rare that the RIAA’s recent announcement that it would sue some actual infringers sent shock waves through the legal community.  Instead, copyright owners sue direct facilitators like Napster,[4] makers of software that can be used to share files,[5] those who provide tools to crack encryption that protects copyrighted works,[6] search engines that help people find infringing material,[7] and quasi-ISPs like eBay or Yahoo Auction.[8]  All of these suits rely on theories of secondary liability, focusing on those who provide services or write software that can be used in an act of infringement.[9]

 

In addition, there is a new theory of what might be called “tertiary” liability that seeks to reach those who help the helpers.  Cases in this vein include lawsuits filed against those who help others crack encryption, for example by providing links to software that can be used to crack encryption,[10] the copyright lawsuit against backbone providers for providing the wires on which copyrighted material flows,[11] the claims filed against the venture capital firm of Hummer Winblad for its role in funding Napster,[12] and (with an unusual twist) the malpractice suit against the law firm of Cooley Godward for advising mp3.com that it could assert defenses to copyright infringement.[13]  The anticircumvention provisions of the Digital Millennium Copyright Act (DMCA) provide for one particular type of tertiary liability (for providing tools that circumvent encryption protecting a copyrighted work and that can be used to infringe the work’s copyright),[14] and there have even been suggestions that there should be a new tort of contributory violation of the DMCA’s anticircumvention provisions, which should perhaps be termed quaternary liability for copyright infringement.[15]

 

            Further, a number of doctrines that were designed to protect these secondary and tertiary “facilitators” – the “safe harbor” for online service providers,[16] the restrictive standard for contributory copyright infringement for equipment providers announced by the Supreme Court in the Sony Betamax case,[17] and the requirement that vicarious infringement be limited to cases of direct financial benefit[18] – are under attack.  The ALScan and Perfect 10 cases undo much of the benefit of section 512’s protection for Internet service providers.[19] Napster rewrites the rule of Sony in a way that significantly limits its application.[20] And both Napster and Fonovisa have all but eliminated the requirement of direct financial benefit in vicarious infringement.[21]  It’s also worth noting that the doctrine of the corporate veil, which protects investors from liability in most areas of law, appears not to function in copyright.[22]  And proposed legislation would go even further in regulating the behavior of those who do not themselves infringe, injecting Congressional oversight into how software and consumer electronics are built[23] and permitting content owners to unleash destructive hacks of computer networks without fear of liability.[24]

 

            There is of course a good reason copyright owners are filing such suits.  They see themselves as under threat from a flood of cheap, easy copies and a dramatic increase in the number of people who can make those copies.  The high volume of illegal uses, and the low return to suing any one individual, make it more cost-effective to aim litigation at targets as far up the chain as possible.  From the perspective of the music industry, it’s easier and more effective to shut down Napster than to sue the millions of people who traded files illegally on Napster.  So far, the courts have been largely willing to go along, shutting down a number of innovative services in the digital music realm, though refusing to ban the provision of peer-to-peer (p2p) software by Streamcast and Grokster.

 

In this article, we focus on one strand of these cases against those who allegedly facilitate copyright infringement – those dealing with distribution of digital content over p2p networks.  We argue that unrestricted liability for anyone who is in any way involved with copyright infringement is a bad idea.  Indirect liability is a continuum, in which acts most closely related to infringement and with the fewest affirmative benefits are the easiest to condemn.  Napster was a relatively easy case for liability, because the service was limited to trading music files and virtually all of the files actually traded were illegal.  The Grokster case is a substantial step further removed from infringement, both because the defendants’ involvement is less (indeed, resellers like Grokster are arguably merely conduits for providing software, an activity which should be legal under most circumstances)[25] and because the noninfringing uses of KaZaA and similar software involved in the case are greater.  Lawsuits against ISPs, search engines, telephone companies and other indirect providers, while not the focus of our attention here, are even more problematic because of the many legal uses of these services.  The key policy point is that going after makers of technology for the uses to which their technologies may be put threatens to stifle innovation.  Similarly, going after necessary third parties like investors and law firms will stifle investment in innovation.  The fundamental difficulty is that while courts can make decisions about direct infringement on a case-by-case basis, lawsuits based on indirect liability necessarily sweep together both socially beneficial and socially harmful uses of a program or service, either permitting both uses or condemning both. 

 

A middle ground has so far largely been lacking in this debate.  Our aim in this article is to seek such ground.[26]  Optimal digital copyright policy with respect to p2p networks would do two things: stop deterring innovators, and permit cost-effective enforcement of copyright in the digital environment.[27] Economically, one can estimate the cost to society from enforcement of the indirect liability rules against p2p providers as a function of the legal uses that that law effectively forbids, plus the foregone efficiency of the p2p distribution mechanism relative to industry-driven distribution of copyrighted content, plus the social value of foregone innovation that results from deterring would-be innovators.  If we compare this cost to the benefits accrued by giving digital copyright owners another, more convenient forum to sue, it is not at all clear that the benefits of the new, expanded indirect liability rules exceed the costs in most cases. 

 

            Moreover, we need not make this difficult tradeoff at all unless it is clear that copyright owners have no effective alternatives to suing facilitators.[28]  The basic economics of copyright enforcement, however, do suggest alternative approaches.  It is not currently cost-effective for copyright owners to sue individual infringers, because there are tens of millions of them, because lawsuits are expensive, and because many infringers would only be liable for (or able to pay) minimal damages.  Copyright owners are happy to sue facilitators instead, because there are fewer of them and both damages and the benefits of injunctive relief are substantial. Copyright owners have no incentive to permit optimal innovation by facilitators, because they do not benefit from that innovation except indirectly. Individual infringers in turn have no incentive to change their behavior or to subscribe to fee-based services, because they suffer none of the costs of infringement.  In this paper, we suggest three possible alternatives that might provide ways out of the digital copyright morass. 

 

            One solution is to change the incentives of individuals potentially engaged in copyright infringement.  Because individual users of p2p networks know that it is extremely unlikely they will be sued, economic theory suggests that the only way to effectively deter infringement is to increase the effective sanction substantially for those who are caught.[29]  Were the government to criminally prosecute selected users of peer-to-peer services, or were copyright owners to sue those users and obtain extremely large monetary judgments, we suspect it could have a substantial deterrent effect on many illegal users.  Selective enforcement has other advantages as well – the suits could target the relatively few keystone providers of illegal files on p2p sites, and those are precisely the users whose activities are most likely infringing.  While particular prosecutions won’t stop illegal file trading altogether, copyright owners have never been able to prevent all piracy.  All they need to do is reduce piracy enough that they can make a return on their investment.

 

Another solution is to change the incentives for copyright owners to pursue individual infringers by reducing the cost of such enforcement.  One such approach would be a levy system of the type proposed by Neil Netanel.[30]  Levies on equipment or services have the virtue of permitting automatic collection of royalties, reducing the enforcement cost dramatically, but at the cost of taxing legal as well as illegal uses.  A levy solves the enforcement problem at the front end, but as with the current approach of suing facilitators, it imposes the costs of copyright enforcement on innovators.  The main difference is that under a levy system the copyright owner is protected by a compulsory license rather than a property rule. 

 

A second way to reduce the cost of enforcement is to create some sort of quick, cheap arbitration system that enables copyright owners to get some limited relief against abusers of p2p systems.  The existing domain name trademark arbitration system is a model in some respects – its speed and low cost -- but a cautionary tale in others – its lack of due process protections.[31]  Digital copyright law also differs in some significant ways from the law governing domain names, and the design of a private arbitration system would have to reflect those differences.  One way to design this system is not to mandate it, as ICANN does for domain name registrants, but to offer a safe harbor to any intermediary that requires its users to participate in the system.  Once users have registered and agreed, they could be challenged by copyright owners if they are uploading (not just downloading) copyrighted files.  The system could also be designed to improve accuracy relative to the binary choice the courts face in indirect infringement cases today.  We could design the system so that it is limited to "clear cases" – say uploading of more than 10 files to a network in a one-week period.  We could also build in a defense for arguable fair uses, so that a user who could prove she was uploading only out-of-print works or that she was space-shifting CDs she already owns might have a defense.[32]  Such a system would permit low-cost enforcement of copyright infringement against direct infringers, reducing the need for content owners to sue facilitators.  Relative to levies, an arbitration system would trade off some increase in cost for accuracy, targeting only those making illegal uses rather than all users of computers or p2p networks.  It would be more fair than selective criminal or civil prosecution, because the burden of paying the penalty for infringement would fall more evenly on each wrongdoer, rather than imposing stark punishment on a few in order to serve society’s interest in deterring the rest. 

 

            None of these approaches is perfect.  Each has its advantages and disadvantages, and is likely to work better in some contexts than in others.  But it is clear that something must be done to escape the current linkage between stopping copyright infringement over p2p networks and stopping innovation.  The economics of copyright enforcement suggest two basic types of alternatives – raising the cost of direct infringement, or lowering the cost of suit.  Pursuing a combination of these approaches—selective enforcement, levies, and an arbitration safe harbor—is preferable to the status quo.

 

            In Part I, we make the case that there has been a seismic shift in copyright infringement in the digital environment, away from suing direct infringers and towards suing facilitators with less and less connection to the act of copyright infringement.  Our discussion in this part focuses on issues relating to p2p networks, though these cases are part of a broader trend towards suing facilitators rather than direct infringers online.  In Part II, we examine the economics of digital copyright infringement.  This part explains why copyright owners are suing facilitators, why doing so is bad for society, and outlines the possible alternatives at a theoretical level.  Part III makes those alternatives more concrete by applying them to the problem of infringement over p2p networks.  Section III.A explores how a system of criminal prosecution of, or severe civil penalties against, high-volume uploaders might work and discusses its likely consequences.  Section III.B evaluates the pros and cons of a p2p levy system and proposes an additional alternative: a safe harbor for p2p facilitators who agree to implement an arbitration system designed to stop digital piracy.  Part III also discusses the limitations and potential problems of these approaches.  We conclude that implementing a combination of these strategies may offer copyright owners effective protection without unduly hampering innovation in p2p networks.

 

I.          Suing Facilitators

 

            Suits seeking to hold someone other than a direct infringer liable for copyright infringement are not new.  Although the Copyright Act throughout the twentieth century was essentially silent on the issue of liability for anyone other than a direct infringer, courts read the statute as imposing such liability in certain circumstances, and Congress  endorsed that view in the 1976 Act.[33]  Two doctrines of secondary liability have emerged in copyright law:  contributory infringement and vicarious liability.  With respect to contributory infringement, “one who, with knowledge of the infringing activity, induces, causes or materially contributes to the infringing conduct of another, may be held liable as a ‘contributory’ infringer.”[34]  With respect to vicarious liability, “one may be vicariously liable if he has the right and ability to supervise the infringing activity and also has a direct financial interest in such activities.”[35] 

 

            The digital era has so far seen an expansion of secondary liability in two main ways.  First, producers and suppliers of technology that has both infringing and noninfringing uses have increasingly been held liable for infringements committed by their users.  Second, the directness of the financial interest in infringing activity required before a defendant is held vicariously liable for that activity has been significantly loosened.  And although Congress has provided some limitations on the liability of online service providers, those limitations have not significantly cut back on secondary liability, and in any event are often a poor fit for the activities of p2p providers.

 

            A.        Indirect Liability and “Dual-Use” Technologies

 

            The impact on innovation of imposing indirect liability for copyright infringement is particularly important with respect to what might be called “dual use technologies.”  These are products or services that can be used by the consumer in noninfringing ways, but that can also be used to infringe copyright.  The phenomenon of dual-use technologies is not a new one.  After all, musical instruments can be put to both non-infringing uses—such as performances of the performer’s own musical works or uncopyrighted works, as well as private performances (or licensed public performances) of copyrighted musical works—and infringing uses, such as unlicensed public performances of copyrighted musical works.  So can typewriters, printing presses, and photocopy machines.  But the question of whether the developer or supplier of such dual-use technology can be held liable for copyright infringements committed by a purchaser of the technology has attracted substantial legal attention only in the last 25 years or so, and the principal precedent on the question is the Supreme Court’s 1984 decision in Sony Corp. v. Universal City Studios, Inc.[36] 

 

            That decision imposed an important limit on secondary liability in the context of the manufacture and sale of dual-use devices.  Universal and Disney, which own copyrights in many motion pictures and television shows, sued Sony over its manufacture and sale of videocassette recorders (VCRs), alleging that people who bought VCRs and used them at home to tape broadcasts were engaged in copyright infringement, and that Sony was liable for contributing to that infringement.  The Court, by a 5-4 vote, declined to impose secondary liability on Sony, announcing a test borrowed from patent law for holding liable those who manufacture and market devices that buyers might use to infringe copyright:  “[T]he sale of copying equipment, like the sale of other articles of commerce, does not constitute contributory infringement if the product is widely used for legitimate, unobjectionable purposes.  Indeed, it need merely be capable of substantial noninfringing uses.”[37]  Because the Court determined that a VCR was capable of substantial noninfringing use, Sony was not liable for the infringing uses committed by VCR owners merely because it sold the machine. 

