Copyright and the Rule of Reason

In this essay I will describe several ways to reduce the growing disconnect between the utilitarian rationale that undergirds copyright law and the copyright law itself. Before I come to the arguments, however, a few words regarding the premise: What do I mean by the "disconnect" between copyright law and its justification?

Copyright is sometimes justified as the appropriate reward for an author's creative labor. 1 And copyright is also sometimes justified as a way that we acknowledge an author's strong interest in a creation that reflects and embodies his or her personality. 2 But the dominant justification for copyright, at least in the United States, is explicitly utilitarian. Congress's power to create patent and copyright laws is provided for explicitly in our Constitution, and-uniquely among the Associate Professor, University of Virginia School of Law. Thanks to participants at a conference on antitrust and intellectual property organized by Tom Cotter at the University of Minnesota Law School. Thanks also to Albert Choi, Scott Hemphill, Bill McGeveran and Dan Ortiz for helpful comments on an earlier draft, and to Brady Cox for expert research assistance. All errors are, of course, mine.

1. For a description and critique of the labor justification for copyright, see Seana Valentine Shiffrin, Lockean Arguments for Private Intellectual Property, in NEW ESSAYS IN THE LEGAL AND POLITICAL THEORY OF PROPERTY 138 (Steven R. Munzer, ed., 2001).

2. E.g., Justin Hughes, The Philosophy of Intellectual Property, 77 GEO. L.J. 287, 330-65 (1988) (describing, critiquing, and providing arguments to bolster Hegelian personality justification for copyright).

higher than competitive price. We hope that this lure will stimulate investment in creative works.

Of course, stimulating the production of more creative works is not the only thing required to "promote progress." We must also worry about access to the new works we have incentivized authors to produce-artistic and literary works are valuable because they create our culture, but they do so only if they are widely accessible. Technological developments, including, most importantly, the rise of digital platforms and the Internet, have lowered dramatically the cost of distributing many creative works, thereby lowering the cost of access. But too-strong copyright law is a threat to the increased access that technology would otherwise permit: by raising the price of commercially-valuable new works, copyright threatens to restrict access. This is true in the simple sense of pricing people out of access to works that they would be able to afford in a competitive market. But there is another, very important element to access-i.e., the ability of creators to use pre-existing works as building blocks for new works. Copyright law interferes with this process, and raises the cost of creating new works.

For these reasons, copyright law must seek a balance between private incentives to create new works, and public access to the works created. To pursue balance, copyright law cannot simply provide rightsholders with complete control over their works, and it never has. So, for example, although copyright owners have the right to control the initial distribution of their work, under copyright law's first sale doctrine, owners of copyrighted works are and long have been unable to restrict the price or terms at which their works are re-sold. 8 Similarly, copyright owners' rights have long been limited by a general (albeit narrowlyfocused) exemption from liability for uses deemed "fair." 9 And of course copyright rights have always been subject to time limits-the Constitution's "limited Times" proviso proscribes perpetual copyright, 10 and a term that ends, thereby eventually sending works into the public domain where they will forever remain as "free as the air to common use", 11 is an important way in which American law has balanced private copyright's reformalization -i.e., for the reintroduction into the copyright law of a set of copyright formalities such as registration, notice, and renewal -as a means for focusing copyright on the commercially valuable works for which property rights are salient. See Christopher Sprigman, Reform(aliz)ing Copyright, 57 STAN. L. REV. 485 (2004) incentives with public access. The particular limitations point toward a more general axiom: a copyright law built on utilitarian foundations should be limited to those rights necessary to create and maintain an ample supply of new works. All other uses should be left unregulated.

To achieve this balance, copyright law must be based on a theory of harm-i.e., we must understand the kinds of uses that cause significant harm to authors' incentives, and those that do not, so that we might focus on regulating the former and not the latter. And here is where the problem arises: although we understand copyright's concept of harm at an abstract level-i.e., copyright "harm" arises from any use that threatens to suppress author incentives significantly below the optimal level-the theory is exceedingly difficult to apply in many cases. One frequently-occurring example is the creation of a derivative work-i.e., a work based on a pre-existing copyrighted work that adds significant new creativity. Current copyright law gives rightsholders the exclusive right to make or to authorize the production of all works that use a not-insubstantial amount of material from their work-even where the second creator adds substantial new creativity of his own. 12 What is the justification for this very broad limit on others' creativity? At the very least, the rule seems overbroad, for authors' incentives are not automatically implicated whenever someone makes and distributes an unauthorized derivative work. If a derivative is close enough (both in subject matter and time) to the original that it competes with it, or perhaps preempts a product or licensing market that the author might otherwise plan to enter, then the derivative work at issue may harm the original author. In such an instance, if the copyright law did not reach conduct in this category, we might fear overall harm to author incentives to create. However, if (as is often the case) the derivative is not closely similar to the original, does not compete with it for audience patronage, and does not preempt a market that the original rightsholder realistically is positioned to exploit, then the existence of the derivative is unlikely to harm the original author. In a large number of cases, plaintiffs have claimed, as we would expect, that they are harmed because the defendant 12. See 17 U.S.C. § 106(2) (2006). A property right of this sort is far from an inevitable feature of copyright. Indeed, the early U.S. copyright statutes contained no derivative work right, but limited copyright to the "sole right and liberty of printing, reprinting, publishing and vending" -i.e., exclusive rights of reproduction and distribution. Copyright Act of May 31, 1790, ch. 15, 1 Stat. 124 (1790) (repealed 1802). It is also entirely possible, moreover, that a copyright system could give copyright owners a right to get paid for derivatives, but not to control them. The law could do this by, for example, subjecting derivative works to a liability rule -i.e., the second author would be free to make the derivative, but would be required to pay compensation. The current copyright laws already use this intermediate type of right for derivative works based on copyrighted musical compositions-known colloquially as "cover songs"-which may be produced under a compulsory license so long as the basic character of the work is preserved. 17 U.S. C. § 115 (2006).

did not pay for a license to make the derivative. And courts have tended to approve these claims. 13 They do so in part because they lack a theory, or indeed much evidence, to help them distinguish cases in which control is necessary to maintain author incentives from those in which defendants' use does not pose any meaningful chance of harm.

In the absence of such a theory, the courts' tendency has been to give to rightsholders more and more complete control over derivative works. This is true especially because copyright is a strict liability tort. Once a plaintiff proves unauthorized copying of any not-insubstantial protected material, the burden shifts to the defendant to prove that its use is fair. 14 And, importantly, fair use is currently not well designed to produce information about potential harms. Defendants bear the burden of proving fairness, and yet it is plaintiffs who are most often in possession of any evidence relevant to the likelihood that a particular use will harm them. In the absence of evidence, courts are reduced to casual empiricism, or, worse, abstract theorizing about markets and effects. In sum, copyright law's current structure and distribution of proof responsibilities has contributed to copyright's growing disconnect from its utilitarian justification.

