Protecting Sony and the Internet:

A Discussion and Critique of Imposing Harsher Secondary Copyright Infringement Rules to Inhibit Peer-to-Peer File-Sharing

Derek Slater

01-02-03

Today, copyright holders are increasingly viewing the digital age as a choice between peer-to-peer (P2P) and pay-to-play. P2P file-sharing applications allow users to search each other's hard drives and, at the click of a button, initiate a file transfer between computers. Over 60 million Americans[1] and many more around the world are downloading files, primarily in ways that infringe copyright.[2] Initially, copyright holders were able to sue P2P providers successfully under the theory that, since the system operators had some level of control over the network and its users, they should be responsible for preventing infringement. In particular, the record industry was able to shut down the infamous Napster,[3] which boasted access to over 1.3 billion songs before it was shut down.[4]

However, Sony v. Universal, a Supreme Court ruling from what seems like a bygone technological era, has protected "decentralized" P2P systems, in which the P2P providers have no control over the user after distributing the software.[5] In 1984, the Supreme Court ruled that Sony, a VCR manufacturer, was not liable for its customers’ infringing uses because the device was "capable of substantial non-infringing uses."[6] In spite of potential harm to copyright holders, the Court feared that liability in this case would unduly constrain development and legitimate use of new technologies. That same fear is present in today's P2P cases.

Regardless, many scholars have argued that even decentralized P2P systems should be prohibited. To them, the systems exist primarily to aid infringement of a proportion that far exceeds what was possible with the VCR; in turn, preserving copyright in the Internet age requires creating harsher liability rules.

In this paper, I will argue the opposite: that Sony's approach to liability is needed more than ever to protect the Internet. Rules proposed to inhibit decentralized P2P would have sweepingly harmful effects on the Internet as a whole and thus should be rejected. In particular, I will focus on a negligence rule, first, because a form of it came up in recent P2P ruling, In re: Aimster,[7] and, second, because it incorporates features common to many standards that go beyond Sony and would endanger decentralized P2P.[8]

My purpose here is not to hash out all the nuances in secondary copyright liability, nor will my argument rely on court precedent. Rather, I seek to support Sony's overall approach and explicitly link it to protecting the Internet. Though similar arguments about Sony and alternative rules have been formulated in recent P2P cases, they have lacked this particular perspective.[9] More narrowly, by placing rules that could limit decentralized P2P outside the range of possible policy options, I seek to shape the framework for further debate on how copyright should be preserved while protecting the Internet.

My argument will proceed in the following steps. First, I will explain what underlies the Internet as enabler of innovation. Then, I will address how Sony is technology-protective and aligns with what the Internet demonstrates about promoting innovation. Moving towards the P2P context, I will note that the stakes involve not just the protection of innovation, but of speech. Against this backdrop, I will briefly describe A&M Records v. Napster and MGM v. Grokster, which protected Sony to a significant extent, as a lead up to an in-depth critique of the negligence rule. Finally, I will conclude by considering policies that could protect copyright and the Internet.

How the Internet Enables Innovation

Though the Internet allows for illicit copying outstripping that which the VCR enabled, it also has far more socially beneficial uses. The Internet has enabled untold innovation, ushering in a litany of “news”—new economy, new politics, new communities, new communication. In this section, I do not seek to discuss all of the Internet's potential benefits. Rather, I seek to explain why innovation thrives on the Internet. In short, it thrives because the Internet is open to all applications and innovators. Inasmuch as this is true, the Internet demonstrates the value in minimizing control over technology creation, enabling it to develop in unpredictable ways.

Cyberlaw scholar Lawrence Lessig argues that the Internet's “end-to-end” design is critical to supporting innovation online. The computers within the Internet, the ones that transport data from your computer in Boston to another in Taiwan, are “dumb”; they route information as best they can, “perform[ing] only very simple functions that are needed by lots of different applications.”[10] The “intelligence” in the network exists at the “ends,” the applications residing on the computers that access the network.

