Reconciling Intellectual and Personal Property

better part of the last century, we witnessed a well-balanced equilibrium. 4 Copyright law successfully mediated this tension through the principle of exhaustion-the notion that once a rightsholder transfers a copy of a work to a new owner, its rights against that owner are diminished. 5 The rightsholder's power to prevent distributing, using, or sometimes even reproducing a work must yield to the personal property interests of consumers. As a result, libraries can lend their books, 6 museums can display their paintings, 7 and consumers can back up their software. 8 Rather than an idiosyncratic carveout or exception, exhaustion is an inherent part of copyright law's balance between the rights of creators and the rights of the public. It is a fundamental component of almost every intellectual property system. 9 Just as fair use balances the competing interests of original and follow-on creators, exhaustion accounts for and accommodates the rights of both creators and consumers. And it recognizes that those rights are not at odds with the goals of the copyright system, but at its core. Meaningful consumer rights to use and transfer their personal property are "intellectual property rights often operate to restrain the owners of tangible things from their ordinarily privileged uses of those things").

4 Some have contended that personal and intellectual property rights are fundamentally incompatible. See Tom G. Palmer, Intellectual Property: A Non-Posnerian Law and Economics Approach, 12 HAMLINE L. REV. 261, 281 (1989) ("[A]ny system of 'property rights' that requires the violation of other property rights, e.g., the right to determine the peaceful use in one's home of one's own videocassette recorder or to purchase blank tapes without paying a royalty to a third party, is no system of rights at all. In short, a system of intellectual property rights is not compossible with a system of property rights to tangible objects . . . ."); see also Hinton v. Donaldson (Sess. 1793), in THE DECISION OF THE COURT OF SESSION, UPON THE QUESTION OF LITERARY PROPERTY 19 (Edinburgh, James Boswell 1774) (describing a common law copyright claim as "far from being founded on property, [but] inconsistent with it . . . [i]t is a rule in all laws, that the commerce of moveables ought to be free; and yet, according to the pursuer's doctrine, the property of moveables may be subjected to endless limitations and restrictions that hitherto have not been thought of, and would render the commerce of moveables extremely hazardous").

5 (a). Exhaustion has its limits, of course. The owner of a copy of a video game cannot reproduce copies for all of her friends. See Perzanowski & Schultz, supra note 5, at 927-28 (discussing the common law limits on exhaustion). 9 Exhaustion doctrines have a long history in patent and trademark law as well. See Quanta Computer, Inc. v. LG Elec., Inc., 553 U.S. 617, 621 (2008) ("For over 150 years this Court has applied the doctrine of patent exhaustion to limit the patent rights that survive the initial authorized sale of a patented item."); Sebastian Int'l v. Longs Drug Stores Corp., 53 F.3d 1073, 1074 (9th Cir. 1995) ("Since 1924, courts have recognized a basic limitation on the right of a trademark owner . . . . [C]ourts have consistently held that . . . the right of a producer to control distribution of its trademarked product does not extend beyond the first sale of the product.").

technologies to prevent the resale and use of pre-owned media; 16 they have used spurious takedown notices to remove used items from secondary markets like eBay. 17 These efforts have met with mixed success.

But two other developments have proven much more effective in curtailing exhaustion and threatening consumer interests. First, content owners, particularly in the software industry, have endeavored to eliminate the personal property interests of consumers, redefining the notion of ownership by characterizing their transactions with consumers as licenses to use the works or the purchase of a license as opposed to the purchase of a copy. 18 Because ownership triggers exhaustion, this approach has allowed rightsholders to assert control over subsequent uses and transfers of those copies, unchecked by countervailing consumer property interests. And because digital media content, as both a legal and practical matter, is increasingly indistinguishable from software, the entire copyright economy could soon be governed by this same licensing regime. 19 Second, the tangible copy is rapidly disappearing from copyright markets. Once characterized by the distribution of physical products, these markets are now defined by exchanges of networked information. 20 In the past, a copy was easy for rightsholders, consumers, and judges to identify and accommodate. Copies were physical artifacts that resided on store shelves. Today, however, digital sales of books, 21 music, 22 video games, 23 and applications 24 are outstripping traditional tangible media. Rather than picking up books and records from store shelves, we stream, download, and store con-tent in the cloud. And rather than enjoying a single copy in multiple contexts-reading a book at home, on the bus, at a friend's house, or on vacation-we now shift between multiple copies rapidly and dynamically across various devices and platforms. This shift has profound implications for exhaustion and consumer property rights. Since transferring digital content from one consumer to another generally implicates the reproduction right, such actions have been found to fall outside the literal scope of copyright law's statutory first sale doctrine. 25 So although a consumer is entitled to sell a used paperback, transferring a used e-book presents legal uncertainty. More broadly, digital distribution changes our relationship to the copy. Many consumers no longer experience the texture of a printed page, the groove of a vinyl record, or even the smooth plastic of a DVD. Instead, our access to the copies we own is often mediated by third parties like cloud storage providers. And even when we store copies locally, our experience of them is less immediate and less tangible than their analog counterparts.