It's time to think of a new approach. There has been useful work recently that begins the work of restructuring copyright law along more defensibly utilitarian lines. I refer in particular to important articles by Shyamkrishna Balganesh 15 and Christina Bohannan. 16 Balganesh suggests the addition to the copyright plaintiff's prima facie case of a generally-applicable test of "foreseeable copying"-i.e., that the defendant's copying was of a type reasonably foreseeable as of the time that plaintiff created the work in suit. 17 The aim of this additional element of the plaintiff's case, Balganesh argues, is to limit copyright to the scope of its utilitarian justification: if the copying was not of the type that an author reasonably would have foreseen ex ante, it could not have 13. See, e.g., Bridgeport Music, Inc. v. Dimension Films, 410 F.3d 792, 801 (6th Cir. 2005) (holding, in a case involving audio sampling of two-second guitar chord, that defendant must "[g]et a license or . . . not sample"); Rogers v. Koons, 960 F.2d 301 (2d Cir. 1992 affected (one way or the other) his incentives to create. 18 Like Balganesh, Bohannan argues for a concept of copyright harm that requires proof that a particular use of a copyrighted work is likely to have a material effect on a reasonable copyright owner's ex ante incentive to create a new work. 19 Bohannan would, however, employ a foreseeability standard not as an element of the plaintiff's prima facie case but rather to reform the fair use defense-in fact, Bohannan argues that, properly understood, the Supreme Court's fair use cases already hew to this model. 20 The Balganesh and Bohannan articles make an important contribution by using foreseeability analysis as a means to re-link copyright law with its utilitarian justification. With that said, in my view, adding a foreseeability requirement, either to plaintiff's prima facie case or to the fair use doctrine, will not in itself supply an administrable theory of copyright harm. The reason for this can be glimpsed in the Supreme Court's recent opinion in Eldred v. Ashcroft. 21 In Eldred, the Court upheld against Copyright Clause and First Amendment challenges Congress's 20-year extension of subsisting copyrights. 22 The gravamen of the Copyright Clause challenge was a seemingly commonsensical proposition: the Copyright Clause premises Congress's copyright lawmaking on the promotion of progress, but extension of existing copyrights cannot possibly promote progress because authors create any particular work only once, and so extending terms retrospectively is simply a quid pro nihilo. 23 In her opinion for the Court, Justice Ginsburg made quick work in dismissing this argument. Noting that Congress consistently has extended subsisting terms each time it extended future terms, Justice Ginsburg wrote, "the author of a work created in the last 170 years would reasonably comprehend, as the [protection] offered her, a copyright not only for the time in place when protection is gained, but also for any renewal or extension legislated during that time." 24 This is obviously a point about foreseeability, and it illustrates what I take to be an intractable defect that makes the concept difficult to employ as a stand-alone limitation to copyright's theory of harm. What is "foreseeable" in any particular copyright case is uncertain and readily manipulable. If Justice Ginsburg and a majority of the Court believe that the prospect of distant future copyright term extensions helps to shape 18. Id. at 34. 19. Bohannan, supra note 16, at 970. 20. Id. at 991-1002. 21. 537 U.S. 186 (2003). 22. Id. at 218-19. 23. Id. at 189. 24. Id. at 214-15 (footnote omitted).

the incentives of rational authors, then who is to gainsay that the prospect of licensing revenues from the use of plot incidents from a famous television comedy as fodder for a quiz book, 25 or from turning a children's book involving a trouble-making cat into a satirical comment on the O.J. Simpson murder trial, 26 or from the use of a picture postcard photograph as the basis for a modern art sculpture, 27 will foreseeably shape author incentives as well and thus class among the uses subject to copyright law?

Copyright scholars will continue the quest for a useful theory of copyright harm. But in the meantime I suggest we lower our sights. There are other, more modest, ways to nudge copyright law back toward its utilitarian justification. We can rely on a set of indirect strategies to push the incentives of rightsholders in a direction that will helpfully separate unauthorized uses that reduce author incentives from those that do not. I want to briefly suggest and defend two related strategies.

First, we should distinguish between conduct we know will harm author incentives over the run of cases, and conduct with more ambiguous effects. So creation and distribution of exact copies of a work should be treated differently than creation of a derivative work. The first we know will almost always be harmful; whether the second is depends on the facts of a particular case.

Second, we should re-structure copyright's burdens of proof to better filter harmful from harmless uses. This second strategy grows out of and is aimed at implementing the first. For cases involving infringing conduct that is very likely to cause harm, we should preserve copyright's current strict liability rule. Indeed, perhaps we should strengthen it by limiting the availability of the fair use defense in these cases of "per se" copyright liability. But for cases involving infringing conduct in our second category-i.e., where the effect of the infringing conduct is ambiguous-we should require plaintiffs to prove that they have been harmed in some substantial way. 28 There are two principal benefits of such a change. First, by requiring that plaintiffs show substantial actual or likely harm in these "rule of reason" copyright infringement cases, we will encourage plaintiffs who have suffered substantial harm to come forward, while discouraging 25. Castle Rock Entm't, Inc. v. Carol Publ'g Group, 150 F.3d 132 (2d Cir. 1998 suits by rightsholders who suffer no harm, or only speculative harm. Second-and perhaps most importantly-altering the plaintiff's prima facie case in this way will produce information about harms and benefits of different uses of copyrighted works. To do this effectively, the law needs to place the burden on the party most likely to have information about the harm-in virtually all cases, that is likely to be the plaintiff.

The law as structured now does not reliably produce this information, with the result that copyright litigation does not help us to know more about how creative incentives are or are not harmed. If we hope to improve our understanding over time, we should re-structure the law so that litigation produces the information about harm that we currently lack.

Given the costs of copyright regulation, the wide range of conduct regulated, and the differing effects on authors' incentives arising from different forms of infringing conduct, one would expect to find some feature in the copyright law that separates the unauthorized uses of copyrighted works that we suspect will, in almost all cases, strike at authors' incentives from those that have more ambiguous effects. There is, however, no such mechanism in current copyright law. Copyright operates, for all practical purposes, according to a blanket per se rule.