Lessig notes two main consequences of this design.[11] First, the network is neutral, open to any type of application. Because the Internet is designed in this way, it allows for as yet unimagined technologies to be implemented. Second, innovators can create without asking anyone’s permission or changing the network itself. Rather, someone at an “end” must simply create an application that comports with the Internet’s “simple functions.” The result is what Lessig terms an “innovation commons," a resource that others are free to build on.[12]

For those reasons, an array of technologies can all coexist together. For instance, let us compare two models often used for file-distribution, client-server and P2P. Client-server, the approach for a typical website, is hierarchical—individuals request information from the server, a computer dedicated to sending out that information with a stable address and connection.[13] On the other hand, in P2P, users are suited to both receive and send information to each other,[14] have "significant autonomy" from central servers, and can have "variable connectivity and temporary network addresses."[15]

P2P file-sharing systems can be implemented in various ways, separated broadly into centralized and decentralized models. Centralized P2P, like Napster, requires users to connect to some central location first before transmitting and receiving data from peers on the network.[16] Though the former aspect mimics the typical client’s dependence on a server, the actual file transfers occur between two individuals using the system. To avoid any dependence, a decentralized design is necessary. To connect to the decentralized Gnutella network, for instance, one connects to another person using it, who is connected to others using it, and so on.[17] When you search for a file, the request spreads from person to person, not through any centrally controlled apparatus. In turn, no one can shut down file-sharing using Gnutella without stopping each individual.

Why is this multiplicity of possible designs important? Lessig suggests that the “commons” approach is most beneficial when we cannot predict the best use for the resource.[18] If we knew ahead of time that, for example, a client-server model that transported a certain source’s data fastest was the most socially valuable use of the network, we could redesign it accordingly. But because that best use is not known, end-to-end is superior.

By letting all models flourish, all potential benefits can be reaped. People can experiment, and neither the Internet nor any central authority picks a winner ahead of time. In a time dominated by client-server tools, P2P itself was an unexpected technology, developing rapidly only in the last few years for file-distribution and many other applications.[19] While the client-server and P2P models each have their own advantages and disadvantages,[20] there is no way of knowing them precisely ahead of time. For example, though some state that decentralized P2P file-sharing will never work efficiently with numerous users,[21] others are trying to fix this "scaling" problem.[22]

Indeed, these technological architectures and associated software tools do not align with narrowly-defined activities. Because the network itself has general purposes, it lends itself to the development of many multipurpose tools. It would be wrong, for instance, to think of Gnutella as only a tool for infringement, for it can be used to transfer any type of file. It would also be incorrect to consider it strictly as a P2P file-sharing application. Gnutella is not a discrete program, but rather a protocol for exchanging information that many types of software can communicate with and that can be implemented in many different ways.[23] In fact, early versions of Gnutella and current World Wide Web applications use the HyperText Transfer Protocol (HTTP) to communicate information,[24] and Gnutella has been implemented as part of a Web-based search engine.[25]

The value of multipurpose tools is similar to that of the Internet itself. The analogy to the Internet is imperfect inasmuch as these devices are not truly “commons”es; however, the beneficial flexibility for users is similar. Technology creators can make tools to which users can put to any number of uses, some of which the creator may not have imagined. The expected use may differ from the actual uses, but that is a benefit, not a fault, of the tool.

To say that the Internet increases innovation in this way is only part of the story. It also creates a broader array of potential innovators. The ease at which one can connect to and transmit information over the Internet lowers barriers to entry substantially. To distribute software, one need only setup a website or connect to a P2P network. If someone is so disposed, he can release a program onto the Internet and never update it or personally distribute it again.

This is precisely what happened with Gnutella. In March 2000, Justin Frankel and Tom Pepper developed the protocol in fourteen days, released it into the wild, and then stopped distributing it hours after its release at the request of their employers, America Online.[26] Certainly, Frankel and Pepper might not be average amateur programmers, having developed their own software company. At the same time, to release Gnutella, they did not have to create a business, a manufacturing plant, or a distribution scheme. They did not even need to maintain a server or any other infrastructure for very long before the program started spreading. In fact, since the program's introduction, volunteer and commercial development has continued absent the help of its original designers.[27] As such, this example illustrates that anyone at the Internet's "ends" can potentially become the next Frankel or Pepper.

How Sony Supports Internet Innovation

We can now adequately assess how Sony protects innovation in general and aligns with what the Internet demonstrates to us about innovation. Before delving deeply into this analysis, it is worth addressing why courts employ secondary liability, since a copyright holder could sue the underlying infringer instead. Three general reasons suggest the benefit of this additional copyright enforcement tool. First, pursuing third parties, who are often in a superior position to monitor the direct infringement,[28] is more efficient and effective than pursuing myriad individual infringement suits.[29] Second, individuals are often poorly positioned to defend themselves, and levying enormous fines on a limited number of infringers seems unfair.[30] Third, copyright holders will be able to capture the value that they are contributing to the third party's product or service.[31]