In a marketplace that undermines ownership and deemphasizes copies, the future of consumer property interests looks bleak. According to copyright holders, you do not own the digital media you purchase online. According to the Ninth Circuit, you do not own the plastic disc on which your software programs are encoded. 27 And according to the Librarian of Congress, you may not own your cell phone or gaming console. 28 In short, the very notion of personal property rights in any embodiment of a work protected by copyright is under attack.

The equilibrium between personal and intellectual property that exhaustion enabled depends on doctrinal assumptions about the copyright marketplace that are quickly becoming outdated. One response to these developments is to concede that both exhaustion and consumer property rights are a thing of the past. But a copyright system governed solely by terms dictated by rightsholders tips the scales too far in their favor, risking broad damage to copyright as a body of law. Rather than abandoning the counterweight of consumer rights, we argue it must be preserved, but in a manner that recognizes the differences between digital and analog distribution.

This Article builds on our earlier work on exhaustion. We have previously emphasized the common law origins of copyright exhaustion, arguing 25 See Capitol Records, LLC v. ReDigi Inc., 934 F. Supp. 2d 640, 655 (S.D.N.Y. 2013) (holding that "the first sale doctrine does not protect ReDigi's distribution of Capitol's copyrighted works" because "it is distributing reproductions of the copyrighted code").

26 See infra notes 143-56 and accompanying text. 27 See Vernor v. Autodesk, Inc., 621 F.3d 1102, 1111 (9th Cir. 2010). 28 See MARIA A. PALLANTE, REGISTER OF COPYRIGHTS, SECTION 1201 RULEMAKING (2012) (rejecting proposed exemption that would have enabled maintenance or repair of a gaming console under § 117); id. at 92 (noting the "lack of certainty in the law [that] makes it impossible" to determine the question of ownership of wireless handsets). To be clear, the formal question at issue in the rulemaking was whether consumers own the copies of the software embedded in their devices. But given the deeply integrated nature of hardware and software in both contexts, ownership of the embedded software copy and ownership of the device effectively collapse.

for a judicial interpretation that is more expansive than the narrow statutory first sale rule. 29 Subsequently, we have advocated increased reliance on exhaustion to resolve a range of disputes over personal use by consumers that are typically analyzed through the lenses of fair use and implied license. 30 And, most recently, we have outlined two competing legislative frameworks for a contemporary exhaustion regime as a part of the broader copyright reform effort. 31 This Article examines both the forces undermining copy ownership and the important functions it serves within the copyright system in order to construct a workable notion of consumer property rights in digital media.

Part I begins by examining the relationship between intellectual and personal property. Sometimes courts have treated those rights as inseparable, as if transfer of a copy entails transfer of the intangible right, or retention of the copyright entails ongoing control over particular copies. But Congress and most courts have recognized personal and intellectual property as interests that can be transferred separately. Although the better view, this approach frequently overstates the independence of copyrights and rights in copies. Those interests interact; each helps to define the boundary of the other. The exhaustion principle, though historically associated with a clear distinction between copy and copyright, is in fact the primary tool in copyright law for mediating the somewhat indistinct line separating the copy and the work.

Part II begins to outline the breakdown of this once stable equilibrium, focusing on the erosion of the notion of consumer ownership. In recent decades, courts have created two distinct regimes for resolving questions of copy ownership: one that applies to software and one that applies to everything else. The software regime endorses rightsholders' efforts to "license" particular copies of their works, in contrast to the general skepticism with which courts regard such efforts. This dichotomy is driven in part by software exceptionalism-the notion that for a variety of reasons software should be treated differently. But the growing acceptance of the licensing model also reflects changing views of property. Those shifts opened the door to the substitution of statutory property rights with unilateral contract terms. As the line separating software from other media becomes increasingly blurred, the thinking reflected in the software cases suggests a creeping erosion of copy ownership.