By "per se", I mean that copyright liability is imposed without requiring any showing of actual harm or that the incentives of the plaintiff in an individual case (or of authors generally) have been or would be impaired in any way by the infringing conduct. Instead, current law imposes liability wherever there is conduct that transgresses one of the copyright owner's exclusive rights. 29 Harm to the plaintiff is presumed based on proof of the infringing conduct.

If we decided to re-structure copyright to account for the differing propensities of various types of infringing conduct to harm creative incentives, our first task would be to identify the forms of infringing conduct that we believe are likely in most cases to harm authors' incentives. Conduct in this category should be treated as "per se" copyright infringement and condemned without any need for an individual plaintiff to show that he, or authors in general, have been harmed or are likely to suffer harm in the future. However, for conduct that, although infringing, has ambiguous effects on authors' incentives, 29. 17 liability should be reserved for instances in which individual plaintiffs can show that they have in fact been harmed.

We see such a bi-partite liability structure in the law of antitrust, where a small (and shrinking) category of conduct is subject to a per se rule of illegality, but where most forms of potentially anticompetitive conduct are made subject to antitrust's rule of reason. Conduct in this latter category is, unlike that subject to antitrust's per se rule, not the type of conduct that experience or economic theory has identified as reliably anticompetitive across the run of cases. Rather, conduct subject to antitrust's rule of reason has ambiguous welfare effects. Depending on the facts of a particular case, rule of reason conduct may be anticompetitive, or it may be competitively neutral or even procompetitive. And since we don't know which it will be in advance, plaintiffs in antitrust rule of reason cases are required, as an element of their prima facie case, to demonstrate that the particular conduct at issue harms competition. 30 Copyright law would benefit from the adoption of something approximating antitrust's strategy of calibrating the liability trigger to different forms of potentially harmful conduct. Toward that end, we might consider the importation of a "rule of reason" governing infringing conduct with ambiguous effects on authors' incentives. A copyright law reformed to differentiate between per se and rule of reason infringement would better balance protection of authors' incentives with the widest possible public access to creative works. The introduction into copyright law of an antitrust-like approach could, therefore, yield a substantial efficiency gain. Such a move would also take considerable pressure off of copyright's fair use doctrine, which was never meant to serve-and has in practice never sufficed-as a general mechanism to weed out claims where authors' incentives are not at stake. Finally, and perhaps most importantly over the long term, the introduction of a per se/rule of reason distinction into copyright law would incentivize the production of more information about the incentive effects of a variety of uses of copyrighted works. Progress in our understanding of copyright rules and the effect of unauthorized uses on authors' incentives requires empirical data illuminating the effects of different uses on the markets for the huge variety of creative works that copyright law governs. And yet at the moment, copyright is structured in a way where such information has little salience, and is therefore not produced.

I should pause to make a quick point about terminology. I am 30. See, e.g., Geneva Pharms. Tech. Corp. v. Barr Labs., Inc., 386 F.3d 485, 506-07 (2d Cir. 2004).

arguing for a copyright rule of reason, but the "rule of reason" is not properly a rule at all. It is, rather, a standard.

Rules are legal commands that differentiate between legal and illegal conduct in a way that readily may be determined ex ante. The way in which we typically regulate speeding is an example-a speed limit is posted (e.g., 65 mph), and you violate the law if you exceed it. Speeding usually is subject to selective enforcement; it may be that in a particular jurisdiction few drivers are pulled over for doing less than 70 when the posted speed is 65. But selective enforcement does not change the fact that the offense of speeding is defined according to a rule. The proscribed conduct is identified and readily understood in advance.

In contrast, standards are legal commands that differentiate between legal and illegal conduct in a way that requires significant decisionmaking ex post. An example would be a speeding rule that directed drivers to "drive reasonably according to the current weather and traffic conditions." Enforcement of a standard against speeding would require police officers, and then courts, to assess reasonableness under the conditions in each case. This may sound bizarre as a regime to govern speeding. It is, however, the regime that governs the accidents caused in part by speeding, for our conduct as drivers in general is assessed according to a standard. All of us are required when driving to conform to an objectively reasonable standard of conduct, assessed according to circumstances in individual cases.

There is an enormous and well-developed literature setting out the trade-offs between rules and standards in structuring legal commands. 31 Rules promise simplicity, at the cost of accuracy. Staying with the speeding example, 65 mph is not an ideal speed limit for all drivers and all vehicles in all conditions traversing any stretch of road so marked. Speed limits are nonetheless structured as a rule because of the cost and difficulty of enforcing a standard, where neither drivers nor police would know with certainty, until a judge or other fact-finder sorts through the (1974) (arguing that the desire to minimize costs is a dominant consideration in the choice between precision and generality in the formulation of legal rules and standards); Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 DUKE L.J. 557 (1992) (concluding that the desirability of rules and standards is most influenced by the frequency of the conduct that will be governed by the law); Carol M. Rose, Crystals and Mud in Property Law, 40 STAN. L. REV. 577, 604-10 (1988) (arguing that since "muddy" standards and bright-line "crystal" rules serve different functions, neither is necessarily preferable to the other); Kathleen M. Sullivan, The Supreme Court 1991 Term: Foreword: The Justices of Rules and Standards, 106 HARV. L. REV. 22 (1992) (outlining several arguments commonly used in support of both rules and standards). particular case, where the borderline lies between lawful and unlawful conduct. So in the case of speed limits we favor a rule because of the expense of enforcing a standard through individualized determinations, combined with the expectation that we have little to gain from more precisely tailoring the speed limit via a standard (especially given the availability of selective enforcement of the rule, which also serves a tailoring function, albeit incompletely).

Standards promise the converse: they provide accuracy, at the cost of complexity. An example would be parties in a closely contested tort suit arguing about the reasonableness under the circumstances of specific conduct. The tort system's negligence standard requires substantial investment to determine whether the conduct in question falls below the threshold. Undertaking that investment is worthwhile, however, wherever the possible varieties of negligent conduct are so great that defining a rule in advance is likely to lead to unacceptable over-and under-inclusiveness.

Current copyright law contains both rules and standards, although they are distributed idiosyncratically. The plaintiff's prima facie case in a copyright infringement lawsuit is structured as a per se rule-i.e., if the plaintiff proves infringing conduct (even if undertaken unconsciously), the defendant is liable. This is so whether or not the plaintiff was harmed by the particular infringing conduct. Once the rule is shown to have been breached, harm is presumed.