Though Congress has never created secondary liability standards for copyright,[32] courts have created two doctrines: contributory and vicarious infringement. Contributory infringement is imposed on "one who, with knowledge of the infringing activity, induces, causes, or materially contributed to the infringing conduct of another."[33] Vicarious liability applies when a party has the right and ability to control the infringer and financially benefits from the activity.[34]

The Supreme Court addressed the doctrines' limits in Sony. Universal City Studios and Walt Disney, owners of copyrights in many television programs, contended that Sony should be liable for consumers' recording copyrighted television programs. The Court noted that typical indirect liability cases involved an "ongoing relationship" between the parties in which "the 'contributory' infringer was in a position to control the use of copyrighted works by others and had authorized the use without permission from the copyright owner."[35] Though Sony admittedly knew that the VCR could be used to make infringing copies, it had no relationship to the infringing acts and no idea which consumers would use VCRs to infringe.[36] Thus, a ruling against Sony would target not just the actions of the third-party, but the existence and functioning of the technology. Sony further argued that "time-shifting"—recording a program to watch it once later—is a non-infringing use.

After a rare re-argument, the Court ruled in favor of Sony 5-4. Though somewhat unclear as to whether its ruling applied to vicarious infringement,[37] the Court held that devices like the VCR "need merely be capable of substantial non-infringing uses" to avoid contributory liability.[38] After affirming this principle, the majority proceeded with a rather fact-intensive analysis of current VCR usage, finding that time-shifting is a non-infringing use and was "substantial."[39] Notably, the majority did not address other arguably infringing uses, like making permanent copies or fast-forwarding through commercials.

Throughout its opinion, the majority expressed its fear that, in spite of secondary liability's benefits, it would unduly hinder this new technology. The majority noted that liability here was "the functional equivalent of holding that the disputed article is within the monopoly granted" to the copyright holder.[40] Even if damages were used instead of an injunction, the effect could be significantly harmful; as Professor Paul Goldstein notes, the precise harm of an infringing copy is difficult to assess, and, inasmuch as damages are passed onto consumers in the form of higher prices, legitimate users would be impeded.[41] In turn, copyright would diminish the public's benefit from legitimate uses by overextending the control of private parties. For that reason, the majority attempted to "strike a balance between … copyright … and the right of others freely to engage in substantially unrelated areas of commerce" (emphasis added).[42]

In creating this balance, Sony builds on principles similar to those that the Internet teaches us about innovation: innovation thrives in an environment of minimal control and its benefits are often unpredictable. Under Sony, unless banning a given general purpose tool will certainly not harm significant legitimate uses, it should be allowed to exist. In this way, Sony protects multipurpose tools like those that flourish on the Internet.

Sony is technology-protective in three key ways. First, because the majority used a clear rule that has no referent to infringing uses, rather than a flexible balancing test, Sony creates the utmost certainty for innovators.[43] Innovators do not have to worry about being dragged into court because of a use they did not consider or the degree of infringement. Once substantial non-infringing uses are identified, the inquiry ends.

Second, because a technology "need only be capable of substantial non-infringing uses" (emphasis added), the rule contemplates that a technology's uses will emerge over time.[44] In so doing, it creates a significantly low threshold for multipurpose tools creators. Many emerging uses might not even implicate copyright, but, even when they do, copyright's "fair use" defense[45] to infringement allows for new non-infringing uses to develop. Before the VCR, no one would have imagined that some day people could easily copy TV programs; because fair use is not discrete, the doctrine could evolve with the technology.[46] Though the majority's fact-intensive fair use analysis has led some to assert that Sony’s result was tied to current VCR uses and could change over time,[47] the word "capable" suggests otherwise. Reading the ambiguous words "capable" and "substantial" together, Sony suggests that a product, used minimally for non-infringing uses today, is still protected because the uses could become substantial tomorrow.

Third, the majority's reasoning suggests that innovators should not be forced to redesign their multipurpose technologies, ignoring arguments made earlier in the case that Sony should do so.[48] As Professor Jane Ginsburg notes, the majority faced an "all or nothing" choice: it could prohibit the device and all its non-infringing uses, or allow it and all its infringing uses.[49] Forcing Sony to change the VCR's design was problematic because the same mechanisms that allowed people to make infringing uses also enabled non-infringing uses.[50] Given that non-infringing uses could not be fully predicted ahead of time, the effects of a redesign on legitimate uses would be unpredictable as well.