 

The substantial noninfringing use test was designed to reconcile the need to give copyright owners effective protection for their works and “the rights of others freely to engage in substantially unrelated areas of commerce.”[38]  The Court’s concern, easily discernible in its analogy to the “staple article of commerce” doctrine in contributory patent infringement cases, was that copyright owners not be allowed to control the development of new technologies used in connection with copyrighted works.  Although the issue directly before the Court in Sony was a claim of contributory infringement,[39] the opinion strongly suggested that its analysis applied to secondary liability for copyright infringement generally, and that the principles in its decision would bar using copyright’s vicarious liability doctrine to hold Sony liable for infringements committed by VCR users.[40]

 

            The Sony doctrine clearly provides significant protection for innovation in technologies that are related to the use of copyrighted material.  Where such innovation leads to a dual-use product or service—that is, a product or service capable of substantial noninfringing use—the innovator can generally rely on Sony for assurance that she will not be held liable for those infringements that consumers commit using the new technology.

 

            In the context of peer-to-peer networks, however, lower court decisions have called into question the protection that the Sony doctrine offers developers of dual-use technologies, though the courts’ opinions leave some uncertainty about how far the cutback goes.  The Ninth Circuit’s decision in A&M Records, Inc. v. Napster[41] is emblematic of this trend.  Music copyright owners sued Napster, charging that users of the Napster network were infringing their copyrights and that Napster was liable for the users’ infringements.  Napster argued that its software and network were capable of substantial noninfringing use and that Sony therefore shielded it from liability for users’ infringements.  The Ninth Circuit, however, read Sony narrowly.  The Sony opinion, the Ninth Circuit concluded, merely barred a court from imputing to a defendant constructive knowledge of another party’s infringement if the defendant was the maker of copying equipment that was “capable of substantial noninfringing use.”[42]  Thus, making and selling equipment capable of noninfringing use could still lead to secondary liability for users’ infringements if a copyright owner could establish by other means that the maker knew, or perhaps should have known,[43] of the users’ infringements and materially contributed to them.  The Ninth Circuit found that Napster had actual knowledge that infringement had occurred on its network and that Napster provided the facilities for that infringement, so Napster could be liable for contributory infringement.[44] 

 

            It is worth noting that it is not clear that Sony itself would have escaped secondary liability under the Ninth Circuit’s reading of the Supreme Court’s test.  The Napster court based its finding of actual knowledge on notices provided by copyright owners to Napster, charging specific past instances of infringing uses of Napster.  It seems quite likely, however, that Disney and Universal would have been able, in the wake of the Sony decision, to provide notice to Sony alleging specific infringing uses by particular VCR owners.  Survey evidence in the case indicated that “a substantial number of [survey respondents] had accumulated libraries of tapes,”[45] and the Supreme Court’s opinion did not address the question of whether library building was a noninfringing fair use.  It therefore seems likely that the studios could have given Sony actual knowledge of infringing use of its VCR by some users and thus, under Napster’s reading, perhaps overcome the Sony court’s limitation on secondary liability.  Since making and selling a VCR seems likely to be a material contribution to the infringing recording of television broadcasts, the consumer-electronics maker might well have been liable under Napster’s interpretation of Sony.[46] 

 

            The Napster court also limited the protective reach of Sony by holding that a product’s capability for substantial noninfringing use was entirely irrelevant to the issue of vicarious liability.  As noted above, the Supreme Court’s opinion in Sony suggested that the maker of a product capable of substantial noninfringing use would not be indirectly liable for user’s acts of infringement under either a contributory infringement or a vicarious liability theory.  The Napster opinion, however, suggests that the noninfringing uses to which an innovator’s technology can be put are irrelevant to the question of vicarious liability.

 

            Post-Napster decisions on peer-to-peer systems have taken conflicting approaches in applying Sony to those systems. 

 

            1.         Aimster 

 

            Aimster provided a file-sharing service over the instant messaging (IM) networks of AOL, ICQ, and Yahoo!. The IM networks allow users to share files with select lists of “buddies,” by providing connections to openly available files and by allowing users to send specific files to one another. Aimster built upon this capability by allowing users to designate all Aimster members as “buddies,” thus allowing users to search for files available on any Aimster member’s open space. In 2001, Aimster began offering an additional fee-based service that provided lists of popular downloaded songs without any additional requests by the user.

 

            Following the Napster decision, Aimster filed for declaratory relief in 2001; and the RIAA and record companies filed suit for copyright infringement.  Aimster filed for bankruptcy in May 2002, and the bankruptcy court ordered an immediate decision on the copyright owners’ pending motion for a preliminary injunction.  In October 2002, the court allowed Aimster to continue operations, but ordered it to refrain from allowing any downloading of the record company plaintiffs’ works, and to post a $500,000 bond.[47]  In a December follow-up order, the court ordered Aimster to disconnect its website and servers entirely, and gave the RIAA permission to request Aimster’s ISPs to disconnect service if Aimster did not do so voluntarily.

 

                        The District Court Decision

 

            Aimster argued that the Sony doctrine shielded it from liability because the Aimster software could be used for noninfringing purposes, including transferring noncopyrighted files (or presumably files in which the transferring party owned the copyright) to other users.  The district court rejected the argument, advancing several grounds for distinguishing Sony and concluding that the doctrine did not prevent Aimster from being held liable for infringements committed using its software.

 

            First, the court found that even though the Sony court had framed the question as whether a product was “capable of substantial noninfringing uses,” the actual facts in Sony established that the VCR’s “principal use” was noninfringing, whereas there was no evidence before the court that any Aimster user had actually used the software for any of the potential noninfringing uses that Aimster identified.  And the court stated that such evidence would have to establish not merely that Aimster was capable of such use, or that it had actually been used for noninfringing purposes, but that such use “constituted Aimster’s primary use.”[48]  In essence, the District Court read the Sony majority to embody an even stricter view of when makers of copying equipment are protected against claims of secondary liability than that held by the Sony dissenters, who would have ruled that “if a significant portion of the product’s use is noninfringing, the manufacturers and sellers cannot be held contributorily liable for the product’s infringing uses.”[49]  None of the verbal formulations used in either Sony opinion indicated that secondary liability could be imposed so long as a product did not have a primary noninfringing use.

 

            In addition, the court suggested that Sony only immunized a supplier of copying equipment against private, home-use copying done using its equipment, and not against the “widespread distribution of infringing works.”[50]  The court also suggested that Sony did not apply when a product was “specifically manufactured for infringing activity,” even if the product did have noninfringing uses, and the court found that Aimster’s service was in fact specifically designed to assist users in infringement.[51] 

 

            Next, the court ruled that Sony applied only to a “staple article of commerce,” and Aimster was not such an article.  A VCR was a “discrete product” that was sold to a buyer who then used the machine as she saw fit.  The court viewed Aimster not as such a product but as an ongoing service that involved an ongoing relationship between Aimster and its users.[52]  Finally, the court read Sony as applying only if the defendant did not influence or encourage the infringement by the users of its product, and found that Aimster did both.[53]

 

            The district court’s reading of Sony in Aimster narrowed even more substantially than did the Napster court the space that the doctrine would provide for developers of new technologies that have both infringing and noninfringing uses to produce and market those technologies without being held liable for copyright infringement committed by the technologies’ users.  A technology’s capability for substantial noninfringing uses would appear to be irrelevant to the innovator’s secondary copyright liability if the product’s principal actual use was infringing.   

 

                        The Seventh Circuit’s Decision

 

            The Seventh Circuit affirmed the preliminary injunction, but on substantially narrower grounds.  The Seventh Circuit expressed concern about the impact of secondary liability on the development of new online services.  It held that in applying the Sony doctrine to the provider of an ongoing service (rather than a discrete product), the service provider’s ability “to prevent its customers from infringing is a factor to be considered in determining whether the provider is a contributory infringer.”[54]  But the court recognized that ability to prevent infringement should not in itself determine liability because such a rule would have “alarming” adverse consequences for the provision of “dual use” services:

 

If a service facilitates both infringing and noninfringing uses, … and the detection and prevention of the infringing uses would be highly burdensome, the rule [that imposes liability whenever the service provider knows of infringing activity and could prevent it] could result in the shutting down of the service or its annexation by the copyright owners (contrary to the clear import of the Sony decision), because the provider might find it impossible to estimate its potential damages liability to the copyright holders and would anyway face the risk of being enjoined.[55]

 

            The Aimster court also expressed disagreement with the Ninth Circuit’s position in Napster, which it characterized as “suggesting that actual knowledge of specific infringing uses is a sufficient condition for deeming a facilitator a contributory infringer.”[56]

 

            The court did not, however, merely reaffirm the Sony opinion’s language that secondary liability would not be imposed on the supplier of a technology that is “capable of substantial noninfringing use.”  Rather, it held that in order to determine whether the supplier of a dual-use service was liable for user’s infringements, “some estimate of the respective magnitudes of [noninfringing and infringing] uses” must be made.[57]  The court made clear that it was not enough for Aimster to show that “its file-sharing system could be used in noninfringing ways.”[58]  The fact that a product or service is capable of noninfringing uses would not exempt the supplier from liability if the product or service “in fact is used only to infringe.”[59]  Because there was evidence that Aimster’s system was in fact being used for infringing purposes, the court said that the burden shifted to Aimster, at least at the preliminary injunction stage, “to demonstrate that its service has substantial noninfringing uses.”[60]  It turned out, however, that the court did not in fact think that it was sufficient that Aimster service had noninfringing uses.  What the court actually required was that Aimster quantify how much of the use of its system was noninfringing.  Thus, although the court itself explained several possible noninfringing uses of the Aimster system, it concluded that “[i]t is not enough … that a product or service be physically capable, as it were, of a noninfringing use.  Aimster has failed to produce any evidence that its service has ever been used for a noninfringing use, let alone evidence concerning the frequency of such uses.”[61]  This lack of evidence relieved the court from having to decide how frequent noninfringing uses would have to be for the service provider to escape liability for infringing uses of the service, though it quoted the district court’s language on Aimster’s failure to show that the “primary” use of the system was noninfringing.

 

            The court further suggested that even if a new technology was not only capable of substantial noninfringing use but was in fact used in noninfringing ways, that would not in itself be enough to avoid secondary liability for actual infringements:

 

Even when there are noninfringing uses of an Internet file-sharing service, moreover, if the infringing uses are substantial then to avoid liability as a contributory infringer the provider of the service must show that it would have been disproportionately costly for him to eliminate or at least reduce substantially the infringing uses.[62]

 

The requirement, at least in the context of peer-to-peer services, that a supplier design her service to prevent or reduce infringement unless it is excessively costly to do so appears to go beyond what Sony required: although the dissenters in Sony discussed design alternatives available to Sony that would have reduced infringement, the majority made no mention of those possibilities as relevant to the question of Sony’s liability for its users’ infringements.[63]

 

            While the Seventh Circuit’s Aimster decision preserves more of the Sony doctrine’s protection for innovators of dual-use technology than does the Ninth Circuit’s Napster opinion, Aimster’s interpretation of Sony does pose significant challenges to innovation.  Someone who develops a new dual-use technology must be concerned about whether noninfringing use of that technology will not only be “substantial,” but perhaps whether it will be the primary use, as well as whether she will be able to prove that substantial or primary use in court.  Perhaps more significantly, even if the innovator is confident as to how the technology will be used, she will have to consider, at least in the case of services, whether she can design the technology to reduce or eliminate the possibility of infringing uses of the technology, what the costs of doing so are, and whether a court will decide that those costs are “disproportionate” and so need not be expended.

 

            2.         Grokster 

 

            Yet another approach to applying Sony to peer-to-peer services emerged in the decision of the district court in California against Grokster Ltd. and StreamCast Networks, providers of peer-to-peer software.  Record and movie studios sued several defendants, alleging that users of their software infringed on the plaintiffs’ copyrights and that the defendants, as providers of the software, were secondarily liable for that infringement.  In April 2003, the district court granted summary judgment to two of the defendants, Grokster and StreamCast, which disseminate the “Grokster” and “Morpheus” software, respectively.[64]  The Grokster decision follows Napster in reading Sony as only barring the imputation of constructive knowledge to a defendant who has supplied a product or service that is capable of substantial noninfringing use.  But in applying the principles announced in Napster to the providers of peer-to-peer software, the Grokster court appears to have given innovators of dual-use technologies more breathing room. 

 

            The court described the operation of the Grokster and Morpheus networks as follows:

 

            Although novel in important respects, both the Grokster and Morpheus platforms operate in a manner conceptually analogous to the Napster system …

            In both cases, the software can be transferred to the user’s computer, or “downloaded,” from servers operated by Defendants. Once installed, a user may elect to “share” certain files located on the user’s computer, including, for instance, music files, video files, software applications, e-books and text files. When launched on the user’s computer, the software automatically connects to a peer-to-peer network … and makes any shared files available for transfer to any other user currently connected to the same peer-to-peer network.