Copyright law includes a defense to infringement liability that is constructed as a standard, or, in Congress's conception, as an "equitable rule of reason." 32 This is the fair use defense, which arose as judge-made doctrine but is now set out in the statute itself: § 107. Limitations on exclusive rights: Fair use Notwithstanding the provisions of sections 106 and 106A, the fair use of a copyrighted work, including such use by reproduction in copies or phonorecords or by any other means specified by that section, for purposes such as criticism, comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research, is not an infringement of copyright. In determining whether the use made of a work in any particular case is a fair use the factors to be considered shall include-32. See H.R. REP. NO. 94-1476, at 65 (1976), reprinted in 1976 U.S.C.C.A.N. 5659, 5679 ("since the [fair use] doctrine is an equitable rule of reason, no generally applicable definition is possible, and each case raising the question must be decided on its own facts.").

(1) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes;

(3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and

(4) the effect of the use upon the potential market for or value of the copyrighted work.

The fact that a work is unpublished shall not itself bar a finding of fair use if such finding is made upon consideration of all the above factors. 33 Is reliance on the fair use defense a workable way to focus copyright on harmful conduct? Judging from the life of the defense thus far, the answer appears to be no. Current law constructs fair use as a defense to infringement, with the burden of proof on the defendant. 34 But in most copyright cases it is the plaintiff that has superior access to information about harm-harm to himself, directly, and, by extension, harm to other authors similarly situated. Fair use does not oblige the plaintiff to produce that information. Courts are instead reduced to theorizing about harm, and they do this badly.

There is another reason-probably of greater practical importance-that fair use cannot function as a general mechanism for sorting harmful from harmless uses. This has to do with copyright's remedies provisions. These are constructed in a way that makes it difficult to predict, with any precision, what penalties are likely if an infringement is not excused as a fair use. Under current law a copyright owner who successfully brings a lawsuit for infringement of a registered copyright is entitled to choose between two principal measures of damage. The plaintiff may elect to recover his "actual damages" as well as "any additional profits of the infringer" that are "attributable to the infringement and are not taken into account in computing the actual damages." 35 Under this formulation, the plaintiff is entitled to recover his actual losses (including the value of any license that might have been obtained absent the infringement) plus the disgorgement of any profits realized by the infringer. Alternatively, in instances in which the plaintiff's copyright was timely registered, the plaintiff may choose to recover statutory damages in amounts set out in the copyright law and 33. 17 U.S.C. § 107. 34. See sources cited supra note 14. 35. 17 U.S.C. § 504(a) (2006).

not linked to any particular showing of harm. Current law permits statutory damages ranging between $750 and $30,000 per work infringed for ordinary infringement, 36 and up to $150,000 per work if the infringement is deemed willful. 37 Additionally, and importantly, the court is permitted, but not required, to award prevailing parties' court costs and attorney's fees 38 and (most likely) prejudgment interest. 39 And injunctive relief is available and is freely granted on both a preliminary and permanent basis. 40 The wide variation in possible damages, the easy availability of equitable relief, and the possibility that plaintiffs may obtain wideranging remedies even in the absence of any showing of harm, means that would-be users are unlikely in many instances to rely on fair use as a sorting mechanism. The risks of failure are too great.

At this point we can assess the merits of the current per se rule in copyright. Our current approach has the benefit of (relative) simplicitythe plaintiff's prima facie case is premised solely on proof of conduct defined in advance and the fair use defense applies not in cases involving "ordinary uses of copyrighted works," but only to those involving "unusual or marginal activities." 41 But reliance on the per se rule produces inaccuracy-current copyright law condemns all infringing conduct according to the same rule, whether or not the particular conduct is likely 36. Id. § 504(c)(1). A plaintiff's ability to collect statutory damages is contingent upon registration of the work either prior to its infringement, or within three months of its publication. Id. § 412.

37. Id. § 504(c)(2). Conversely, in cases of innocent infringement -i.e., infringement unaccompanied by any intent to infringe or knowledge regarding the unlawfulness of the infringing conduct -a court may reduce statutory damages to as little as $200 per work infringed. Id. The availability of these special statutory damages is conditioned on the plaintiffs' registration of the infringed work prior to the infringement's commencement. Id. § 412. In addition to money damages, courts are permitted to order the seizure and destruction or other disposition -including transfer to the prevailing plaintiff -of infringing articles. Id. § 503.

38. Id. § 505. Like statutory damages, the availability of attorney's fees is conditioned on timely registration of the work. Id. § 412.

39. In Design v. K-Mart Apparel Corp., 13 F.3d 559, 569 (2d Cir. 1994) (noting that availability of pre-judgment interest an unresolved issue in Second Circuit); Kleier Adver., Inc. v. Premier Pontiac, Inc., 921 F.2d 1036, 1040-1042 (10th Cir. 1990) (awarding pre-judgment interest). In addition to these civil remedies, the copyright laws also impose criminal penalties for certain instances of copyright infringement. The federal Copyright Act imposes prison terms of up to ten years and substantial criminal fines for infringement of registered works if the infringement has been undertaken either "for purposes of commercial advantage or private financial gain," id. § 506(a)(1), or for reproduction or distribution, within a 180-day period, of copies with a total retail value exceeding $1000, id. § 506(a)(2). Under these definitions, the prospect of criminal penalties hangs over many, if not most, instances of infringement. Thus far, however, the federal government has been sparing in its application of the criminal penalties.

The costs and benefits of a copyright rule of reason are precisely counter to those inhering in the current rule. Under a copyright rule of reason, liability would be premised not simply on proof of an infringing act, but on proof of the actual or threatened imposition of the type of harm that is anticipated. Employing this liability standard would present greater complexity-the plaintiff's prima facie case in every rule of reason infringement action would include not just proof of infringing conduct, but an assessment of whether harm is likely. The benefit of such an expanded inquiry would be a better fit between means and ends-i.e., between copyright's application and the preservation of authors' incentives.

Antitrust, like copyright, is an economic regulatory system focused on avoiding a form of market failure-in the case of antitrust, the possibility that firms or groups of firms with market power will suppress competition, raise prices, and deter innovation and investment. 42 And like copyright, antitrust must perform a balancing act. Most conduct that is potentially anticompetitive is also potentially procompetitive. Which outcome is likely in any particular case depends on the characteristics of the firms, products, and markets at issue. There is the persistent worry, moreover, that the welfare effects of most forms of potentially anticompetitive conduct often is ambiguous even upon close examination, and courts therefore may err in their assessment of these effects. As a result, antitrust liability rules that are too aggressive may deter or preempt procompetitive, as well as anticompetitive conduct.