One could argue that Sony did not actually preclude such a remedy for all types of technologies. After all, the court explicitly distinguished this case from one's where the technology provider had an "ongoing relationship" and control over the infringing actor. However, given Sony's logic, redesigning the technology itself would still not need to be an option. If Sony had watched TV with its customers, then it could have been expected to prevent infringement. That result has little to do with whether the VCR should be enjoined, but whether Sony's conduct should be.[51] There, a remedy would only target the innovator's actions, not the technology itself.

Internet Innovation and Speech

The argument thus far has set the stakes for an alternative to Sony in terms of innovation. Here, moving back to considering the Internet in particular, I will add one important element to the equation. While the Sony secondary liability rule is agnostic towards the legitimate use's character, its protection of the VCR and thus "time-shifting" was particularly significant because it enhanced consumers' access to televised information. In the Internet and P2P context, inhibiting technology will affect both reception and dissemination of speech on a far larger scale. Because the Internet’s benefits are closely connected to these important social values, they are worth paying special attention to.

The Internet is capable of enabling and disintermediating communication. Just as no one controls the applications that run on the Internet, anyone can send communication directly to another "end" on the network. Unlike "one-to-many" mediums, like television and radio, the Internet can be "many-to-many," allowing for mass two-way communication.[52] Communication on the Internet is also of limited expense compared to traditional mediums. In these ways, the Internet obviates the need for and power of intermediaries in information distribution.

This transformation furthers democratic values by creating a richer marketplace of ideas. Scholar Niva Elkin-Koren suggests that diminishing power of intermediaries will increase diversity of views, in part because corporate sponsorship and thus marketability will be less relevant.[53] Not only will more people be able to express themselves to others, but they will also have greater access to knowledge and information. As the ACLU amicus brief in Grokster points out, many libraries and archival projects distribute public domain works and other materials freely on the Internet and P2P.[54]

Indeed, the benefits of P2P file-distribution are most pronounced in this context. P2P further decreases costs and reliance on intermediaries compared to the client-server model. Before P2P systems flourished, distribution typically occurred by setting up a website through a third-party—a website hosting service—or by setting up a server and maintaining a stable Internet connection. Practically speaking, one also had to be indexed by a search engine, which generally favor large, static, American, commercial sites.[55] With P2P distribution, your computer can become a file distributor with some readily available software, and your files become automatically accessible using the search tool.[56]

Decentralized P2P in particular diminishes intermediary control and enables speech. To understand the full extent of this, it is helpful to imagine a situation in which someone wants to suppress the spread of an important document.[57] With client-server or centralized P2P designs, a clever attacker or an accidental server error can shut down the network, silencing the speech. Even a high level of traffic at the center could slow down the entire network.[58] In decentralized models, however, there is no such chokepoint. Furthermore, by expressing to a centralized source that you have or that you want the document, your actions can be more easily recorded and reported to others. Though privacy can still be compromised in decentralized models,[59] there is no centralized entity that one must trust. Hoping to provide a space for open dialogue, Freenet, a distinct decentralized system, was designed with these dangers in mind. The system is resistant to attack and technically hinders attempts to track users’ activities.[60]

The Internet and P2P under Sony and the Negligence Rule

With the stakes now laid out, I will turn to the current state of Sony in P2P cases, focusing primarily on Aimster's negligence rule approach. I will argue that a negligence rule will substantially undermine Internet innovation and legitimate speech by generating uncertainty and constraining technological designs.

a) Background: Napster and Grokster

In the fall of 1999, Shawn Fanning created the infamous Napster. As noted above, Napster was a centralized P2P system that required users to connect to its servers and search its central index.[61] Napster appealed after losing at the district level,[62] and the 9^th Circuit Court of Appeals ruled that, once a copyright holder told Napster of infringement on its system, Napster was obliged to block it. Eventually, the courts forced Napster to shut down for failing to prevent all infringement.