            Both the Morpheus and Grokster software provide a range of means through which a user may search through the respective pool of shared files. … Once a search commences, the software displays a list (or partial list) of users who are currently sharing files that match the search criteria…

            The user may then click on a specific listing to initiate a direct transfer from the source computer to the requesting user’s computer. When the transfer is complete, the requesting user and source user have identical copies of the file, and the requesting user may also start sharing the file with others. Multiple transfers to other users (“uploads”), or from other users (“downloads”), may occur simultaneously to and from a single user’s computer. [65]

 

            Finding no dispute that at least some users of the defendants’ software engaged in direct copyright infringement, the court turned to the question of whether Grokster and StreamCast could be liable as contributory infringers.

 

            With respect to the defendants’ knowledge of end-user infringement, the court followed the Ninth Circuit’s Napster decision in reading the Supreme Court’s Sony opinion as addressing only the knowledge required for a finding of contributory infringement where a product is capable of substantial noninfringing use.  The court found that the defendants’ software is capable of such use, including dissemination of works with the consent of the copyright owner and dissemination of works not protected by copyright, and the defendants offered evidence of such actual use by its customers.[66]  As a result, the court applied the standard announced in Napster, which it read as prohibiting the court from imputing knowledge to a defendant based on the fact that its software could be used to infringe copyrights.  The district court, however, focused on the timing of a indirect-liability defendant’s knowledge of infringing activity:

 

            Rather, liability for contributory infringement accrues where a defendant has actual—not merely constructive—knowledge of the infringement at a time during which the defendant materially contributes to that infringement.

            In other words, as the Ninth Circuit explained, defendants are liable for contributory infringement only if they (1) have specific knowledge of infringement at a time at which they contribute to the infringement, and (2) fail to act upon that information.[67]

 

            The court noted evidence (including internal documents and searches by company executives) that showed that Grokster and StreamCast “clearly know that many if not most of those individuals who download their software subsequently use it to infringe copyrights.”[68]  In addition, the court noted that the plaintiffs had sent defendants thousands of notices of claimed infringements.  But in the court’s view the crucial question was whether the defendants had “actual knowledge of infringement at a time when they can use that knowledge to stop the particular infringement”[69]—that is, “whether actual knowledge of specific infringement accrues at a time when either Defendant materially contributes to the alleged infringement, and can therefore do something about it.” [70]

 

            With respect to the defendants’ material contribution to their users’ infringements, the court framed that question as “whether Grokster and StreamCast do anything, aside from distributing software, to actively facilitate—or whether they could do anything to stop—their users’ infringing activity.”[71]  Neither Grokster nor StreamCast operated the network over which the users of their software connected and exchanged files, and the court emphasized the decentralized nature of those networks: when users search for and initiate file transfers, no information is transmitted to or through any computers owned or controlled by the defendants.[72]  Further, the court explained that when a user wishes to connect to the Grokster or Morpheus P2P networks, the user must locate another user to whom to connect, but emphasized that neither defendant was involved in the process that allows a user to locate a network connection.  The court sharply contrasted Grokster and StreamCast’s operations with Napster’s: 

 

            Plaintiffs appear reluctant to acknowledge a seminal distinction between Grokster/StreamCast and Napster: neither Grokster nor StreamCast provides the “site and facilities” for direct infringement. … Users connect to the respective networks, select which files to share, send and receive searches, and download files, all with no material involvement of Defendants. If either Defendant closed their doors and deactivated all computers within their control, users of their products could continue sharing files with little or no interruption.

            In contrast, Napster indexed the files contained on each user’s computer, and each and every search request passed through Napster’s servers.  Napster provided the “site and facilities” for the alleged infringement, affording it perfect knowledge and complete control over the infringing activity of its users. If Napster deactivated its computers, users would no longer be able to share files through the Napster network.[73] 

 

The court therefore concluded that Grokster and StreamCast did not provide active and substantial contribution to end-user infringements[74] in a way that justified holding the companies liable as contributors to those infringements:

 

Defendants distribute and support software, the users of which can and do choose to employ it for both lawful and unlawful ends. Grokster and StreamCast are not significantly different from companies that sell home video recorders or copy machines, both of which can be and are used to infringe copyrights. While Defendants, like Sony or Xerox, may know that their products will be used illegally by some (or even many) users, and may provide support services and refinements that indirectly support such use, liability for contributory infringement does not lie “merely because peer-to-peer file-sharing technology may be used to infringe plaintiffs’ copyrights.”[75]

 

The Grokster decision thus offers innovators of dual-use technologies substantially more protection against the danger of secondary liability for their users’ acts of copyright infringement than do the Napster or Aimster opinions, at least where the innovator creates a dual-use product and does not have an ongoing service relationship with the user.  Whether this approach will continue, however, will depend on the Ninth Circuit, as the plaintiff copyright owners have appealed the decision.

 

B.        Expansion of Vicarious Liability and The “Direct” Financial Interest Requirement

 

            While the Sony doctrine’s protection of developers of dual-use technologies has been interpreted so as to make it of uncertain use to innovators of peer-to-peer technologies, the doctrine may be undermined entirely by recent developments in the law of vicarious infringement.  Vicarious liability for infringement committed by a third party has expanded in recent years, offering another possible approach for copyright owners to hold peer-to-peer developers liable for infringement committed by users of their technologies.  The basic rule is that “one may be vicariously liable if he has the right and ability to supervise the infringing activity and also has a direct financial interest in such activities.”[76]  In recent years, courts have substantially expanded what constitutes a sufficiently “direct financial interest” in an infringers’ activity to hold a third party liable for that activity.  The result is that ever more parties are potentially subject to vicarious liability for others’ copyright infringements, including innovators who may be deterred from pursuing innovations because of such potential liability.[77] 

 

            Vicarious liability in copyright originated from the doctrine of respondeat superior, holding employers liable for infringements committed by their employees.  The doctrine expanded to hold defendants liable for infringements committed by independent contractors as well.  The seminal case of Shapiro, Bernstein and Co. v. H.L. Green Co.,[78] involved a department store whose record departments were operated by an independent concessionaire.  Green received ten to twelve percent of the concessionaire’s gross receipts from record sales.  The concessionaire sold infringing recordings, and Green was held liable.  The court found that Green had “an obvious and direct financial interest in the exploitation of copyrighted materials” by the concessionaire—indeed, the court viewed Green as having “a most definite financial interest in the success of [the] concession; 10% or 12% of the sales price of every record sold by [the concessionaire], whether ‘bootleg’ or legitimate, found its way … into the coffers of the Green Company.”[79] 

 

            In recent years, the doctrine has far outgrown the employment and independent contracting contexts, and the financial interest that a defendant must have in a third party’s infringing activities in order to be held liable has become more attenuated.  The Ninth Circuit’s 1996 decision in Fonovisa, Inc. v. Cherry Auction, Inc.,[80] is generally viewed as a major case in the expansion of vicarious liability.  The defendant in that case operated a flea market where it rented space to third-party vendors; Cherry Auction advertised the flea market to the public and charged customers for parking, admission, and food sold at the market.[81]  Fonovisa sued, seeking to hold Cherry Auction liable for sales by one of its vendors of infringing recordings.  Under the doctrine of vicarious liability, the Ninth Circuit held that Cherry Auction “reap[ed] substantial financial benefits from admission fees, concession stand sales, and parking fees, all of which flow directly from customers who want to buy the counterfeit recordings at bargain basement prices,” and that this was sufficient for the imposition of vicarious liability.[82]  For the court, because the infringing activity “enhance[d] the attractiveness of the venue to potential customers” or served as a “draw” for customers, the venue operator could be held liable for the infringing activity.

 

            Fonovisa’s interpretation of the “direct financial interest” standard for vicarious liability allows imposing liability for what seems to be a somewhat indirect financial interest.  The flea market earned nothing directly from the sale of infringing recordings by one of its vendors.  Instead, the court assumes that the vendor’s offering of infringing recordings attracted to the flea market customers who otherwise would not have attended, and those additional customers would result in revenues to the flea market not from their purchase of infringing material but from ancillary fees.  This financial connection seems fairly clear in the traditional “dance-hall” cases, which hold the operator of a dance hall vicariously liable for infringing public performances of copyrighted musical works committed by a band that the operator hired to play in the dance hall.  Most customers pay admission to the dance-hall operator largely because they wish to hear music performed.  Thus, the operator’s financial interest in the infringing performances seems fairly direct.  It seems far less clear that most flea market shoppers pay admission to a flea market largely because they wish to purchase infringing recordings.  But the existence of infringing activity is assumed to draw customers in greater numbers than noninfringing activity, and any money those customers pay to the defendant appears to count as revenue “directly” related to the infringing activity for purposes of vicarious liability.

 

            In Napster, the Ninth Circuit in a single paragraph loosened the “direct financial interest” requirement even further.  The court followed Fonovisa in ruling that the availability of infringing music on the Napster system served as a “draw” for users.  But because Napster disseminated its software to users, and permitted them to use its system to list and locate titles, at no charge, it did not, unlike Cherry Auction, make money off of the customers attracted by the infringing material.  Indeed, it did not even make money indirectly, by selling advertising to users of the service.  The Ninth Circuit concluded, however, that because Napster would likely charge users in the future, and because that “future revenue is directly dependent upon ‘increases in userbase,’” Napster had a sufficiently direct financial interest in infringement committed by its users to warrant holding the company vicariously liable for that infringement.  Thus, not only can a defendant be held liable if it earns money from ancillary services to customers attracted by infringement, it can be held liable if it is likely to earn such money in the future.

 

            As a result of the loosened requirement for direct financial interest, innovators are more likely today to be found vicariously liable for copyright infringement committed by users of their innovations.

 

            C.        Statutory Safe Harbors for Internet Service Providers

 

            For innovators who are also Internet service providers, the Sony doctrine is only one source of limitation on liability for copyright infringement.  In 1998, Congress enacted, as part of the Digital Millennium Copyright Act, statutory limitations on the liability of those who provide online services.  Essentially, Congress provided online service providers (OSPs) with several safe harbors:  if an entity qualifies as an OSP[83] and meets two basic eligibility requirements,[84] then it is exempt from all monetary relief and most injunctive relief for copyright infringement with respect to four specified categories of activities if specific conditions (which vary with the type of activity) are met.  If an OSP fails to qualify for the safe harbor on any basis, then its liability for copyright infringement is to be determined by ordinary principles of copyright law.

 

            Congress enacted the safe harbors in 1998 as part of the Digital Millennium Copyright Act in response to concerns expressed by online service providers about their potential overwhelming liability for copyright infringement committed by their users.  These statutory safe harbors, however, have not provided significant protection from indirect liability to innovators of dual-use technologies, particularly in the peer-to-peer context.  The main reason for this is that the most relevant safe harbor for peer-to-peer systems primarily protects service providers against liability for acts of direct copyright infringement committed by the provider.  The safe harbor largely preserves the availability of relief against service providers on the basis of secondary liability for infringement committed using the service, though the safe harbors may somewhat heighten the requirements for holding the provider secondarily liable.  In addition, judicial interpretation of the safe harbors has curtailed the usefulness of their protection for P2P innovators.

 

            The safe harbors apply to any “provider of online services or network access, or the operator of facilities therefor.”[85]  Some innovators whose products involve copyrighted works in digital format will likely not meet this definition and therefore not be eligible for the safe harbors at all.  A company that distributes peer-to-peer software, for example, may be disseminating a product that its customers use over an online network, but the company itself may not be providing (or operating any facilities for) online services or network access.  Similarly, a company that merely makes a digital video recorder capable of automatically skipping commercials would not necessarily be a service provider within the scope of the safe harbors.  Other innovators, however, will likely qualify as service providers eligible for safe harbor protection.  Napster and Aimster, for example, seem at least to have offered their users “online services;” a Napster user would connect over the Internet to Napster’s own computers in order to identify music files available for copying.

 

            The safe harbors protect service providers from copyright infringement liability for four kinds of activity.  The first harbor, in § 512(a), essentially protects against liability for merely transmitting or retransmitting someone else’s material over a computer network—that is, essentially for serving as a mere conduit for Internet transmissions, as ordinary Internet service providers such as Earthlink do when they transmit a customer’s e-mail message over the Internet to its addressee or when they retrieve a Web page from a third-party’s computer and transmit it to a customer’s at that customer’s request.[86]  The second safe harbor protects a service provider who temporarily caches or stores online material on its own system or network in order be able to transmit that material at a later time to other of the provider’s users who request it.[87]  A third safe harbor limits the liability of a service provider that stores information on its own system or network at the direction of a user, such as an ISP that hosts a user’s Web site on the ISP’s computers or an online auction Web site such as eBay that hosts a customer’s auction information on its computers.[88]  Finally, a fourth safe harbor shields service providers who offer “information location tools.”[89]  These include not only directories, indices and search engines that direct users to information on the Internet, but also any “reference, pointer, or hypertext link” to such information.[90]

 

            For peer-to-peer service providers that meet the statutory definition of “service provider,” the § 512 safe harbors may offer little protection from liability for copyright infringement committed by the service’s users.  The first court to address this issue in any depth is the district court in the Napster case.[91]  That court ruled that Napster was not protected from liability for its users’ copyright infringements under the § 512(a) “conduit” safe harbor.  It reasoned that the safe harbor only protects a service provider against liability for the provider’s “transmitting, routing, or providing connections for, material through a system or network controlled or operated by or for the service provider.”[92]  Because any infringing transfer of files in Napster’s system occurred directly between two Napster users over the Internet and not through Napster’s own system, the court concluded that the secondary liability claims against Napster were not based on the activity shielded by Section 512(a) and thus were not precluded by the safe harbor.  The district court in the Aimster case reached the same conclusion about that company’s system.[93] Given that peer-to-peer services, by their very nature, involve decentralized transmissions directly between users, the § 512(a) safe harbor, at least as interpreted by the Napster court, seems likely to offer little protection against secondary liability claims.