Unlike copyright, however, antitrust law features a mix of rules and standards tailored to different forms of potentially anticompetitive conduct. For conduct known to harm competition in most cases-e.g., conspiracy to fix prices, rig bids, or divide markets 43 -antitrust imposes liability according to a "per se" rule. For conduct in this category, a plaintiff demonstrates a violation of the antitrust law by showing that the defendant intentionally engaged in the proscribed conduct. The plaintiff is not required to prove that the defendant intended-or even was aware of the possibility of-harm to competition. Indeed, the existence of such 42. See Harry First, Controlling the Intellectual Property Grab: Protect Innovation, Not Innovators, 38 RUTGERS L.J. 365, 390-91 (2007).

43. N. Pac. Ry. Co. v. United States, 356 U.S. 1, 5 (1958) (identifying price fixing, division of markets, group boycotts, and tying arrangements as unlawful activities "in and of themselves"); see also Verizon Commc'ns Inc. v. Law Offices of Curtis V. Trinko, LLP, 540 U.S. 398, 408 (2004) (identifying collusion for purposes such as price fixing as the "supreme evil of antitrust").

harm is irrebuttably presumed. 44 A very different liability standard applies, however, to conduct with ambiguous welfare effects-i.e., to conduct that may harm competition in some instances, but that may be neutral or even procompetitive in others. In such cases, antitrust law assesses liability under a "rule of reason". Courts employing the rule of reason do not presume harm to competition; rather, the plaintiff must demonstrate it, and show that the competitive harm outweighs any associated and offsetting precompetitive effects ascribable to the conduct at issue. 45 In order to show harm, plaintiffs in rule of reason cases are typically required to describe the relevant market-i.e., the economic market in which the products or services at issue compete. The likelihood of competitive harm is assessed by analyzing the effect of the conduct at issue on competition within that market or markets. In addition, it often is said in the rule of reason context that plaintiff must show that defendant acted with the intent to harm competition. This does not mean that plaintiff must show that defendant subjectively intended to cause competitive harm, but rather than such an intent can be inferred objectively from the character of the conduct. 46 Importantly, plaintiffs in a rule of reason antitrust case must demonstrate that the specific conduct at issue threatens, on balance, to 44. N. Pac. Ry. Co., 356 U.S. at 5 ( [T]here are certain agreements or practices which because of their pernicious effect on competition and lack of any redeeming virtue are conclusively presumed to be unreasonable and therefore illegal without elaborate inquiry as to the precise harm they have caused or the business excuse for their use. This principle of per se unreasonableness not only makes the type of restraints which are proscribed by the Sherman Act more certain to the benefit of everyone concerned, but it also avoids the necessity for an incredibly complicated and prolonged economic investigation into the entire history of the industry involved, as well as related industries, in an effort to determine at large whether a particular restraint has been unreasonable -an inquiry so often wholly fruitless when undertaken. ) (emphasis added).

45. Antitrust courts have been quite adept at using burden-shifting methodologies when plaintiffs have made initial showings that a particular form of conduct is quite likely, in the circumstances to harm competition. See, e.g., United States v. Microsoft Corp., 253 F.3d 34, 59 (D.C. Cir. 2001) (noting that after plaintiffs make a prima facie case demonstrating anticompetitive effects, the burden shifts to the alleged monopolists, who may offer procompetitive justifications for their conduct in order to shift the burden back to the plaintiffs to rebut the claim).

46. See Ronald A. Cass & Keith N. Hylton, Antitrust Intent, 74 S. CAL. L. REV. 657, 659 (2001). In contrast to antitrust, copyright infringement is a strict liability offense -there is no knowledge or intent requirement, and even accidental or unconscious infringement is actionable. See, e.g., De Acosta v. Brown, 146 F.2d 408 (2d Cir. 1944); Bright Tunes Music Corp. v. Harrisongs Music, Ltd., 420 F.Supp. 177 (S.D.N.Y. 1976), aff'd, ABKCO Music, Inc. v. Harrisongs Music, Ltd., 722 F.2d 988 (2d Cir. 1983). In this aspect, copyright's per se rule is even more categorical than antitrust's -no one has ever been held liable for "accidental" or "unconscious" price-fixing, market allocation, or bid-rigging.

harm not just the plaintiff but competition in some relevant economic market. This dominant form of antitrust analysis requires case-specific inquiries into market harm. 47 By varying liability rules in this way, antitrust doctrine tailors the law's application to account for the varying likelihood, given different forms of potentially anticompetitive conduct, of actual threat to competition. Perhaps just as importantly, antitrust law's mix of rules and standards incentivizes the production of more and better information about the market impact of various forms of potentially anticompetitive conduct.

Again, the difference between antitrust's per se and rule of reason methodologies is the difference between rules and standards. Antitrust mixes a select number of per se rules (entering into an agreement to fix prices is unlawful) with a relatively large area where conduct is judged according to a standard (an exclusive dealing contract will be judged unlawful if harm to competition from the arrangement's exclusion of rivals outweighs any efficiencies gained via the arrangement). For the conduct subject to per se rules, antitrust sacrifices some accuracy-even price-fixing does not harm competition in all cases. (Think, for example, of a situation involving an incumbent and a potential entrant, and where entry (absent agreement) is likely to trigger marginal-cost pricing. Entry under these conditions might be unattractive to the potential entrant, especially if the incumbent is a lower-cost producer and would thus be able to underprice the entrant in a competitive market. But if the entrant is able to credibly pre-commit to entry and reaches an imperfect price agreement with the incumbent, the result will be duopoly pricingpricing above marginal cost, and therefore not as good as full competition, but better for consumers than monopoly pricing. Pricefixing under these conditions makes entry and hence lower-than-47. Cont'l T.V., Inc. v. GTE Sylvania Inc., 433 U.S. 36, 49-50 (1977) ("Under [the rule of reason], the fact-finding weighs all of the circumstances of a case in deciding whether a restrictive practice should be prohibited as imposing an unreasonable restraint on competition. Per se rules of illegality are appropriate only when they relate to conduct that is manifestly anticompetitive.") (citations omitted); Chicago Bd. of Trade v. United States, 246 U.S. 231, 238 (1918) (

The true test of legality is whether the restraint imposed is such as merely regulates and perhaps thereby promotes competition or whether it is such as may suppress or even destroy competition. To determine that question the court must ordinarily consider the facts peculiar to the business to which the restraint is applied; its condition before and after the restraint was imposed; the nature of the restraint and its effect, actual or probable. The history of the restraint, the evil believed to exist, the reason for adopting the particular remedy, the purpose or end sought to be attained, are all relevant facts. This is not because a good intention will save an otherwise objectionable regulation or the reverse; but because knowledge of intent may help the court to interpret facts and to predict consequences. ). monopoly pricing possible). But the law also gains clarity and predictability-if you agree to fix prices, you have violated Section 1 of the Sherman Act, full stop. And the prophylactic rule that the law sets up against per se conduct also conserves prosecutorial and judicial resources: antitrust criminal prosecutors and civil plaintiffs need not offer proof of harm in individual price-fixing cases to establish liability, and courts are spared the task of assessing that evidence.