Scholars have extensively debated the precise extent of Napster's holding, with some even arguing that Sony's VCR would not succeed under it.[63] The court's opinion is convoluted, yet, the key point is threaded throughout it: though Napster's current and potential uses were clearly substantial,[64] its continuing control over infringing uses made it unlike Sony.[65] In this way, the 9^th Circuit followed the essence of Sony. Because Napster could ban users and remove specific file titles from its index, the specific infringing uses could be potentially severed from the current or future legitimate ones, and thus the danger to creating liability was less significant.[66] The court explicitly confirmed that the Napster technology was shielded as distinct from "Napster's conduct in relation to the operational capacity of the system."[67] Furthermore, though the court said Sony did not apply to vicarious infringement, it came to much the same result as in the contributory section, repeating that Napster had control but any remedy must fit its current technological architecture.[68]

Why, then, was Napster shut down? Even though the 9^th Circuit recognized that blocking infringing files was not an "exact science" because Napster did not inspect the actual contents of the song files,[69] the district court's remedy amounted to forcing Napster to alter its design until it could do what it feasibly could not—prevent all infringement.[70]

As will become clear in my discussion of the negligence rule, this forced redesign and shut down are a troubling precedent. The court effectively killed off development of centralized P2P file-sharing. For the purposes of this essay, however, I will not discuss Napster's numerous problems specifically; putting aside the remedy and the court’s not applying Sony directly to vicarious liability, the case’s overall logic follows Sony relatively well.

This faithfulness to Sony played out in the only case thus far involving decentralized P2P, Grokster. The case involved two distinct decentralized systems that allow for transferring of any file type and, to an extent, were created in reaction to Napster's focus on control. The first, StreamCast Networks' Morpheus, accesses the Gnutella network, and, as such, StreamCast has no central control.[71] On the other hand, Grokster accesses the FastTrack network, which designates particular users within the network to acts as indexes, rather than passing requests from peer to peer as Gnutella does.[72] Still, this network allegedly has no central point of control, and, even if it did, Grokster is a third-party licensee that has no access to the network's underlying code.[73]

On April 25, 2003, the judge followed Sony and Napster and ruled in favor of the defendants. Noting that the systems had substantial non-infringing uses, the judge cited distribution of public domain, government, and copyrighted materials authorized for distribution.[74] Both services, unlike Napster, had no ability to control its users' searching for and transmitting infringing files after distribution of the software; in particular, he noted that even if the companies went out of business, neither service would be affected.[75] Those factors provided the rationale for dismissing contributory infringement claims, and, similarly, the judge set aside vicarious liability because the companies lacked control. Making clear the connection to Sony (at least as far as the contributory section went), the judge affirmed:

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.’[76]

b) Aimster and the negligence rule

Not long after Grokster, the 7^th Circuit Court of Appeals fashioned a secondary liability rule in Aimster that threatened the immunity of decentralized P2P providers. Aimster integrated search and file-sharing features into America Online's Instant Messenger chat tool. While the district court did not determine whether Aimster centrally indexed files,[77] Aimster at least had some central control over its network such that pulling the plug on its servers would have some effect.[78] In his ruling against Aimster on June 30, 2003, the esteemed Judge Richard Posner asserted that liability rests on weighing "the respective magnitudes of the [infringing and non-infringing] uses",[79] including how probable they are as well as whether the "prevention of infringing uses would be highly burdensome."[80] Because Aimster had provided no evidence of any non-infringing activity, it easily failed this test.[81] Furthermore, in affirming that this "cost-benefit trade-off" does not treat control and actual knowledge of the infringing act as exclusively determinant, Posner explicitly rejected Napster's interpretation of Sony. In a seeming divergence from Sony's holding, Posner asserted in dicta: "Even when there are noninfringing uses of an Internet file-sharing service, moreover, if the infringing uses are substantial then … the provider … must show that it would have been disproportionately costly for him to eliminate or at least reduce substantially the infringing uses."[82]

This test bears striking similarity to suggestions made by the Grokster plaintiffs[83] as well as William Landes and Doug Lichtman's suggestion that indirect liability be premised on a negligence rule.[84] Like Posner, Landes and Lichtman would weigh all costs and benefits of applying liability, attempting to maximize social welfare.[85] In particular, they would ask whether a "party’s failure to take economically reasonable precautions results in a harm," thus encouraging people to take such precautions ahead of time.[86] Though imposing liability will impair legitimate users, if that harm can be mitigated and the benefit to limiting infringement is great, then on balance liability would be justified. In some special cases, even if redesigning a product to eliminate infringing uses would be too difficult, damages could still apply if a social welfare gain could be achieved.[87]

Such a rule encompasses the necessary predicates for suing decentralized P2P systems. First, a negligence rule would tend to disfavor P2P, because it is disproportionately used for infringement. Because so many people are effortlessly using P2P to commit infringement, there is a significant potential for harm. In particular, P2P critics point out the significant decline in music sales since P2P's genesis.[88] Second, control would not be a wholly determinative factor. The Grokster plaintiffs argue that decentralized systems could and should have included some mechanism to filter out infringing uses.[89]

Before addressing some broader problems with this approach, it is worth focusing on this filtering point, for it is where the negligence rule is seemingly at its strongest. Let us assume for the moment that all decentralized systems had to incorporate filtering systems but were not enjoined altogether. Let us further assume that the filtering would not be overbroad, only blocking infringing files, and that this rule only affects decentralized P2P file-sharing services. If we eliminate the mass infringement and do not inhibit legitimate uses, seemingly avoiding the "all or none" problem faced in Sony, would there be any deficiency with this approach?