 

            The safe harbors in Sections 512(b) and 512(c) will generally not offer peer-to-peer innovators protection against secondary liability claims because those provisions cover infringement claims arising out of the storage of material on the defendant’s own computer system or network, and P2P systems, involving as they do transmissions directly between two users of the system, typically do not involve such central storage on a provider’s computer.[94]

 

            Section 512(d) governing “information location tools” might cover activities of some P2P providers, since those providers may supply to users information about where a particular file is available on the P2P network.  Indeed, the district court in Napster ruled that the defendant “undisputedly performs some information location functions.”[95]  Nonetheless, Section 512(d) provides little protection to innovators against secondary liability claims, because the safe harbor primarily shields defendants against liability for acts of direct copyright infringement while placing essentially no limitation on claims that a service provider is secondarily liable for the direct infringements of its users.  As a result, someone who directs users to infringing online material or activity remains liable under Section 512(d) if that person actually knows that the material or activity is infringing or is aware of facts or circumstances from which the infringing activity is apparent.[96]  This essentially mirrors the basic test for contributory infringement, which allows liability if a defendant knows or has reason to know of infringing activity and materially contributes to that activity.[97]  And the safe harbor also does not protect a provider who receives “a financial benefit directly attributable to the infringing activity” if the provider “has the right and ability to control such activity.”[98]  This, of course, is the basic test for vicarious liability, so if a plaintiff can establish the elements of a claim that the service provider is vicariously liable for its user’s infringement, the 512(d) safe harbor will not limit the provider’s liability pursuant to that claim.[99]  The district court in the Aimster case took just this approach, treating the conditions on the 512(d) safe harbor as being identical to the elements of claims for contributory infringement and vicarious liability; because the Aimster court ruled that the defendant’s conduct came within the scope of these secondary liability doctrines, it therefore concluded that Aimster’s conduct was simultaneously outside the scope of protection of the safe harbor.

 

            Thus, to the extent that claims for indirect liability pose a threat to P2P innovations, the safe harbors for online service providers added to copyright law by the Digital Millennium Copyright Act do little to ameliorate that threat.  Combined with the uncertain application of the Sony doctrine to online digital innovations after the Napster decision and the loosening of the direct financial interest requirement for vicarious liability, suits seeking to hold digital innovators liable for infringements committed by users of their products and services have substantial likelihood of success.

 

II.        The Economics of Digital Copyright Infringement

 

            A.        What Has Changed?

 

            Why have copyright owners shifted from suing infringers to suing facilitators?  The answer lies in a fundamental shift in the economics of copyright in the digital environment.  Copyright in the United States has always been a utilitarian response to a public goods problem.[100]  It costs more to create a work than it does to imitate someone else’s work, and so without some sort of control over imitation creators will not have enough incentive to create.[101]  But this public goods problem has always been an incomplete one.  It was never the case that imitation was costless, only that it was cheaper than creation.  An infringer who wanted to distribute counterfeit copies of a book, record, computer program or videotape in the twentieth century needed the same sort of production and distribution facilities that the copyright owner did.  Counterfeiters had to print books, press records, or record tapes or disks en masse, and then find a way to ship those counterfeit copies to their own network of retailers, who had to be paid to sell the illegal copies.  The costs of distributing counterfeit copies might be somewhat less than the cost of legitimate distribution -- the copies might be sold on a card table on a street corner rather than in a storefront – but counterfeiting required a substantial business of facilities and employees. 

 

            During most of the twentieth century counterfeiters were also clearly distinct from individual end users.  End users might also make copies without authorization by or payment to the copyright owner, and some of those end user copies might be illegal.[102]  But unlike counterfeiting, end user copies weren’t a serious threat to a copyright owner’s sales during this period.  End user copies were made for their own personal use, often did not substitute at all for purchase of a lawful copy,[103] and were at worst only a very small source of substitution for lawful copies.  They were not widely distributed: a college student might tape an album for a few friends, but was unlikely to make several thousand tapes and give or sell them to strangers.[104] 

 

Copyright owners in the twentieth century sued counterfeiters, but generally did not sue end users even if they were making illegal copies.  This made perfect sense given the economics of traditional copyright.  There were relatively few such counterfeiters, and the harm each one caused to copyright owners was large enough to justify spending the money to find them and shut them down.  By contrast, there were a large number of end users making copies, many of them legitimate customers of the copyright owners.  Many of those copies were legal, or at least of debatable legality.  And in any event the injury to copyright owners caused by any single end user was quite small, if not zero.  Even if it was legally possible, it simply wasn’t economically rational to stop the end users.[105]

 

The digital environment is quickly changing this calculus.[106]  The great promise of digital distribution – the virtual elimination of the costs of production and distribution[107] – is a mixed blessing for copyright owners.  Content owner costs go down as they embrace digital distribution,[108] but so do the costs of counterfeiting.  Indeed, as the costs of production and distribution approach zero, the public goods problem gets worse, because the ratio of the cost of creation to the cost of imitation approaches infinity.[109]  Further, as the cost of counterfeiting approaches zero, the sharp division between professional counterfeiters and end user copiers breaks down[110]  Anyone can give copies of software or music to others on the Internet in a variety of ways: put it on a Web page, email it to friends or to a listserv, swap it on IRC, or make it available for download on a peer-to-peer file sharing site.  It costs virtually nothing to do so.  And unlike end user copying in the analog environment, online copying by end users can be quite substantial.  If I tape a CD to give to a friend, I have deprived the record company of at most one sale.  If I post the CD online, thousands or tens of thousands of people might download the music, and the company might lose a large number of sales.  This problem is exacerbated by the fact that it is easier to make a copy of digital content than it is to photocopy a book or tape a CD, and that unlike photocopies or analog recordings digital copies do not degrade in quality from generation to generation, permitting those who obtain copies to make copies of the copies.  The massive decline in the cost of copying has made copyright infringement a more significant problem in the digital environment. 

 

The economics of digital copyright have also rendered traditional solutions to counterfeiting obsolete.  The wide distribution of copies made by end users over the Internet means that content owners can no longer ignore end user copies and focus on professional counterfeiters.  In order to stop counterfeiting online, copyright owners must stop the end user copies as well.[111]  But it simply isn’t cost effective to sue each end user for copyright infringement.  Napster had 70 million users at its peak; estimates of usage for the various components of the Morpheus network are even higher.[112]  It may cost as much as $400,000 to take even a straightforward copyright case to trial.[113]  Suing even a fraction of the end users could bankrupt the content industries.[114]

 

Copyright owners have understandably cast about for an alternative to suing end users.  The strategy they have settled on is to sue facilitators.  Suing facilitators is cost-effective for the content industries because a single lawsuit can eliminate the distribution mechanism for a large number of end user copies.[115]  The Napster case, for example, shut down what was then the single largest forum for distributing music copies;[116] the music and movie industries now hope to do the same with other p2p networks like Morpheus.  The more attenuated facilitator suits have two analogous goals: to prevent access to Web sites that contain digital content by shutting down search engine or wireline connections to those sites,[117] and to make investment in or assistance to new digital distribution companies risky.[118]  If copyright owners can shut off the distribution channels for digital content, they don’t need to worry nearly as much about the low cost of making any given copy of that content.  From their perspective, suing facilitators is a logical response to the changing economics of copyright law.  Unfortunately, as we show in the next section, it is not a socially optimal response. 

 

In an important article, Ray Ku argues that digital distribution technology has eliminated the need for middlemen, and therefore solved the public goods problem.[119]  If Ku is right, there is no justification for copyright protection at all in a digital world.  We are skeptical of both parts of Ku’s syllogism, however.  First, while it may well turn out to be the case that digital distribution mechanisms such as p2p networks can replace the physical movement of goods, they certainly haven’t done so yet.  Despite massive downloading of music, CD sales have declined only somewhat, and DVD and book sales seem even less affected by digital distribution.[120]  There is still a demand for physical copies of works of authorship, and likely will be for the foreseeable future – particularly for the sizeable minority of Americans who are still not online.  Further, intermediaries can play a useful role in screening and selecting content; a consumer faced with 100,000 songs by independent artists on a p2p network may simply not know where to turn.[121]  For these reasons, among others, we are not persuaded that intermediaries will disappear in the digital environment. 

 

Even if digital distribution fulfilled the promise Ku sees for it, we think copyright incentives will remain important.  One need not agree with Samuel Johnson’s quip that “no man but a blockhead ever wrote, except for money”[122] to think that the incentives provided by copyright encourage a substantial amount of creativity.  Ku bases his argument on the assumption that the “creation of the work does not depend upon funding derived from the sale of copies.”[123]  Ku is surely right to suggest that writers and artists create for a variety of reasons, and that many would create without any hope of recompense.[124]  But fewer would do so, particularly in industries like Hollywood where production costs are substantial, and that additional creativity is what copyright is designed to encourage.[125]

 

            B.        What’s Wrong With Suing Facilitators?

 

                        1.         Lumping Legal and Illegal Content Together

 

            Suing intermediaries and facilitators differs in fundamental ways from suing counterfeiters.  A lawsuit against a direct infringer allows the court to make a determination about the accused infringer’s conduct.  A court holds an accused infringer liable only if it determines that he did in fact infringe.  Any accused infringer can defend such a suit by arguing that he did not infringe, or that his infringement was justified or excused by a recognized defense.  The same is true of certain types of indirect liability for infringement in the traditional copyright system.  If I am held liable for inducing another to infringe, it can only be because the court has concluded that I had the required relationship with a party who is found to infringe.  The specific facts of the case matter.[126]

 

            Suits against third parties in the digital environment do not – indeed cannot – make this distinction.  Suits against facilitators premised on individual cases of infringement would pose the same economic problem for copyright owners as suits against the individual infringers themselves.  Rather, the class of suits we consider in this paper involves efforts to shut down a facilitator entirely[127] or to require modification in the way the facilitator operates.[128] 

 

            The problem with these claims is that they lack the granularity of suits against direct infringers.  For example, in the Grokster case, the Central District of California had to decide either to ban the distribution of software that permits users to connect to the Morpheus network or not to ban it.[129]  That binary choice is ill-suited to the realities of the Morpheus network, over which individual end users trade lots of plaintiffs’ content, trade some content that is either in the public domain or which the copyright owner permits, trade some files of a type that tends not to be copyrighted at all, and trade significantly more content that might be copyrighted, but whose owner has neither granted permission nor sought its removal by joining in the lawsuit.  Lawsuits against end users can distinguish between those who post infringing content and those who don’t.  A lawsuit against the software maker can’t draw that distinction.  And not surprisingly, lawsuits against facilitators are even worse at determining whether individual end users have made lawful use of the plaintiff’s content, a fact-specific inquiry involving the acts and motivations of millions of people who are not parties or even witnesses in the case. 

 

            Thus, courts in the facilitator lawsuits are put to an unpleasant choice: they must either ban unquestionably lawful conduct in order to get at the infringing conduct, or must let the infringing content remain online in order to protect the legal trading of content.  Neither alternative is particularly attractive as a general matter.  The balance between the two harms will tilt different ways in different cases, however.  The closer the facilitator is to direct infringement, and the more closely tailored the facilitator’s system is to infringing content, the less collateral harm an injunction will cause to legitimate users.[130]  At some point, as Landes and Lichtman note, “the benefits in terms of increased copyright enforcement come at too high a cost in terms of possible interference with the sale of a legitimate product.”[131]

 

In Napster, for instance, the service was limited to the trading of music files by users, and the evidence submitted to the court suggested that at least 87% of the files traded, and perhaps as many as 99% of the files traded, were copyrighted by the plaintiffs.[132]  Some of those copyrighted files were doubtless downloaded for purposes that the law would allow, but even so it seems reasonable to conclude that shutting down Napster stopped far more illegal conduct than legal conduct.  As we move further away from services tailored to facilitating infringement, however, the balance shifts.  Unlike Napster, the Morpheus network permits the transmission of any type of file.  While we have not seen definitive evidence on usage, it appears that the plaintiffs in MGM own copyrights covering rather less than 75% of the content shared on the network.[133]  Banning the distribution of software that allows users to connect to the Morpheus network would therefore stop more legal conduct and less illegal conduct than an injunction against Napster.  As lawsuits move further and further from the actual infringer in their effort to find a lever to stop piracy, the balance shifts even further against the copyright owner.  Suits against ISPs or search engines are likely to target far more legal conduct than illegal conduct, and the net social harm to shutting such a site down is correspondingly greater. 