The situation is, of course, reversed for conduct subject to antitrust's rule of reason. The rule of reason standard sacrifices clarity and predictability by subjecting conduct with uncertain welfare effects to an inquiry that varies significantly in individual cases. (The rule of reason inquiry necessarily varies because both the details of competition in different relevant economic markets and the varieties of competitive harm and procompetitive efficiencies that may arise in these settings are as diverse as the forms of potentially anticompetitive conduct). This loss is-in theory-balanced by increased accuracy. If courts are able reliably to distinguish between procompetitive and anticompetitive conduct, then only the latter is subject to sanction. Plaintiffs are required to shoulder the burden of establishing harm to competition, which suggests a bias toward tolerance of type II errors (i.e., false negatives) over type I errors (i.e., false positives).

One final point: based on the differences in proof requirements set out above, it appears to me that antitrust and copyright are premised on divergent understandings of the workings and potential failures of markets. Copyright is justified, as has been stated, by a fear of market failure created by uncontrolled copying, and resulting in sub-optimal incentives to create new artistic and literary works. Judging from the expansion of copyright scope, duration, and the law's indiscriminate use of a per se liability rule, copyright appears to proceed from a view that market failure is the ordinary expectation in the absence of legal intervention. That is a contestable premise, and it is interesting to note that the trend in antitrust law over the past quarter-century is in some tension with copyright's view of markets as fragile. Antitrust-which, because of its dominant rule of reason methodology has developed a much deeper understanding of the likely harmfulness of a wide spectrum of potentially anticompetitive conduct-has migrated toward a concept of markets as robust and not frequently subject to failure, and therefore toward more skeptical treatment overall of claims that particular forms of conduct will suppress competition. Accordingly, the domain of the per se methodology in antitrust has been shrinking. Still in this category is price fixing-the "supreme evil of antitrust" 48 -and related forms of conduct 48. Trinko, 540 U.S. at 408. such as bid-rigging, certain horizontal group boycotts, and market division (whether by territory or customer). But the trend in antitrust is to move other forms of conduct, once carefully fenced within the per se enclosure, toward more circumspect treatment under the rule of reason. Just two years ago, the Supreme Court overturned a century of precedent subjecting minimum resale price maintenance to per se treatment, holding that economic experience had shown that the practice often had procompetitive effects and was therefore appropriately analyzed under the rule of reason. 49 And less dramatically, but no less surely, the Supreme Court and lower federal courts have whittled away at the previous per se test for tying, such that at this point there is little, if anything, left to distinguish the antitrust analysis applying to potentially anticompetitive ties from the ordinary rule of reason. 50 In sum, the clear trend in antitrust law is to subject the great bulk of potentially anticompetitive conduct to the rule of reason. The per se category remains for the few forms of conduct that are believed reliably to lead to competitive harm. This raises an important question: given antitrust's view that markets are in general robust against a variety of threats to competition, should we be skeptical of copyright's implicit assumption that markets for innovation fail readily in the face of copying-even copying that does not appear to displace demand for the plaintiff's original work? Copyright, as it is structured now, does not produce information useful to address this question. This should concern us.

At this point, I will provide a quick summary. Technology has increased enormously the potential uses of creative works. Copyright law has expanded to embrace-i.e., to characterize as infringement-almost all of these uses. Some unauthorized uses of copyrighted works strike directly at authors' incentives to create new works. Other unauthorized uses do not threaten the kind of market failure that copyright exists to preempt. Freeing these uses would create welfare gains, measured in terms of access to works that otherwise would be unavailable.

One would expect, given the above, that copyright would sort the harmful from the harmless uses. Unlike in the case of antitrust, copyright lacks an internal mechanism to do so. Copyright relies instead on fair use, which is not up to the job. What can we do about this? We should consider importing into copyright law an internal sorting mechanism. In designing this mechanism we might take cues from the law of antitrust. If we were to do so, we would define categories of "per se" and "rule of reason" infringement. We would re-structure the copyright law to condemn conduct in the former category based solely on proof that the conduct has occurred. Conduct in the latter category, in contrast, would be condemned only where a plaintiff establishes harm to authors' incentives.

Which brings us to the final question: what kind of conduct belongs in each category? I should preface what follows with an important observation about the stakes-any initial allocation of infringing conduct into per se and rule of reason categories is just that: an initial allocation. One of the signal virtues of an antitrust-style reformulation of copyright liability rules is that the application of rules and standards will be informed over time by information about the harms (or lack thereof) produced by different forms of infringement.

With that in mind, we can identify a very significant category of infringing conduct that belongs in the per se category. I refer to "consumptive" infringement. By "consumptive" I mean forms of infringement involving the reproduction and distribution of copies that are either exact or near enough so that they are almost certain to compete with the original work for patronage.

As an example of consumptive infringement, consider the reproduction and distribution of exact copies of a copyrighted song. Although there are cases in which unauthorized copying and distribution of this kind could benefit an author (for example, by creating demand for a previously unknown work, which the author is able later to exploit), this wholesale copying is likely, over the run of cases, to harm authors by diverting to copyists sales that authors otherwise would have made.

The category of consumptive infringement will apply to a large percentage of copyright cases that plaintiffs will wish to bring. Most copyright violations involve the making and distribution of exact or nearexact copies. Virtually all of the infringement via peer-to-peer networks falls into this category. So for all of these instances, the per se rule for consumptive infringement preserves current law.

Let me offer a few caveats. First, this per se category of "consumptive" infringement should apply only if an exact or near-exact copy has been made and distributed. Accordingly, the per se liability standard should not apply in cases of copying for personal use, including copying for the purpose of time-shifting (i.e., personal use of a VCR or digital video recorder), or device-shifting (as in copying songs from a person's CD collection for listening on that same person's iPod), or to make a back-up copy. Nor should the per se standard apply in cases of "intermediate" copying, by which I mean any copying done not for the purpose of offering the copy in competition with the original, but rather to create some other product or service. The Google Book Search service-a fully-searchable database that eventually will contain many millions of both public domain and copyrighted books-is made possible by intermediate copying. Google makes wholesale digital reproductions of copyrighted books to construct its search database, but the search service returns only "snippets" showing the occurrence of a searched-for term, and does not distribute the copies themselves (except in the case of public domain books or other books for which the rightsholder has agreed to allow distribution of a full-text or partial copy). 51 Because intermediate copying done to offer a separate, non-competing product or service is not very likely to harm copyright owners and depress incentives to invest in new works, this type of conduct should be actionable under a copyright rule of reason only in cases in which a plaintiff is able to establish harm.