Indeed, there would, because a company-controlled filtering implementation would have to sacrifice some aspects of a decentralized design.[90] Any filtering list will become obsolete almost immediately, for new copyrighted works are created all the time. Rather than deploying the software with a filters list, the software would have to acquire a new filters list each time it is used. Even if searches and indexing were not performed centrally, P2P software would have to depend on a centrally-controlled filters server.

In its practical effect, this conflicts with the benefits of the Internet's end-to-end design. That is, the Internet would still technically retain its end-to-end architecture, but the law would limit it by banning the decentralized P2P design. Even though the same legitimate file-transfers would still be possible through other means, legitimate uses would lose the decentralized system's current and potential technical benefits. The precise harm cannot be fully known because development of decentralized P2P will be stopped in its tracks.

At the very least, in terms of using the system to distribute speech, the negligence rule would sacrifice the specific benefits mentioned above and substitute the Internet's disintermediation of speech for mandated intermediary control. The ACLU amicus brief in Grokster notes that mandated control could have its most dramatic effect on politically controversial speech, for filtering and a central point of control can easily become a tool of censorship.[91] Projects like Freenet, dedicated to protecting freedom of speech by eliminating centralized control, would be entirely impossible.

Moreover, this rule would have a substantially negative effect on who can innovate. As noted, one advantage of decentralized P2P is that providers need not maintain any costly infrastructure. Far fewer budding programmers would attempt to write the next Gnutella if doing so requires hosting a centralized infrastructure.

Even in this hypothetically ideal situation, the negligence rule would have demonstrably harmful effects. If we move towards a more realistic account, these problems—undermining the benefits of end-to-end in terms of hampering innovation and speech—are compounded by many orders of magnitude.

First, it is important to note that whatever effects the negligence rule has on innovation will be widespread. Legally, there is no obvious reason why the rule would apply to just decentralized P2P; certainly, Landes, Lichtman, and Posner are all proposing generally applicable rules. As a technical matter, there would be no principled reason why the negligence rule could not apply to essentially any technology that helps one receive and transmit data over the Internet. Decentralized P2P is not unlike many other stand-alone file-distribution tools. As the defendants in Grokster suggest, email clients like Microsoft Outlook could be forced to screen out infringing files.[92] Since Web browsers, like Gnutella, interpret content sent using HTTP, they, too, could be forced to screen downloaded content. Indeed, if it applies to decentralized systems, where the link between the infringer and innovator is weak, then the negligence rule should apply to non-stand-alone software as well. Centralized search engines like Google, which have control over their indices, could be forced to perform filtering beyond their architecture. America Online uses a central logon server for its Instant Messenger program and allows for user-to-user file-transfers (Aimster simply made this easier and added a search tool). This is not to say that liability would be imposed, but rather than any of these services could have a conceivable case brought against them.

Second, the negligence rule is not suited to technological advances because a technology’s costs and benefits cannot be easily assessed. The rule's underpinning is that a judge can narrowly tailor a solution that maximizes benefit while minimizing the harm of infringement. However, in indirect infringement cases, a court will not be able to determine the value of future legitimate uses.[93] Any given case will only look at a technology's costs and benefits as understood at that particular moment. In turn, a judge will not be able to account for the legitimate uses that have yet to emerge fully. Also, as the decentralized P2P example above suggests, the effects of forcing someone to redesign a technology are often subtle and cannot be fully understood in the present. Even if it were simply a matter of determining the added expense of rebuilding the tool, it will be exceedingly difficult for a judge to comprehend fully the costs and feasibility of redesigning a tool. [94]

Moreover, because the negligence rule replaces Sony's clear rule with a balancing test, innovators will suffer due to heightened uncertainty.[95] No innovator will be able to estimate ahead of time whether his particular technology is protected. Determining what quantum of non-infringing uses is needed, or what architectural constraints make a redesign "disproportionately costly," is left up to a judge's discretion. Technology developers will have to be constantly wary of being sued.