 

                        2.         Requiring the Facilitator to Police Is Not a Solution

            A court might try to get around this problem by enjoining the distribution of infringing material on a facilitator’s network, rather than shutting down the site altogether.  The Ninth Circuit took this approach in Napster, seeing it as a compromise that preserved the legal uses of the network while stopping copyright infringement.[134]  This approach echoes the increasingly common approach of building safeguards against copyright infringement into devices or into the network itself, an approach known as “digital rights management” (DRM).[135]  Congress has also considered requiring device manufacturers to build in encryption and other tools to prevent infringement.[136] 

 

Such an intermediate approach is unlikely to work, for several reasons.  First, Napster itself is a caution.  The district court ultimately went further on remand than the Ninth Circuit seemed to authorize, holding that Napster must design its system so that no infringing content can get through before being allowed to provide its users with access to noninfringing content.[137]  The end result was that the parties fought for months about how to redesign the Napster system, and the system never went back online.  The “intermediate” injunction was no different than an outright prohibition on the Napster system.  Second, there will always be disputes over what content is infringing.  Copyright law is full of gray areas,[138] and copyright owners have a history of trying to extend the law beyond its bounds.[139]  A court that decides to stop infringing content while letting the rest of the service continue will either have to enjoin all infringing content in advance (in which case no rational defendant will operate their system at all, for fear of going to jail for contempt) or will be signing up to arbitrate an endless series of oversight disputes about particular cases. 

 

Third, and most important, the idea of enjoining only the infringing material presupposes control by the facilitator over the material that is distributed on the system.  Napster could in fact exercise such control, because it ran a central directory service that customers had to use in order to find music on the system.  The same cannot be said of most other facilitators, however.  Sony has no control over the uses to which its VCRs are put.  Companies like Streamcast and Grokster sell software manufactured by others and used by individuals who make files available on the Morpheus network; they have no ability to remove certain files from the network while retaining others.  Similarly, Jon Johansen can’t somehow make DeCSS available only to those who want to use it to view DVDs on a computer, but keep it from those who want to make illegal copies of a DVD.[140] 

 

3.         Agency Cost Problems

Even if it were feasible, the idea of compelling facilitators to stop some but not all content would not be socially optimal because facilitators don’t have the proper incentives to distinguish lawful from infringing content.  Assaf Hamdani and others have noted that third parties are too quick to take down material posted on their Internet sites by others when they receive a complaint of copyright infringement.[141]  ISPs, auction sites, search engines, wireline providers and other intermediaries capture only a tiny part of the value of a third party posting.  Indeed, if the third party pays a flat rate, the intermediary may not suffer any financial consequence from removing a posting or link.  Indeed, the problem of automatic takedown is so great that when Congress passed the safe harbor for ISPs in the Digital Millennium Copyright Act, it felt compelled to require ISPs to put back disputed content if their customers complained about taking it down.[142]  The fact that these intermediaries don’t bear the full social cost of taking down challenged content means that enforcing copyright law by requiring them to do so creates negative externalities, tilting the law too far in favor of copyright owners.[143]

 

            4.         Loss of the p2p Distribution Network

p2p networks are a particularly efficient means of distributing content.[144]  They have several advantages over both the existing distribution networks for CDs and over the “top-down” distribution models such as MusicNet and Pressplay that have been implemented by content owners to date.[145]  First, p2p networks are distributed, while authorized download sites tend to be centralized.  As a result, functioning p2p networks are less vulnerable to bandwidth constraints and the crash of a central server, for the same reason that the Internet is resilient in avoiding such problems.  Second, p2p file sharing is inherently responsive to content demands.  The fact that consumers are also suppliers means that if a large number of people want to download the latest Shakira song, a large number of people are likely to make that song available for upload too, because uploaders by definition provide only the music they themselves download from others or rip from a CD.  The music industry doesn’t need to print more CDs or decide which songs have sufficient demand to support giving them server space to make this happen; it happens on its own.  Third, p2p networks may affect the creation as well as the distribution of information by facilitating what Yochai Benkler has called “peer production.”[146]  Finally, and most significantly, p2p networks harness volunteers providing essentially free computing resources.[147]  Just as millions of users support the Search for Extraterrestrial Intelligence (SETI) by donating idle processing power,[148] p2p file sharers are donating their idle computer resources to the cause of music distribution. 

 

It is an unfortunate fact of modern life that this efficient distribution mechanism is used to distribute illegal rather than legal copies in many cases.  But shutting down p2p networks to solve the infringement problem forces us to rely on a less-efficient mechanism for distributing digital content.  This lost efficiency represents a cost to society, one that could be avoided if there were a way to harness the benefits of p2p networks in the service of distributing authorized content.

 

            5.         Harms to Innovation

Another, potentially even more corrosive, problem with suing facilitators is the danger such suits pose for innovation.  Traditional copyright suits against direct infringers do not threaten innovation, since they target only the infringing user of that innovation.  Suits against facilitators, by contrast, seek to outlaw a service entirely or to declare a device or program contraband.  Banning the sale of a device or computer program obviously restricts innovation directly, and therefore reduces social welfare by the net social value of that innovation.  If the courts declare p2p networks illegal altogether (or indirectly do so by ordering modifications and filtering), for example, the social cost will not only be the foregone legal uses of those networks but the unanticipated future benefits those networks could have brought.  Economic evidence strongly suggests that those unanticipated future benefits, or “spillover” effects, exceed the immediate value of most new technologies.[149]  The VCR is an obvious example of a technology that the copyright industries tried to ban,[150] but which later developed in unanticipated ways, creating a new market that has provided tremendous benefit to the very copyright owners who would have outlawed it.[151]  The early history of radio offers a similar lesson.[152]

 

The alternative discussed above – requiring programmers to change their sites to build in screens against illegal copying – is little better, because it puts the courts or Congress in the untenable position of dictating to programmers how they should design their products.[153]  Innovation works best when it is unfettered by governmental requirements, particularly the kind of detailed oversight that the Napster case ultimately entailed.

 

Over and above the direct restrictions on innovation, the threat of lawsuits or criminal prosecutions against innovators is likely to deter a significant amount of innovation, some of which would unquestionably have been legal.[154]  The anecdotal evidence of such deterrence is quite strong. When programmers started being prosecuted criminally for writing code that violated the DMCA’s anticircumvention provisions,[155] and online magazines were sued for writing stories that linked the reader to allegedly unlawful sites,[156] the result was to chill programming, deterring some from working on encryption at all and steering others away from work in certain areas perceived as sensitive.[157]  A number of programmers went so far as to file suit against the content industries, seeking a declaratory judgment that their conduct was lawful.[158]  Litigation is expensive, uncertain and time-consuming; the fact that computer scientists wanted to go to court to gain the assurance that they wouldn’t be prosecuted suggests that they were quite worried about what would happen if they continued to innovate.  Lawsuits against direct infringers might deter conduct close to infringement, but they don’t deter innovation, except to the extent that innovation is funded only by infringing activity.  Only lawsuits against facilitators deter innovation that might facilitate legal uses as well as infringement. 

 

A final threat to innovation is more systematic.  Courts can see the advantages of well-established technologies, even if those technologies also facilitate infringement.  No court is likely to ban unlicensed printing presses or photocopiers, even though doing so might be a much more effective way of dealing with piracy than suing counterfeiters.  The social value of printing presses and photocopiers has become quite clear over time.  Further, they have become accepted as a part of the status quo, and banning them would look like a social disruption.  The same is likely true of the VCR: while it narrowly escaped being declared contraband in 1984, it is highly unlikely that any Supreme Court Justice would vote to outlaw the VCR today. 

 

New technologies, by contrast, are much more vulnerable to legal challenge.  In part this is because their ultimate value may not yet be clear; as noted above, the VCR is a good example of a technology that turned out to have substantially more value to society than was originally perceived.  It is also because stopping the deployment of a new technology won’t cause the disruption of settled expectations that rooting out an existing technology would.[159]  When courts shut down new technologies, the world may literally never know what it is missing.[160]

 

Traditional copyright law dealt with the risk of harm to innovation in the same way patent law still does: by sharply cabining the circumstances in which copyright owners could sue makers of technology.  The Sony decision set an intentionally tough standard for such suits; even if the seller of a device was otherwise guilty of contributory infringement, the court would ban a technology only if the technology was not even capable of a substantial noninfringing use.[161]  Recent developments have significantly undermined this rule, however.  The DMCA expressly rejected the “substantial noninfringing use” test in favor of one much more generous to copyright owners.[162]  And in Napster, the court appears to have radically rewritten the Sony test in a way that renders it inapplicable in virtually any case, including Sony itself.[163]  The result is that so far, at least, courts in digital copyright cases have shown little hesitation about banning technologies that clearly have at least some social value.[164]

 

            Suing facilitators reduces innovation.[165]  By the very nature of innovation, it is hard to quantify this harm.[166]  But it surely exists, and it must be added to the social harm caused by banning existing legal uses in evaluating the economic effects of permitting suits against facilitators.

 

            C.        What’s the Alternative? 

            The arguments in the preceding sections seem to create a classic policy tradeoff: suing facilitators is much more cost effective than suing direct infringers in the digital world, but it also causes social harm.  In order to decide whether suing facilitators made policy sense, the traditional approach would be to try to compare the magnitude of the benefit to the magnitude of the harm.[167]  David McGowan may well be right that this is an inquiry that will never have a definitive answer.[168]

 

            We don’t have to ask the question, however, if there are alternatives to suing facilitators that are cost-effective but do not create the same social problems.  In exploring potential alternatives, it is helpful to start with the basic economics of deterrence.  The foundational work in this field is Gary Becker’s analysis of the economics of criminal law.[169]  Becker’s insight is that a rational actor will adjust her behavior in response to the expected sanction – that is, the penalty that she will pay if caught multiplied by the probability that she will be caught.[170]  If the punishment for a particular bad act (say burglary) is set equal to the defendant’s gain from that act, the act will not be deterred unless the chance of being caught is 100%.  The intuition is simple: if the only cost to being caught is having to give up what you stole, a rational criminal will commit a burglary if there is any chance she might get away with it.  The corollary is that the more the sanction exceeds the defendant’s gain from conduct, the more rational actors will be deterred from engaging in crime, even if they are less likely to be caught.  If our burglar must pay a fine that is 10 times what she stole, she would be wise not to steal even if there is only a 10% chance of being caught.[171]  Indeed, Kaplow and Shavell extend Becker’s analysis by pointing out that from a cost-benefit perspective, the maximum possible sanction is the optimal one because it requires the fewest resources to implement.[172]

 

            Becker’s fundamental insight focuses on the chance of detection and the magnitude of the sanction.  Because he is working primarily with criminal law, this approach makes sense: those are the likely variables.[173]  To apply his model to digital copyright infringement, where private actors are the most likely enforcers, we need to make a few modifications. 

 

First, detection is not much of a problem in the online copyright environment.  While crimes are normally concealed, online copyright infringement generally is not.  Indeed, one of the overlooked benefits of the Internet for copyright owners is the ability it gives them to find infringers who would otherwise remain hidden.  Copyright owners are unlikely ever to catch an end user who makes a photocopy of a book, and it is even hard (though certainly not impossible) to detect traditional counterfeiters.  By contrast, a large percentage of the copyright infringement that occurs online is publicly searchable,[174] and copyright owners can more easily identify infringers.  In the digital environment, it is likelihood of enforcement that substitutes for likelihood of detection.  Copyright owners can find online infringers, but for the reasons we discussed in section A they have generally proven themselves unwilling to sue those infringers.  Becker’s point applies with equal force to defendants who know they will be “caught” but who do not expect to be called to account for their behavior.

 

Enforcement is unlikely in the digital copyright environment because the copyright owners would have to bear a litigation cost that exceeds the likely return to a lawsuit.  Becker’s framework largely ignores the transactions costs of litigation, because government enforcement is not sensitive to litigation cost in the same way that private litigants are.[175]  The cost of litigation affects the likelihood of private enforcement, though, and so the probability of “detection” in Becker’s framework is in fact a function of the costs of enforcement.[176] 

 

The second modification to Becker’s model concerns the costs of prevention.  Becker takes the background environment – the architecture of a city, for instance – as a given.  As Joel Reidenberg and Larry Lessig have made clear, however, that background environment is mutable online.[177]  Copyright owners who want to stop digital infringement need not sue more infringers or raise the sanction to infringers if they can change characteristics of the Internet itself in a way that makes copyright infringement more difficult.  One way to do this is to build copy controls into the digital media themselves.  This sort of “digital rights management” is increasingly common.  Another way to change the Internet environment is to sue facilitators.  If copyright owners can shut down p2p networks, or can enlist ISPs and search engines to filter their users’ content for copyrightable material, they may not need to enforce their copyrights directly at all.  The closest parallel to the traditional theory of crime would be a change in the architecture of a city – say, the creation of a gated community. Efforts to change the Internet itself by suing those who build or run pieces of it offer an alternative way to modify behavior in Becker’s model. 