Another important example of intermediate copying is the reverse engineering of software. Most software reverse engineering involves the copying of potentially copyrighted computer object code, and then the reconstruction, using engineering techniques, of the human-readable source code. By reconstructing the source code, rival software firms learn how the targeted software works, and how to provide similar or better functionality in their own products.

Assuming for the moment that the reverse engineering is "clean", the process only involves intermediate copying of object code. That is, the rival firm will eventually market a product that may be based in part on what they have learned from copying the target's object code and reverse engineering it back to a facsimile of the target's source code. The rival's product will not, however, distribute any of the copyrighted object code contained in the targeted software. In this case, what emerges from the process is not a copy, but a separate product.

Reverse engineering has been judged fair use in two important cases in the Ninth Circuit. 52 Under the copyright rule of reason, the same result would be obtained in almost all reverse engineering cases, with the exception that the plaintiff would bear the burden of proving harm.

Second, even in cases of copying and distribution of exact or near-51. See About Google Book Search, http://books.google.com/intl/en/googlebooks/history.html (last visited Apr. 5, 2009). 52. See Sony Computer Entm't, Inc. v. Connectix Corp., 203 F.3d 596 (9th Cir. 2000); Sega Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510 (9th Cir. 1992). exact copies, the per se rule should not apply unless a copyright in the plaintiff's work has been timely registered. For reasons I have explained elsewhere, 53 registration is a signal that the copyright owner places some positive economic value on the property right that the copyright law establishes in a creative work. Only in the instances where a copyright owner expects the work to compete and earn returns can we imagine a copy diverting some of the owner's expected market. Infringement claims involving unregistered works should be treated in all instances under a rule of reason.

Classing unregistered works under a rule of reason would substantially ease the current problem with orphan works-i.e., works that are under copyright, but for which would-be users find it difficult or impossible to identify a rightsholder to ask permission to use. Many orphan works are unregistered, and the use of works that lack either registration or some relatively easily-found data regarding ownership is unlikely, in the run of cases, to produce many instances where authors' incentives are harmed. Why is that? Because under current law preinfringement registration provides important benefits, including the availability in infringement actions of very powerful statutory damages 54 and attorney's fees. 55 Accordingly, the failure to register a work is a signal that the author does not expect that work to produce a significant return. In such cases, any presumption that unauthorized reproduction and distribution harms the author's incentives to create new work is unjustified, for the copying cannot affect incentives if the expected return from a work is at or near zero.

Of course, harm is possible in such cases-for example, if the copying is undertaken as part of the creation of a derivative work of a type that the author can show that he intended to produce, or to license. The plaintiff should be put to the proof of such harm-i.e., the plaintiff should produce evidence that he was likely to enter such a market, and also that presence of a competing derivative was likely to displace demand rather than create it. We should expect that plaintiffs will rarely be able to make such a showing-as Stewart Sterk has noted, situations where control of derivative works makes a substantial difference in an author's ability to recover his costs are quite rare. 56 53. See Sprigman, supra note 7, at 502-14. 54. 17 U.S.C. § 504(c)(1). 55. 17 U. S.C. § 412. 56. Stewart Sterk, Rhetoric and Reality in Copyright Law, 94 MICH. L. REV. 1197, 1215-17 (1996) ( One argument for giving authors copyright in derivative works is that the prospect of profits from derivative works is necessary to create adequate incentives for production of the original. The argument is persuasive only in those situations when (1) the projected returns from the original work are too small to justify the costs of A third caveat is that the definition of consumptive infringement must be carefully drawn to avoid the temptation to which copyright law's general "substantial similarity" liability standard 57 has already fallen victim-i.e., the condemnation of any non-de minimis use of protected material. The gravamen of the consumptive infringement category is our strong intuition that production and distribution of exact or nearly exact copies of commercially valuable works will divert some increment of the potential audience for the original to the copy. That intuition is strong enough to support the application-at least initially-of a per se rule to all such infringing conduct. This is not to say that every instance of distribution of even exact copies will inevitably harm the rightsholder in the original work. One might imagine, for example, instances in which the unauthorized distribution of copies of a song could lead to a bandwagon effect that turns the song into a hit, thereby raising consumption of the original-even in the face of the copying-far beyond what circumstances might otherwise have provided. The question is not whether there will be some inaccuracy-some inaccuracy is in the nature of a per se rule. The question is whether the gain in clear understanding of proscribed conduct, stable expectations, and simplified enforcement that a rule provides are worth the occasional misapplication of the law. For copies that are exact or nearly so, the answer to that question is, at least on current evidence, yes.

Now that we have developed, at least in brief, a concept of consumptive infringement subject to a per se liability rule, let us move to the second and in some ways more interesting category. What kinds of infringing conduct belong under the rule of reason? Anything that does not fall under the category of presumptively consumptive use. I have already mentioned personal uses, intermediate copying, and the use of production, and (2) the projected returns from the derivative work are so large relative to the cost of producing the derivative work that the difference will more than make up the projected deficit on the original work alone. These conditions may apply when the original work is an extraordinarily high-budget movie with the potential for sales of toys, t-shirts, and the like, but they are less likely to apply in more common derivative-works cases. ).

57. "Substantial similarity" has long been the predominant liability standard in copyright law. In assessing whether two works are "substantially similar", a court will first look to whether the defendant in fact copied plaintiff's work. In doing so, the court will assess whether the defendant had access to plaintiff's work, and whether the works are similar enough that an inference of copying is warranted. If the court finds copying in fact, it will then inquire whether that copying went too far -i.e., whether it involved non-de minimis appropriation of protected material -so as to constitute an improper appropriation. See Arnstein v. Porter, 154 F.2d 464, 468 (2d Cir. 1946).

unregistered or orphaned works. These are all important, but the most significant cases in this category are likely to involve the creation of derivative works.

Depending on the circumstances in particular cases, a derivative work may, or may not, compete with the original. What, then, should we do with derivative works which use substantial portions of an original work, but are neither exact copies nor nearly so? For these works, we should apply the rule of reason, conditioning liability on the plaintiff's proof of the particular types of harm that we might imagine.