But as Part III details, the erosion of ownership is only half of the problem. The copy, once the uncontroversial locus of consumer property rights, has transformed as well. Copies were once persistent, valuable, readily identifiable, and easily accounted. But the days of the unitary copy are numbered. Today, copies are discarnate, ephemeral, ubiquitous, and of little value in themselves. In part these changes reflect shifting consumer demand. But more importantly, they signal the increasing disconnect between copyright law's conception of the copy and today's technological reality. We are witnessing a blurring of the formerly clear distinction between the intangible work and the tangible copy, a distinction that has been central to copyright law's approach to balancing intellectual and personal property rights.

In light of the challenge of squaring the existing doctrinal framework for exhaustion with these developments, we turn in Part IV to an examination of the functions served by copy ownership. With a better understanding of the role of copy ownership within the copyright system, we will be better positioned to craft an approach to consumer property rights in the post-copy era. We identify three primary functions of copy ownership. First, locating consumer rights in a particular copy helps preserve the rivalry that distinguishes real property from intellectual property, thus preventing consumer rights from unduly interfering with the ability of the copyright holder's to exploit the work. Second, copy ownership encourages consumers to participate in copyright markets rather than rely on unauthorized sources of content. Third, a stable and reliable notion of copy ownership reduces information cost externalities by eliminating idiosyncratic transfers of rights in copies.

Part V argues that while these three functions historically have been bound up in the single, unitary copy that defined distribution in the nineteenth and twentieth centuries, the copy is not essential to achieving those goals. Instead, the copy served as a token, signaling that each of the three functional concerns at the heart of exhaustion was satisfied. With this new understanding of the place of the copy, we outline the structure of a new exhaustion doctrine that more carefully and transparently interrogates exhaustion's underlying policy concerns, rather than using the unitary copy as a proxy for consumer rights.

Copyright law's fixation requirement guarantees that from the moment of initial protection a work exists in two distinct forms-first, as an incorporeal expression of ideas, and second, as a tangible embodiment of that expression. 32 As a result, copyright law must confront the relationship, and potential conflict, between rights in the ethereal work and rights in the fixed copy. That relationship is not immutable; we can imagine several ways to structure it. We might, for example, think of those two sets of rights as bound together. Some courts adopted this approach, holding that ownership of a tangible copy entails ownership of the copyright in the underlying work. Conversely, some cases addressing ownership of copies of computer software hold just the opposite-that by virtue of copyright ownership, rightsholders can assert ownership over physical copies seemingly purchased by consumers. But for three hundred years, most courts have rejected the notion that control over one set of interests necessarily confers control over the other. Instead, they have taken a third tack, one that recognizes that ownership of a copyright is distinct from ownership of a copy. Under this approach, the transfer of a painting or manuscript does not affect an assignment of copyright. Nor does copyright ownership entitle a rightsholder to lay claim to all copies of a work.

Merely recognizing, however, that the transfer of one set of interests can occur independently from the other does not fully reveal the relationship between them. As this Part will argue, even if these two interests are conceptually distinguishable, they are deeply intertwined. Their boundaries are defined, at least in part, in reference to each other. While most courts have understood that personal and intellectual property are distinct, they have failed to fully appreciate the extent to which they are interdependent.