 

Our modified analysis of the economics of deterrence suggests that there are several different legal ways to attack the problem of piracy in the online environment.  First, copyright owners can try to limit the ability of users to engage in infringement by changing the characteristics of the Internet itself.  This is the approach they have taken so far.[178]  Second, copyright owners can try to deter infringement by raising the cost to the infringers when they are caught.  This means either raising the punitive fine an infringer must pay or imposing a non-monetary penalty such as jail time.  Third, copyright owners can try to stop infringement by increasing the likelihood that infringers will be sued.  As we have seen, this means finding a way to reduce the cost of litigation to the copyright owner.  Economic theory suggests that copyright owners should be indifferent between these approaches at some level; the level of sanction or enforcement can be set to achieve any particular level of deterrence.  As we have seen, however, social welfare is not indifferent between these approaches.  Suing facilitators imposes collateral social costs that can be avoided either by raising sanctions or by lowering the cost of enforcement.

 

In Part III, we explore these alternatives in detail.  Before we do, it is worth emphasizing that the goal of any approach is not the elimination of piracy.  Piracy has always been a feature of the intellectual property landscape.[179]  The U.S. started life as a pirate nation,[180] and the content industries have complained for decades about the billions of dollars in revenues lost to piracy every year.[181]  Yet those same industries have survived and even thrived despite significant piracy.  The content industries have never had or needed perfect control over infringement.  They merely need enough control to give them sufficient incentive to create new works.[182]  This is not to condone piracy or say that it should not be minimized to the extent possible.  Rather, it is to make the point that weeding out all piracy simply isn’t cost-effective.  To try to give copyright owners perfect control would impose dramatic social costs to gain dubious benefits.[183]  In the context of online copyright infringement, the real policy question is how to bring piracy down to a manageable level akin to the rate of infringement in the traditional copyright environment. 

 

III.       Exploring Alternatives to Suing Facilitators

 

            In this Part, we consider two alternatives to suing facilitators in the particular context of p2p file sharing.  These alternatives build on the theoretical options taken from our modified Becker model: (1) raising the sanctions imposed on infringers and (2) lowering the costs of copyright enforcement.

 

            A.        Raising Sanctions

            In the traditional economics of deterrence, raising the sanction is a simple matter of increasing the legislated or judicially imposed penalty for a particular offense.  With digital copyright infringement, things are a bit different.  Copyright law already includes substantial supracompensatory sanctions in both civil and criminal law.  Any copyright infringer – even one who acts innocently[184] -- can be held liable for the plaintiff’s attorneys’ fees and subject to statutory damages in lieu of actual damages at the plaintiff’s sole election.[185]  Those statutory damages normally range from $750 to $30,000 per work copied in the court’s discretion.[186]  The court has the discretion to lower the amount to $200 per work for innocent infringers and to raise it to $150,000 per work for willful infringers.[187] 

 

These damage amounts reflect recent increases by Congress and dealing with infringement over p2p networks offers no reason to raise these damage amounts further.[188]  Because the most likely targets of a civil lawsuit in the p2p context are the “keystone” uploaders, who often have several hundred different songs on their computer,[189] existing statutory damages can easily run into the tens of millions of dollars per individual.[190]  This is likely to be an ample deterrent for the individuals who most often hold keystone positions on p2p networks.  Indeed, it’s arguably far too high already to do much good.  College students don’t have tens of millions of dollars to lose, and conversely those who do have that kind of money don’t tend to spend their time trading music files on p2p networks.  Civil suits with potentially enormous statutory damages may deter uploading because college students (or their parents) will fear bankruptcy.  Indeed, the RIAA was apparently able to eliminate a significant amount of file sharing merely by threatening to sue a few students.[191]  But if so, existing statutory damages will be more than sufficient to achieve that deterrence. 

 

College students are perhaps even more likely to be deterred by the prospect of going to jail.  Copyright law includes rather substantial criminal penalties, including prison time, for willful copyright infringement.[192]  Under the 1976 Act as originally enacted, copyright infringement was a criminal offense only if the defendant acted willfully and for purposes of commercial advantage or financial gain.[193]  Congress expanded criminal penalties rather substantially in the No Electronic Theft Act of 1997, however.  The law now provides that a willful infringer is criminally liable either if they act for financial gain, a term now defined to include the expectation that others will reciprocate by providing copies of other works, or if they reproduce or distribute works worth more than $1,000 retail value in any six-month period.[194]  This latter provision is likely to reach most keystone uploaders on a p2p network, so long as they act willfully.[195]  As with civil penalties, it doesn’t seem that the existing penalties need to be augmented.  Rather, the reason the already substantial civil and criminal penalties have not had a deterrent effect is that they have not yet seriously been pursued against alleged direct infringers on p2p networks.[196]  As Stuart Green put it, “if the state is serious about enforcing intellectual property laws, it cannot simply expect to impose harsh criminal sanctions, stand back, and wait for compliance.”[197]  Only in September 2003 did sound recording copyright owners begin to pursue civil infringement suits against individual p2p uploaders.  In this section, therefore, we consider whether a small number of high-profile civil suits against, or criminal prosecutions of, file traders would substantially reduce piracy.

 

The prospect of spending several years in prison or owing millions of dollars in damages is likely to serve as a substantial deterrent to digital copyright infringement by end users.  The more difficult empirical question is how many people the government must prosecute, or copyright owners must sue, in order to create a credible deterrent to illegal activity.  While we have no hard data on this point, we think a number of factors combine to suggest that the number of cases may actually be quite small. 

 

First, while there are a massive number of users of p2p networks like Morpheus and (before the injunction) Napster, the overwhelming majority of those users only engage in downloading.  Indeed, by one estimate, 3% of the users of a p2p network upload 97% of the files on that network.[198]  These high-volume uploaders are also the users most likely to be engaged in uploading illegal content, rather than providing access to legal files.  They are easy to identify, both because they will repeatedly appear in content searches and because many run so-called “supernodes” that facilitate fast downloads.[199]  Stopping piracy on a p2p network doesn’t require targeting downloaders, who may in any event have a legitimate reason for downloading some copyrighted content.[200]  It just requires targeting the much smaller number of high-volume uploaders.  If there are 3 million users logged onto Morpheus at any one time,[201] perhaps 90,000 of them are high-volume uploaders.

 

Second, high-volume uploaders are likely to be easily deterred.  They are not paid for uploading files, and indeed contribute substantial bandwidth and perhaps time on a voluntary basis in order to make files available to others.  They are persuaded to do so in part because the p2p community inculcates a “norm” of sharing,[202] though the fact that most people do not upload indicates that that norm is not a particularly strong one in the community at large.  But it is possible to participate in the p2p system without uploading, and the threat of bankrupting civil suits or criminal prosecution may induce a substantial number of high-volume uploaders to become passive downloaders instead.  This is particularly true with criminal prosecution because the sort of individuals who tend to be high-volume uploaders seem likely to fear jail more than the average criminal.  Willful digital copyright infringement is a crime committed primarily by college students: young, educated members of the elite with a bright future ahead of them.  The prospect of going to prison – and the attendant consequences, such as being kicked out of school – may worry a college student more than it would those inclined to commit other kinds of crime, such as burglary or assault.  The college student may feel she has more to lose and less to gain from the criminal activity.  And since she has no strong stake in being an uploader, she may simply decide to quit.  While it is only a guess, it might be reasonable to say that a 1% chance of criminal prosecution for uploading files in any given year would be enough to deter the majority of uploaders, and the same may be true for the chance of being adjudged liable for a multi-million dollar damage award.  This means that if we must deter 90,000 people, we need only prosecute or sue 900.

 

Even this number might overstate the number of suits or prosecutions needed to significantly reduce p2p piracy.  While it is possible that deterrence occurs only after a threshold – that is, that no one will be deterred by the threat of legal action until the chance of prosecution reaches 1% -- we think it more likely that deterrence is at least partially linear, because some high-volume uploaders are more risk-averse than others.[203]  Prosecuting fewer than 900 people – say, 300 – might deter some but not all uploading of illegal content.  Partial deterrence will not only reduce the piracy on p2p networks, but will also increase the burden on the remaining high-volume uploaders, as the mass of downloaders in a network shifts to the remaining uploaders.  The result may be a cascade effect, in which causing some uploaders to stop providing illegal content (and deterring others from starting to provide such content) imposes technical burdens that in turn cause more uploaders to drop off the network, further increasing the technical burden (and the percentage risk of prosecution) for the remaining uploaders. 

 

We can foresee three main objections to the use of criminal or severe civil sanctions to enforce the law.[204]  First, imposing such liability, especially criminal liability, on a few individuals in order to deter thousands of others may seem unfair to those who are singled out for prosecution.[205]  This unfairness may have no great legal consequence; selective prosecution occurs in a variety of fields and courts have consistently rejected constitutional challenges to the arbitrariness of making examples of a few defendants, at least where racial animus is not at issue.[206]  But it does put the burden of stopping piracy squarely on the backs of a few uploaders, rather than distributing it more evenly among the population of infringers, and many people might find that morally objectionable.[207] 

 

Second, the downside of effective deterrence is the risk of overdeterrence.  Criminal penalties are particularly likely in white collar cases to deter legal conduct that is near the borderline of illegality and may be wrongly perceived as illegal.[208]  In this case, however, we think the risk of overdeterrence is minimal.  We envision criminal prosecution or civil suits for significant monetary damages focused entirely on high-volume uploaders – say, those who upload more than 100 copyrighted songs.  It is highly unlikely that these high-volume uploaders are in fact engaged in legal conduct.[209]  If all high-volume uploaders are acting illegally, and if it is clear how to avoid being in that category, overdeterrence doesn’t seem a significant problem.  It is somewhat more likely that courts will err by punishing high-volume uploaders who are not in fact willfully infringing copyright, but who instead genuinely believe that their conduct is legal.  While this would be a miscarriage of justice, since willfulness is an element of criminal copyright infringement and of enhanced statutory damages,[210] it will not create a deterrence problem.  Those who don’t know they face criminal prosecution or higher damage awards can’t be deterred by the threat of that prosecution or those damages.  Finally, as with any criminal law, mistaken prosecutions will impose significant costs on those wrongfully targeted.[211]  Mistakes will certainly be made, though the straightforward nature of the case and the detailed electronic trails file transfers create may actually make the risk of mistaken prosecution rather small. 

 

Third, criminal prosecution requires the initiative of U.S. Attorneys, and they may find the prospect of prosecuting college students for uploading music politically unpalatable.[212]  This isn’t really an objection to the application of criminal liability as much as skepticism that criminal sanctions will really be enforced.  It is true that a large number of people participate in p2p file sharing, and it is possible that they would protest criminal prosecutions, making the person who brought those prosecutions unpopular.  On the other hand, some of the most powerful lobbying groups in the world are behind stronger criminal copyright enforcement.  They managed to persuade Congress to pass the NET Act, strengthening criminal penalties and expanding the definition of criminal copyright infringement.  More recently, a number of Congressional representatives have on two different occasions taken the Justice Department to task for not enforcing the NET Act,[213] suggesting that there is substantial political will in favor of criminal prosecution.

 

Still other objections to criminal prosecution or severe civil penalties stem from broader objections to the enforcement of copyright law in the digital environment.  If you believe copyright law on the Internet in general is a bad idea,[214] or that p2p file sharing should be legal,[215] it follows that you wouldn’t want to see criminal prosecutions of, or substantial monetary penalties for, uploaders.  From the perspective of those who believe in the copyright system, and believe that large-scale file sharing is illegal, however, criminal prosecutions or very large statutory damage awards offer the advantage of dealing with piracy without censoring innovation.  They have disadvantages too, however.  Most notably, it seems unfair and disproportionate to impose the burden of stopping piracy so heavily on a few unlucky defendants.  Further, using criminal or civil enforcement against infringing p2p users would – if it works – over time effectively shut down p2p networks all or almost all of whose content is illegal, since without uploaders there can be no downloaders.[216]  Only those networks that were put to legal as well as illegal uses would be likely to survive.  Thus, enforcing the copyright law against direct infringers may have the effect of eliminating the efficiencies associated with the p2p distribution system.  Because of these shortcomings, in the section that follows we examine alternative methods of targeting enforcement at direct infringers rather than at intermediaries.

 

            B.        Lowering Costs

            A more palatable alternative to raising sanctions by putting a small number of college students in jail (or bankrupting them) in order to deter their peers is to lower the cost of enforcement.  Suing most or all direct infringers currently isn’t attractive because litigation is so expensive and time-consuming.  If enforcement is quick, cheap, and certain enough, the sanction for infringement doesn’t need to be very high in order to achieve the same deterrent effect.  In this section we discuss two possible systems for lowering enforcement costs: an automated compulsory license system implemented though a levy, and a streamlined online arbitration system for resolving copyright disputes.