The plaintiff might seek to prove, for instance, that there is significant cross-elasticity of demand between the original and the derivative, and therefore allowing unauthorized reproduction and distribution of the derivative will in fact divert enough of the sales that the original author might otherwise have enjoyed that the court is able reasonably to conclude that the loss, if assessed ex ante, would have affected the author's incentive to create. A plaintiff might make such a showing by producing survey evidence of cross-elasticity, in just the way that plaintiffs in antitrust cases are required to support allegations of competitive harm by demonstrating (often via evidence about crosselasticity of demand) that they compete with the defendant in a relevant product or service market. Another good example is the burden of proof that must be discharged by plaintiffs pressing trademark confusion claims in Lanham Act cases. 58 These plaintiffs are required to introduce evidence that consumers actually are confused when presented with a senior mark and a similar junior mark-i.e., direct evidence of the kind of harm that the trademark law seeks to prevent. 59 Similarly, in cases where the plaintiff claims preemption of a derivative market or a licensing market, the impugned derivative work may represent a market that the original work's author may wish to enter (as either an author or a licensor), or it may not. In cases involving alleged preemption of licensing opportunities, plaintiffs should be put to proof that they have exploited similar licensing markets in the past, or that similarly-situated rightsholders have done so, or that they stand ready and have taken steps to do so. A bare desire to collect the rents that the defendant has collected is not enough.

Plaintiffs should also be required to demonstrate that the licensing market at issue is one that would be economic for them to enter-i.e., that the transaction costs of licensing would not overwhelm the likely revenues that might be obtained. This is an inversion of Wendy Gordon's suggestion, in a classic article, that the fair use doctrine should free uses from copyright liability where the transaction costs of licensing overwhelm the value of the use. 60 Working within copyright law's current structure, the burden of establishing this market failure rests on the defendant as part of the fair use analysis. Under the revised structure described here, I would place a burden to establish "market viability" on the plaintiff as part of his prima facie case.

Moreover, and importantly, even if the derivative market is one which the author would be likely to enter, and even if licensing is otherwise economic, the plaintiff must show that this market is significant enough that, if viewed from the perspective of the author ex ante the creation of the original work, the freedom of another author to occupy that market with an unauthorized derivative would be likely to have an appreciable effect on the author's innovation incentives. For this showing to be made, the harm of market preemption must be substantial relative to the total expected return from the original work. Substantial harms are more likely to be foreseeable-but I should emphasize that the use of foreseeability that I advocate here is not the same as that advanced by Balganesh or Bohannan. I would not have the court determine whether the use was in fact objectively foreseeable ex ante the creation of plaintiff's work. I would, instead, focus on an importantly different counterfactual-i.e., whether the use, if in fact foreseen ex ante, would have meaningfully affected the author's creative incentives.

I should admit here, if it is not already obvious, that none of the changes to copyright law that I have proposed will make the ultimate question of what counts as harm in copyright less difficult or contentious. The modest but, I believe, nonetheless helpful, effect of my suggested changes to the structure of copyright would be to re-distribute the burdens of uncertainty. At present they rest almost entirely on defendants. Under my division and re-ordering of cases, some of the burden of uncertainty in copyright rule of reason cases would be transferred to plaintiffs.

The effect of this shift would be to shape the incentives of those copyright owners who are considering bringing suit. The burden of establishing harm will create expense and uncertainty for potential plaintiffs in copyright disputes classed under the rule of reason. Plaintiffs who believe that they have been substantially harmed, and who consider themselves in a position to offer evidence about harm, will come forward in the expectation that the burden of proving their case ultimately will be 60. Wendy J. Gordon, Fair Use As Market Failure: A Structural and Economic Analysis of the Betamax Case And Its Predecessors, 82 COLUM. L. REV. 1600 (1982).

borne by the defendant (under copyright law's provision for shifting costs and attorneys fees). 61 Those who are not substantially harmed, will not file suit-litigation costs and the necessity of proving harm will, for these plaintiffs, serve as a screen. 62 This is a desirable outcome-both limiting litigation and forcing prospective plaintiffs to think hard about their damages before filing suit. Of course we should worry about potential plaintiffs who believe that they have been harmed but cannot precisely quantify such harms. Nothing I have suggested requires precise quantification (or even nearly so). To make out a prima facie case in a copyright dispute governed by the rule of reason, plaintiffs will be obliged to describe their harm, and to show that it is substantial enough that it could have affected incentives if considered ex ante. Such evidence will usually be more readily available to plaintiffs than to defendants. It is therefore on plaintiffs-at least in rule of reason cases-that the burden of proof should fall.

These, then, are the structural shifts that should guide an efficient reform of copyright law. The basic principles are, however, friendly to at least two very different implementations. The first, and most direct, I have treated at length. It would be to require, as a predicate to liability in any case involving infringing conduct with ambiguous welfare effects, proof of actual or likely harm to authors' incentives.

The second would be, in some ways, conceptually more modest but nonetheless perhaps more demanding of plaintiffs in actual cases. It would be to limit damages and other remedies for conduct falling within the rule of reason. That is, while plaintiffs in copyright rule of reason cases would not be required to prove harm as a predicate to a liability finding, their recoveries would be limited to proved actual damages. Neither statutory damages nor attorney's fees would be recoverable, and injunctions would be available only where the plaintiff succeeds in showing harm irreparable via an award of damages.

The screening effect of the second implementation would be similar to the first. Only plaintiffs who have suffered substantial harms would bring suit for infringement falling within a rule of reason. Plaintiffs unable to show such harms would find the prospect of filing a lawsuit unattractive. This is precisely the sorting mechanism we would want for an efficient copyright system. This implementation would, however, require plaintiffs to prove their damages with greater precision. § 505 (2006). 62. For a valuable explanation of the potential role of litigation costs in efficiently shaping incentives to sue in the case of incomplete contracts, see Albert Choi & George Triantis, Completing Contracts in the Shadow of Costly Verification, 37 J. LEG. STUD. 503 (2008).

The copyright law should be reformed to differentiate its treatment of conduct likely over the run of cases to lead to harm to authors' incentives from other types of conduct that have more ambiguous and context-dependent effects.

I have identified a category of "consumptive" uses as the type of infringement about which copyright need be most concerned. That conduct should be subject to a strong rule-based proscription. All other conduct should be treated according to a more sensitive standard, where plaintiffs have the opportunity to enjoin the conduct and recover damages when they can show that they have been harmed, but cannot do so otherwise.

These suggestions do not in themselves supply an answer to the most basic and enduring problem in copyright law-i.e., what counts as a relevant harm. They do, however, shape incentives in ways that will teach us more over time about the circumstances in which the many and varying forms of infringing conduct cause harm, and when they do not. This is valuable information the production of which the current structure of copyright law does not encourage.