This approach's deficiencies will be most stark for technologies in their infancy,[96] thus posing a serious threat to the Internet environment. While Sony's greatest fear was that new technologies would be hampered by copyright, the negligence rule makes a finding of liability most likely in these cases. If an innovator releases a new multipurpose tool, he can have no way of definitively ensuring that it will immediately be adopted for its legitimate uses. The technology's initial primary use could be infringement, and in that case the cost of finding liability will seem much lower to a judge. With Internet technologies, judges will have a particularly difficult time because the digital environment is still incredibly young and evolving; many digital tools do not have clear analogues in the analog world, and, as noted, the Internet innovation thrives precisely because it can happen in unexpected ways.

Given these factors, a negligence rule will lead to conservatism in general purpose tool development that will unduly restrict legitimate uses. Uncertainty in how a court will weigh harms and benefits will lead to the creation of overly restrictive technologies because, as scholar Assaf Hamdani points out, in many cases third parties do not receive the social value of legitimate uses.[97] To the extent that including a feature for legitimate uses will risk enabling some infringing use, innovators will stay on the side of caution. Rather than general purpose technologies, innovators will allow only those uses that they know will pass the court's test.

Not only does this constrain what uses the innovator can create, but also the ones that consumers can. Whether because of a court-forced redesign or an innovator’s undue precautions, people will only be able to put the technology to legitimate uses that the innovator or the court had considered worth protecting. In this way, fair use will not be able to develop in the way it did in Sony. That is, as lawyer Fred von Lohmann explains, if the VCR had initially contained copy-protection that did not protect time-shifting, the Supreme Court never would have had the chance to consider whether it is a fair use.[98] With regard to P2P, the appeals court in Napster rejected numerous fair use arguments for P2P,[99] but that is not to say that other uses cannot emerge.

Such technological conservatism would enhance the harmful effects discussed in the decentralized P2P example. First, as Professors Alfred Yen[100] and Niva Elkin-Koren[101] stress, many of the legitimate uses on the Internet that harsher liability rules will harm are someone's speech.[102] A negligence rule will curb development of communication tools and, inasmuch as it could mandate increased control over users, it could also "[reinforce] the existing centralized structure" in information distribution.[103] Second, uncertainty will be most harmful to the hobbyist, non-commercial, and small-commercial innovators that typically thrive on the Internet. They have the least resources to fight off a lawsuit, particularly when the consequence might force them to employ expensive, permanent use-filtering infrastructure or face paying substantial damages.

In sum, this essay started with an Internet as a space where innovation and speech thrive because of an architecture of limited control. Under a negligence rule, the practical benefits of the end-to-end architecture would be restricted because the general-purpose tools that run on the network would be severely inhibited. In particular, many stand-alone tools like decentralized P2P might be forced to include points of centralized control. Rather than people on the ends figuring out how to put the Internet and Internet applications to as yet unimaginable uses, innovators will have to create restrictive tools. In essence, we would largely forfeit the benefits of the Internet as an enabler of innovation and speech.

Along with these flaws in the negligence rule, its possible fruitlessness in ending infringement should also be considered. Like the court stated in Grokster, the current decentralized P2P systems cannot be shut down. This point is not dispositive because, in the long run, marginalizing these systems by limiting their growth could have some effect. It does, however, weaken the argument of those who support a harsher rule because it will end infringement.

Regardless, Landes and Lichtman argue that one potential way to reduce the negligence rule's uncertainty problems would be to legislate safe harbor provisions.[104] So long as technology creators restrict infringement to the safe harbor’s preset degree, they will be free from suit. Unfortunately, this hardly seems like a solution to the negligence rule's tendency to force technology creators to be conservative. After all, safe harbors encourage conservatism by making sanctioned designs more desirable. Indeed, because safe harbors can only apply to known types of technologies, it is hard to imagine how they could create certainty for creators of new technologies. Finally, if such safe harbors do not preserve stand-alone tools, then they will still not fix the problematic application to decentralized P2P.

Another similar solution would be to restrict somehow the reach of the rule. As I noted above, if the negligence rule can apply to decentralized P2P, then it could reasonably apply to many other tools. Perhaps counterintuitively, distinctions between various search and file-transferring tools could actually exacerbate the negligence rule's uncertainty. Distinctions would have to be rather fine-grained, premised on specific differences in current technologies. Substantially case-specific distinctions will be of little use for emergent tools that transfer files in new ways.