 

            By any count there are a lot of infringers online.  The task any civil enforcement approach must confront, therefore, is to permit copyright owners to sue enough of these infringers to reduce the problem to manageable proportions[217] without imposing extraordinary costs on the copyright owners.  If we are not to raise the sanction for enforcement to punitive levels,[218] this means lowering the cost of enforcement against individuals to such a degree that copyright owners can cost-effectively pursue tens or even hundreds of thousands of them.

 

                        1.         Levies

            One possibility is to do away with lawsuits altogether in favor of an ex ante enforcement mechanism.  In an important paper, Neil Netanel has proposed that copyright be enforced online in this context through a system of levies.[219]  A levy is a form of compulsory license.  Rather than requiring copyright owners to identify and sue those who owe the license fee, however, the levy would be automatically collected on the sale of software, services or hardware that are likely to be used in infringement.  A similar system is being implemented in Germany and elsewhere in the EU,[220] where purchasers of computers pay a set fee (currently 12 euros) into a fund designed to compensate copyright owners for infringement.[221]  And there is precedent in U.S. law: the Audio Home Recording Act provides for a levy to be charged on all blank digital audio tapes and digital audio recorders, with the revenue to be allocated among music copyright owners.[222]  The AHRA hasn’t seen much use, but that is because digital audio tapes never caught on.

 

            Levies of the type Netanel has proposed offer substantial advantages over the existing regime of secondary and tertiary liability.[223]  They are likely to catch virtually all copyright infringers and force them to pay what amounts to a small license fee.  Because they operate automatically, they can be enforced at a minimum of cost.[224]  And because they replace the existing scheme of legal enforcement, they permit society to make use of the existing, efficient p2p networks to distribute digital content online.  In essence, a levy co-opts illegal file-sharing by charging a fee and then declaring it legal.

 

            A hybrid levy approach would not impose the levy by law on all devices, but would permit facilitators who might otherwise fear indirect liability to buy immunity by paying a levy for their users.  If Grokster paid a levy for each copy of Morpheus software that it disseminates, for example, the company could avoid being sued for facilitating infringement by users of its software.  Companies that specialized in facilitating music downloads would want to pay the fee, since they would face liability under traditional principles of contributory or vicarious infringement.  By contrast, companies with rather less connection to infringing activity could opt not to pay the levy, gambling that they are not infringing. 

 

            Still other proposals that seem to have little in common with levies operate on the same basic principle.  Landes and Lichtman argue for a negligence standard, under which Internet service providers that facilitated copyright infringement by failing to design their system to avoid it would have to pay the damages caused by that infringement.[225]  Presumably a true liability rule, unlike current copyright law, would permit neither injunctive relief shutting down the facilitators nor supracompensatory remedies like statutory damages.  Such a liability rule, like a levy, is a tax imposed on facilitators to pay for the harm they cause to the incentives of copyright owners.  And a liability rule, like a levy, is in effect a compulsory license, with all the problems those entail.[226]  The difference is that the tax is calculated ex post by a court, rather than ex ante by an administrative agency.

 

            One problem with levies is that, like suing facilitators, they target upstream technologies rather than the people doing the infringing.  Indeed, imposing a levy is economically quite similar to suing facilitators – the levy just substitutes a liability rule and a collection mechanism for copyright law’s existing property rule.  To make the levy small, it has to be imposed on a wide range of devices (say, all computers or all modems or all ISP service agreements).  But a levy charged on a range of devices with multiple uses is really just a tax on those devices, paid by both those who download music and those who don’t.  This is akin to a tax on innovation in the Internet environment.  This tax may be better than suing innovators under a property rule, because copyright owners won’t have the power to ban innovation outright, but taxing innovation will naturally discourage it.

 

            Levies will likely have other consequences as well.  If a levy is charged on a single device or service (say a computer or an ISP account), and if paying the levy makes downloading content legal, the levy will create moral hazard problems.[227]  There is every incentive to download as much music as possible if you are paying a flat rate.  Copyright owners may compensate for this problem by setting the rate relatively high, doing further damage to innovation and discouraging casual users from buying the device or service at all.  Further, this flat rate charge will eliminate or at least sharply limit the role of authorized musical services provided by the content owners.  This may not be a problem – as noted above, there are reasons to think that p2p networks distribute content more efficiently than the copyright owners would – but if you think top-down networks are preferable, the fact that p2p networks will replace them is an additional cost.  Finally, a levy requires money to be collected by a facilitator at the point the device, program or service is provided.  This channels innovation in the relevant market into pay form and centralized software distribution, since someone must collect the tax.  It also discourages anonymity.[228]

 

            A better-designed levy would be closely tailored to acts of infringement.  A partial step in this direction would be to charge a per-use rather than a flat rate fee.[229]  Charging a levy on every megabyte downloaded, for example, might correlate reasonably well with copyright infringement, and it would solve the moral hazard problem described above.  Such a bandwidth tax would still have rather severe effects on certain types of innovation, however, notably those that involve high-bandwidth uses of the Internet.  And discouraging the fledgling broadband Internet market seems a bad idea, given the lengths policy makers are willing to go in other circumstances to encourage broadband rollout.[230] 

 

            Charging a levy only on acts that would otherwise be infringing – say, a fee per mp3 file downloaded without authorization from the copyright owner – would be ideal from an innovation standpoint, since it would distinguish between legal and illegal uses of a device or service.  But such an approach would obviously create serious monitoring problems.  Indeed, it doesn’t make sense to talk about a well-targeted levy as a levy at all. Instead it is a compulsory license dependent on identifying and collecting from infringers.[231] 

 

                        2.         An Arbitration System

            Instead of a levy system, another possible alternative for lowering enforcement costs for copyright owners would be to make dispute resolution by copyright owners against direct infringers quick and cheap, so that copyright owners would be more inclined to pursue such direct infringers instead of suing facilitators. 

 

Is it possible to make such dispute resolution quick and cheap?  Traditional arbitration is neither.  There is, however, an online model in the Uniform Dispute Resolution Policy (UDRP) for Internet domain name trademark disputes,[232] which has resolved 10,000 domain name trademark disputes in 3 years, at a cost of $1200 each and little more than a month of time.[233]  The UDRP is designed to resolve only straightforward cases of bad-faith cybersquatting, and to reserve for the court system difficult factual and legal disputes between parties with competing and arguably legitimate claims to the same domain name.[234]  For those straightforward cases of cybersquatting, there are unlikely to be significant factual or legal disputes that need resolving.  A panelist given the basic facts can make a decision fairly quickly.  Like the UDRP, a copyright arbitration system, if properly conceived, could target straightforward conduct that is unlikely to have legitimate justifications, such as high-volume uploading of copyrighted works to p2p networks.  Assertion of a plausible factual or legal dispute – evidence suggesting that the works in question weren’t copyrighted, or weren’t copied, or that the use is fair – should result in denial of the copyright owner’s claim without prejudice to her ability to bring a lawsuit. 

 

Our analogy to the UDRP will raise some people’s hackles.  The UDRP has some serious structural problems.  It lacks some essential due process protections, like an administrative appeal, a fair system for assigning judges, and a penalty for overreaching by complainants.[235]   But these problems can be solved in the copyright context by learning from the problems with the UDRP.  A digital copyright arbitration process could select judges in a fair and balanced® way.  It could permit an administrative appeal.  And it could impose some sanction on frivolous or bad-faith claims made by copyright owners.[236]

 

There are two fundamental differences between the factual setting of the UDRP and the digital copyright cases an arbitration panel would likely be called upon to resolve.  First, the domain name at stake in the UDRP is ultimately under the control of ICANN.  As a result, a UDRP panelist does not have to collect money or property from a losing domain name registrant; she merely needs to instruct ICANN to transfer ownership of the domain name to the trademark owner.  There is no similar control over digital copyright infringers.  A copyright arbitration system needs a substitute sanction, such as an award of money damages or a reliable way to remove infringing material or the infringer herself from the system.  Further, if the sanction is something that must be collected from the defendant, the arbitration system will need some way of ensuring compliance.

 

            Second, the UDRP is imposed by ICANN on all registrars, who impose it by contract on all registrants.  It requires contracts with and reliable identification of users.  There is no central authority that contracts with Internet users generally.  Binding Internet users to arbitration procedures would require them to contract with their ISPs or with providers of specific services, and there is no entity akin to ICANN that has contracts with all the ISPs and could impose this requirement on them. Further, reliable identification of Internet users may be undesirable for privacy or other reasons.[237] 

 

The arbitration system we envision for copyright would not be mandatory like the UDRP.  Instead of requiring it, we envision a system in which facilitators in an ongoing service relationship (Napster, Morpheus, ISPs) with an Internet user accused of copyright infringement could gain statutory immunity from infringement in exchange for imposing and administering such an arbitration system.[238]  Entities who fear the imposition of copyright liability could opt into this statutory immunity by establishing and enforcing an arbitration system for copyright disputes that met minimum statutory standards.  These compliant entities would be responsible for collecting accurate data permitting them to identify the users accused of copyright infringement[239] and to enforce the judgment of the arbitrator.  Depending on the sanction in the arbitration system, compliant entities may need to find a way to remove material or users from the system, or may need a way of charging modest sums to the defendant’s account.[240]

 

Not all facilitators will need or want to participate in the arbitration system we describe.  Facilitators who believe they will not be held liable for copyright infringement under the standards of secondary liability can take their chances with the law; for many classes of facilitators (such as search engines and backbone providers) they will likely prevail under existing law.  Facilitators who wish to preserve the anonymity of their users may opt out of the system,[241] but in doing so they forego our proposed statutory immunity.  This does not mean that they are necessarily engaged in infringement; it simply means their conduct must be evaluated under existing legal standards.

 

An arbitration system might have an important collateral benefit in implementing the DMCA safe harbors.  The DMCA grants safe harbors to ISPs only if they have a policy in place for terminating the accounts of “repeat infringers.”[242]  No one seems to know what makes one a “repeat infringer,” however.[243]  Copyright owners have read the term broadly, to include anyone who is the subject of two DMCA notifications, and possibly even anyone who has posted two or more allegedly infringing works at one time.[244]  It seems wrong to say that one is an infringer merely by virtue of receiving a cease and desist letter, which digital content owners have been sending with reckless abandon and which need not even meet the standards of Rule 11.[245]  The other extreme – that one is not an infringer until adjudicated so by a court, and so repeat infringers must be sued and lose twice – seems equally unworkable.  The arbitration system may provide some middle ground here, by allowing a quick determination by a neutral third party that an individual is in fact an infringer.  Keying the termination obligation to an arbitration finding would protect the due process rights of those wrongfully accused of infringement without rendering the repeat infringer provision altogether ineffective.

 

            C.        Pursuing Multiple Approaches

            As the discussion in the previous sections suggests, none of these three approaches to improving copyright owners’ ability to enforce their rights against direct infringers rather than against suppliers of products or services that can be used for infringing and noninfringing purposes is perfect.  Each seems most workable in particular contexts, and implementing all of these approaches may be the best way to give copyright owners effective relief without overly hampering innovation in independent but copyright-related technology.

 

            The arbitration safe harbor we have described seems best suited to p2p service providers whose operations involve some centralization, providers such as Napster and Aimster.  Such services are likely to have the ability to bind users to the dispute resolution process through their service agreements.  Perhaps more importantly, these services will likely be able to exclude individual users from participation in their p2p networks.  In the case of Napster, for example, because the system operated by maintaining a centralized directory of files available on users’ computers, users had to connect to Napster’s centralized directory in order to locate other users and their files.  As a result, Napster was in a position to screen users when they attempted to connect and to select which users could or could not access the directory.  In the context of an arbitration safe harbor, this ability to exclude could provide an effective sanction, essentially equivalent to injunctive relief, against a user found by the dispute resolution process to be using the system to commit copyright infringement.[246]  Even if a copyright owner were unable to collect any monetary penalties imposed in the arbitration process, she would at least be able to have the infringer removed from the p2p network and would get the benefit of the deterrent effect that such removal would have on other potential infringers.

 

            While p2p technology has in recent years shifted away from centralized services, it is not clear how much of that shift has been driven by the technical superiority of decentralized services and how much is due to the legal liability imposed by Napster on operators of centralized p2p services.[247]  The popularity of Napster during its heyday suggests that centralized p2p networks may well be viable technological and business models in the absence of the prospect of liability for all infringing use by network users.  An arbitration safe-harbor could provide clear limits on that liability and allow centralized p2p networks an opportunity to prove themselves in the marketplace.

 

            Another important category of p2p facilitators probably cannot feasibly comply with the requirements of an arbitration safe-harbor from copyright infringement liability: companies who simply provide users hardware or software for p2p file transfers without entering into a continuing relationship with those users.  In this type of p2p network, the software facilitates file transfers between software users without any need to connect to any computer operated by the creator of the software.  As a result, someone who creates or disseminates such software cannot subsequently control which users of the software can engage in file transfers over the p2p network.  The software at issue in the Grokster decision is an example.  As the district court pointed out in that case, “Users connect to the respective networks, select which files to share, send and receive searches, and download files, all with no material involvement” of the companies that wrote or distributed the software to the users.[248]

 

            Such providers of p2p software are unlikely to be able to take advantage of an arbitration safe harbor such as the one we have sketched.  It is not clear how the provider could bind users of the software to participate in the arbitration system once they have acquired the software, any more than, for example, Adobe Systems could bind users of its Acrobat software to a dispute resolution system for claims that users are creating and disseminating infringing copies in Acrobat’s portable document format.[249]  More importantly, unlike providers of p2p services with some centralized component, the mere distributor of software cannot bar individual infringing users from the network.  Thus, even if a copyright owner pursued a claim against such a user through the safe harbor’s arbitration system, the potential for effectively enforcing an arbitration decision against the user is much lower. 

 

            For p2p innovators for whom the arbitration safe harbor is unlikely to be a practical option for avoiding secondary liability for user infringements, a levy system would provide a more feasible alternative.  The disseminator of p2p software could, for example, pay a levy on each copy of the software disseminated (or perhaps enabled).  The software provider could collect the levy from the user and remit it to the collection agency, or it could distribute the software to users for free and pay the levy out of revenues it earns by other means (such as advertising) on the distribution or use of the software.  Any user of a copy of the software on which the levy had been paid would not be subject to suit for copyright infringement for use of the software, and the distributor would not be subject to secondary liability for such infringement. 

 

            As noted above, however, a levy system would likely channel p2p innovation into centralized, fee-based distribution of software and would require payment not only by those who use the software for infringing purposes but also by those whose use of the software does not infringe.  It seems preferable, therefore, to allow p2p providers to obtain immunity from claims of secondary liability for copyright infringement by paying a levy, but not to require them to do so. Such a targeted levy will probably be fairly high; levies at the low level Netanel suggests[250] are possible only if they are applied widely to a number of devices that are not in fact used to infringe copyright. But the p2p software providers who will opt into such a targeted levy system are those who reasonably fear that they would not be able to avail themselves of the Sony defense because they lack a substantial noninfringing use.  It seems reasonable to impose a fairly high levy on such software. 

 

            Making a levy a condition to immunity from suit rather than mandating it means that some providers of p2p software will not be immune under either the arbitration safe harbor or the levy system, and that copyright owners will not have a lower-cost enforcement alternative to copyright litigation against users of that software.  In those cases, protecting innovators against secondary liability claims should rely on the remaining strategy discussed above:  raising the effective sanction on infringers, rather than lowering the cost of enforcement against each infringer.  Courts should vigorously enforce the Sony doctrine that the producer of a dual-use technology is not liable for infringements committed by users of that technology merely on the basis of supplying the technology to users.  But copyright owners could instead pursue severe civil remedies, and convince the government to pursue criminal penalties, against the keystone uploaders who make widespread dissemination of copyrighted works over p2p networks possible.  Our analysis above suggests that such suits against a limited number of uploaders could be effective in deterring a much broader group of potential infringers and reduce infringing activity enough to allow copyright owners the opportunity to earn a return on their investment in the creation and publication of their works.

 

            D.        Can Enforcement Work on the Internet?

            A policy of targeting direct infringers is workable only if those direct infringers can be found and brought to justice.  Surely, one might object, the Internet makes this infeasible. After all, among the most celebrated characteristics of the Internet are its international character and the potential for anonymity.[251]  On the Internet, the saying goes, no one knows you’re a dog.[252]  A common argument against enforcement of intellectual property law online has been that infringers will simply move offshore,[253] or conceal their identity using unbreakable encryption.[254]  If enough p2p uploaders do so, enforcement targeted at those uploaders will fail.

 

            International enforcement and anonymity will undoubtedly limit the efficacy of efforts to target direct infringers on p2p networks to some extent.  We are skeptical that these effects will be substantial enough to prevent effective antipiracy enforcement, however, for three and a half reasons.  First, the same objections can be made to efforts to sue facilitators.  If individual uploaders can move offshore or conceal their identity, so too can at least some facilitators.[255]  Indeed, these challenges to enforcement are probably more daunting when suing facilitators than when targeting direct infringers.  A company that expects to be sued has significant incentives to incorporate in a foreign jurisdiction, and perhaps the resources to do so.  A college student is unlikely to move overseas in order to be able to continue to upload music.  If facilitators are as much or more likely than individuals to escape justice in these ways, these concerns don’t justify a policy of suing facilitators rather than individual uploaders.

 

            Second, neither anonymity nor moving offshore is nearly as easy to implement as first-generation Internet scholars have predicted.[256]  To take the more obvious case first, physical relocation to another country imposes huge burdens.[257]  Existing individual uploaders are unlikely to view the “benefits” of providing digital music for free to strangers as worth the major economic, social and cultural costs of a move.  This leaves the possibility that new uploaders will spring up in those countries to take the place of those who are deterred.  This is a potential risk: traditional (non-digital) piracy is much more common in non-WTO countries than in more developed nations, and the Internet may permit that higher rate to be transmitted internationally to the developed world.  But even this risk seems overstated, because many of the countries that lack modern copyright laws also lack the wealth and Internet infrastructure to support a large number of uploaders.  Nor will geographic indeterminacy permit uploaders to evade local laws; most IP addresses can ultimately be traced via the ISP or network owner to a user who has provided a geographic address, and in any event a wealth of new technologies permit geo-location of Internet addresses.[258]

 

            Anonymity doesn’t require physical relocation, but it does have social costs.  True anonymity precludes building reputation or establishing repeat play, because if you are truly anonymous you cannot prove you are the same person who logged on yesterday.[259]  It is difficult (though as Michael Froomkin has shown not impossible)[260] to maintain a consistent pseudonymous persona that can build reputation and be found repeatedly online by others, but that cannot be traced to an individual who might be held legally accountable.  Such untraceable pseudonymity[261] requires both the faithful use of unbreakable encryption and the presence of intermediaries willing to provide anonymous remailing services to parties unknown to them.  Unbreakable (generally public-key) encryption is currently available, but it is not widely used, though less powerful forms of encryption such as Secure Sockets Layer is a common feature of ecommerce.[262]  Such remailing services do exist, but they are hardly common.[263]  In part this is because the only way such an entity could get paid is through anonymous, untraceable digital cash, and digital cash never took off.[264]  It is certainly true that high-volume uploaders today aren’t anonymous.  We think it unlikely that the infrastructure exists today to support a widespread shift to anonymity in response to a new copyright enforcement initiative.[265]  Further, while there are some efforts to develop anonymous file-sharing in response to industry lawsuits, those efforts may be self-limiting.  By eliminating any trusted intermediary, most truly anonymous approaches to file-sharing will emphasize sharing among small groups of friends, rather than open sharing with strangers.[266]  They will also be more vulnerable to interference by copyright owners who sabotage the network by offering fake files.[267]  This may produce true anonymity, but at the cost of creating mini-networks that don’t scale.  From the perspective of copyright enforcers, this would be a major victory.

 

            Third, even if a significant number of high-volume uploaders seek to escape legal accountability by moving offshore or distributing files anonymously, it is not at all clear that those strategies of evasion will prove effective.  At the outset, it is worth noting that as of April 2003, 146 countries have adhered to the WTO’s TRIPs Agreement and 151 have adhered to the Berne Convention as of June 2003.  Fleeing to these countries (which includes every major nation in the world) won’t help; the law in those countries permits domestic enforcement against high-volume uploaders.  Proposed international conventions would also permit the enforcement of national copyright judgments in any member country.[268]  And going abroad may not even be necessary to pursue foreign infringers—the recent history of digital copyright enforcement suggests that the reach of U.S. law is long indeed.[269]  U.S.-based content industries have managed to obtain civil jurisdiction over companies based in Nevis, the Netherlands, Australia and Vanuatu,[270] and pushed the U.S. government to bring criminal prosecutions of a Russian company and software engineer.[271]  Courts in France and Australia have also subjected U.S. citizens to jurisdiction.[272]  And even if an uploader resides in a jurisdiction that will not enforce copyright, that uploader needs to worry about criminal as well as civil liability every time she travels to one of the many countries that do enforce copyright.[273]  Dmitry Sklyarov found this out to his detriment when he came to the United States to present a paper and was arrested for violating the DMCA.[274]

 

            Nor is anonymity of the kind commonly used necessarily a barrier to prosecution.  The most famous “anonymous remailer” of the 1990s, anon.penet.fi, disclosed the name of one of its users and folded its service in response to a search warrant issued by the Finnish government.[275]  “Anonymous” posts in chat rooms or on ISP bulletin boards also have proved thin protection.  In an important decision of first impression, the court in In re Verizon Internet Services held that under the DMCA an Internet service provider must disclose the identity of the owner or user of a particular IP address.[276]  Truly untraceable anonymity will permit practitioners to avoid prosecution, but what often passes for anonymity on the Internet today will not.

 

            Finally, one way to structure an arbitration system along the lines we proposed in the last section would be to condition facilitator immunity on the prohibition of anonymity.  This approach wouldn’t ban anonymity,[277] instead permitting facilitators to opt out of the arbitration system, albeit at the cost of losing the immunity promised to facilitators who identify users and contractually bind them to arbitration.  We call this a half of a reason because it applies only to arbitration, not to enforcement against direct infringers.[278]  We aren’t convinced that such a condition is necessary, because we aren’t convinced that true anonymity will prove much of a barrier.  But if it does, designing the arbitration system to force disclosure of identity is a potential solution.[279]

 

            None of this is to suggest that Internet enforcement will be perfect.  Indeed, it is reasonable to assume that some high volume uploaders will live in countries where copyright enforcement is impractical, and that some uploaders will be able to maintain true anonymity.  Further, we must consider not just how file-sharers behave now but how they might respond to more aggressive enforcement.[280]  If file-sharers really care about uploading rather than just downloading digital music – something about which we are skeptical – they may make a significant investment in anonymity to preserve their right to do so.[281]

 

But as we noted above, perfect enforcement isn’t the goal.  Antipiracy enforcement has never been perfect.  The real question is whether the characteristics of the Internet prevent enforcement from being “good enough” to allow copyright owners to reap a reasonable profit.  We think the answer is “no”—or at least likely enough to be “no” as to make enforcement against direct infringers worth a try.

 

            Indeed, imperfect enforcement may have affirmative social benefits.  First, digital distribution by consumers of some classes of works over p2p networks may well turn out to be legal.  The obvious example concerns out-of-print works not made available on official digital distribution sites.  Because of the extraordinary length of the copyright term, there are a large number of such works languishing in the clutches of apathetic copyright owners.[282]  A court may well conclude that reproduction of such works is fair use because it causes no economic harm to the copyright owner, who after all is not exploiting the work, and permits distribution of a work that would otherwise be lost.[283]  p2p networks may serve a useful function in distributing this copyrighted content.[284]  A second argument is broader.  A number of scholars have argued that as digital distribution takes hold, more and more content owners will voluntarily distribute their works for free online,[285] or alternatively that digital distribution companies will come up with a way to compensate authors at least as well as the rather inefficient system of middlemen.[286]  If this is true, a service that starts out as a haven for piracy could over time turn into a legitimate means of distributing content as artists voluntarily embrace it.  This may or may not happen.[287]  But shutting down p2p networks outright will terminate this experiment before it has a real chance to get going.  By contrast, targeting direct infringers may actually facilitate the transfer from illegal to legal uses precisely because it doesn’t shut the network down all at once, but removes illegal content a bit at a time.  The resulting demand for legally-distributed content (from independent bands and filmmakers, of out-of-print works, and the like) may or may not be enough to support a robust online distribution network.  But we will at least get a chance to find out.

 

IV.       Conclusion

 

            Copyright owners sue facilitators online because it is cheaper and easier than suing direct infringers.  Cheaper and easier doesn’t necessarily mean more efficient, however.  The shift towards suing facilitators who are further and further removed from the act of direct infringement imposes substantial social costs on both legitimate users and on innovation, costs the copyright owners do not have to bear.  The solution is to change the economics of targeting direct infringers.  One way to do this is to enforce civil and criminal copyright statutes against high-volume uploaders.  Such enforcement would likely have a substantial deterrent effect on uploading illegal files, though it may have undesirable social or moral consequences.  Alternatively, we could reduce the cost of targeting direct infringers by imposing a levy on the technology they use or by inducing them to agree to a low-cost, quick arbitration system.  As to the latter option, recent experience with such a system in the Internet domain name context suggests both that it is workable and that careful attention must be paid to process concerns in designing such a system.  An arbitration system could be set up to permit facilitators to “opt in,” agreeing to obligate their users to participate in exchange for immunity from liability for indirect infringement.

 

            None of these approaches is perfect.  Further, none will stop the demand for digital content, and so none will work unless accompanied by a serious, sustained effort by copyright owners to offer digital content online in legal form.  But the approaches we discuss are better than the alternatives: quashing innovation by expanding the scope of indirect liability, or doing nothing in the face of rampant digital copyright infringement.  The Internet has changed the economics of copyright enforcement irretrievably.  It is imperative that policy-makers set legal rules with these economic developments in mind.