Conclusion: Alternatives

In Sony, the Court provided one additional rationale for its restraint: a long history of the courts deferring to Congress to solve tensions between copyright holder and technologies.[105] Implicitly, the Court also recognized the history of copyright holders' fighting new technologies only to benefit from them later. For instance, copyright holders sued creators of player piano rolls and cable television operators that retransmitted broadcast signals.[106] In both cases, courts recognized that the law did not prohibit the uses, deferred to Congress, and, after carefully examining the problem, Congress created a compromise: a compulsory license that guaranteed remuneration for copyright holders while preserving the public benefit from the technologies.[107] Though Motion Picture Association of America President Jack Valenti once asserted that "the VCR is to the American film producer and the American public as the Boston Strangler is to the woman home alone,"[108] the VCR and videotape rentals have provided a key part of the industry's business.[109]

Because the potential benefits of the Internet are so extensive, Sony’s cautiousness seems even more appropriate today. To that end, I have argued that the Sony approach to liability should be preserved and a negligence rule should not be adopted. I have more narrowly focused on the negligence rule as one rule that could inhibit decentralized P2P. Once we accept these premises and that decentralized P2P will remain in existence, we can look to alternatives that can protect the Internet and copyright. In concluding the essay, it seems wise to consider two plausible alternatives briefly.

One option is to take no action, and let copyright holders pursue the obvious alternative: suing direct infringers. In fact, the Recording Industry Association of America (RIAA) has begun doing just that.[110] Mark Lemley and R. Anthony Reese note that direct infringement suits and criminal prosecution might work well in the P2P context because the infringing content comes from a highly concentrated group of users.[111] Since these users receive little benefit for sharing, deterrence could be effective.[112] Alongside this approach, the content industries can create services that diminish incentive to use P2P. In the last year, the legal digital music market has grown significantly[113] with the introduction of, among others, Apple's iTunes[114] and Roxio's Napster 2.0[115] (which, despite its name is a client-server based service). Whether these services will ever effectively be able to draw people away from free downloads on P2P remains to be seen.

Even so, enforcement currently has several problems that should make us skeptical of this approach. First, locating targets has so far required privacy violations due to erroneous RIAA subpoenas without any judicial oversight.[116] Second, the penalties for infringement are currently grossly out of proportion, allowing potentially for $150,000 damages per infringement.[117] Third, and most importantly, newer P2P systems are likely to develop anonymizing features that could render enforcement fruitless in the long run. StreamCast, for instance, helps its users find “proxy servers” that mask the address that the RIAA would subpoena,[118] and, though widespread infringement using Freenet is currently impractical,[119] that could change. New tools are already emerging to route around the RIAA's enforcement mechanisms.[120]

Another option would be to turn to Congress for a solution. Courts have limited remedies for copyright infringement: injunctions and damages;[121] in addition, courts can only consider the facts and parties of one case and thus lack the ability to balance the many interests and details of complex copyright issues. Congress, on the other hand, can take the broader view and has all possible policy tools at its disposal. Though, as Jessica Litman has comprehensively detailed,[122] Congress has often failed to create flexible rules that fully consider the public's interest, it has been able to form compromises that preserve technology and copyright.

Of course, Congress' limiting certain technological designs would hardly be better than the courts doing it. Over the past several years, there have been several proposals that would force digital media devices to incorporate copy-protection measures.[123] Such mandates are severely detrimental to innovation, sharing many of the negligence rule’s problems.

Several scholars have recently proposed more useful alternatives[124] often called "alternative compensation systems"[125] (and sometimes referred to as "compulsory licenses," though dissimilar from those mentioned above). Generally, these policies would tax technologies related to P2P file-sharing and distribute the money to copyright holders, while legalizing infringement on the Internet to varying degrees. Such systems have various problems, including the complexities of government administration, cross-subsidization,[126] and ways of cheating the system such that certain copyright holders receive more than they deserve.[127] On the other hand, rather than fighting the ability to disseminate and acquire information, these systems encourage the practice. In that way, they build on the advantages of the Internet.

And that, in the end, should be the goal. Rather than choosing between P2P and pay-to-play, between the Internet and copyright, we should seek a world that protects both. Attempts to assess the cost-benefit trade-offs of a particular technology using the negligence rule will miss out on this broader picture, sacrificing the Internet’s many benefits in the process. Instead, we should protect Sony and turn our attention towards creating a more constructive, comprehensive solution.