A minority of courts confronting the tension between intellectual and personal property interests have staked out two inconsistent and equally extreme positions, both of which effectively merge those interests into an inseparable bundle. Early courts favored copy owners in this tug-of-war. Later, courts confronted with computer software disputes swung the pendulum in the other direction, allowing copyright owners to control copies of their programs distributed to the public. Some courts held that the transfer of title to a particular copy of a work entailed a transfer of copyright ownership in the underlying work. For example, in Parton v. Prang an artist sued the purchaser of an original painting, seeking to enjoin the purchaser's reproduction and sale of copies of the pictorial work. 33 According to Parton, when he sold his painting, "he only parted with the result of his labor as property." 34 In other words, he sold his personal property in the canvas, but "the right to multiply copies of the picture was vested in him as the author and proprietor of the same, and that he still retain[ed] that right notwithstanding the sale." 35 The court disagreed, holding that an unconditional sale of the painting "had the effect to transfer to the respondent the right to reproduce or chromo the picture." 36 According to the court, "it is well settled law that even copyright is an incident to the ownership of a manuscript, and that it passes at common law with the transfer of a work of art." 37 Endorsing an 1860 Irish decision, 38 the court thought it a "strange proposition" that transfer of the copy would not simultaneously transfer "that which principally con- stitutes the value of the thing transferred," namely the right to reproduce copies. 39 Because Parton's sale of his painting was absolute and unconditional, his "entire property was transferred." 40 Seventy years later, the New York Court of Appeals followed similar logic in Pushman v. New York Graphic Society. 41 There, Pushman sold his painting to a gallery without expressly permitting or forbidding its reproduction. When Pushman sought to enjoin the gallery's reproduction, the court sided with the owners of the copy, reasoning that "an artist must, if he wishes to retain or protect the reproduction right, make some reservation of that right when he sells the painting." 42 According to the court, the failure to do so "convey[s] the artist's whole property in his picture"-both intellectual and personal. 43 Although both of these cases were decided as matters of common law copyright, they demonstrate both the mutable relationship between personal and intellectual property rights and the problems that flow from treating one as nothing more than an incident of the other. By failing to distinguish between-and more importantly, balance-these competing interests, the rules applied by these courts would undermine the incentive structure of copyright law. If the sale of one copy, let alone thousands or millions, transferred all rights to the underlying work, incentives to make works available to the public and perhaps to create them in the first instance would be significantly reduced, frustrating the two key policy objectives of copyright law.

Equally troubling are cases that treat copyright ownership as determinative of personal property interests. As discussed in greater detail in Part II, the computer software cases have been anything but uniform in their resolution of the competing claims of intellectual and personal property holders. At their least nuanced, courts addressing disputes over ownership of copies of software have embraced, in the words of one commentator, a "magic words" approach. 44 These courts allowed copyright holders to control the disposition of copies so long as they asserted that they licensed rather than sold their software. In other words, software makers were deemed to own all copies of their programs by virtue of their copyright ownership of the underlying literary work, regardless of the nature of the transaction that placed those copies in the hands of consumers.

Perhaps the most influential instance of this equation of copyright ownership with copy ownership is MAI Systems Corp. v. Peak Computer, Inc. 45 puter owners. 46 To prevent Peak from servicing MAI computers, MAI sued for copyright infringement, alleging that Peak created unlawful reproductions of MAI's software programs by loading MAI software in RAM when it powered on a client's computer. 47 Peak argued that these RAM instantiations were insufficiently durable to qualify as fixed copies. 48 After rejecting that argument, the Ninth Circuit noted in passing that Peak could not avail itself of the "essential step" copy defense in section 117 of the Copyright Act. That provision allows the owner of a copy of a program "to make or authorize the making of another copy" if doing so is "an essential step in the utilization of the computer program." 49 But as the Ninth Circuit tersely explained in a footnote, "[s]ince MAI licensed its software, the Peak customers do not qualify as 'owners' of the software." 50 The Ninth Circuit provided no factual or legal support for its conclusion that Peak customers did not own the copies of MAI software residing on their computers. Presumably, it relied on the district court's factual finding that the software "ha[s] been licensed . . . [and] ha[s] not been sold." 51 But that finding was supported by nothing more than MAI's own declaration. 52 Similarly, the court in Microsoft v. Harmony equated ownership of the copyright in a program and ownership of particular copies of that program. 53 When Microsoft sued for unauthorized distribution of copies of MS-DOS and Windows, Harmony argued its distribution was permitted under the first sale doctrine, which allows "the owner of a particular copy . . . to sell or otherwise dispose of the possession of that copy." 54 But Microsoft contended, and the court agreed, that Harmony was not the owner of the copies at issue because Microsoft "established a course of conduct . . . consistent with an intention to retain all the rights associated with the grant of copyright" in the software. 55 Because Microsoft retained ownership of the intel-lectual property in its programs, according to the court, it retained any personal property interests in tangible copies of those programs as well. 56 Determining whether a sale that transfers a personal property interest to a consumer has occurred is an inquiry that should be guided by the facts and circumstances surrounding a particular transaction. Consumers can, of course, possess copies of works that they do not own. Library patrons are not owners of the books they borrow. Lessees of computers do not own the copies of software installed on them. And subscribers to cloud-based software services do not own the copies they access online. But the courts in these cases did not interrogate the details, or even the basic structure, of the transactions. Instead, they reduced the question to whether the copyright holder asserted that it licensed rather than sold its software. In doing so, these cases effectively link copyright ownership and copy ownership. So long as the rightsholder retains its intellectual property interest, it can assert a personal property interest over copies.

In many ways, this outcome is the mirror image of the common law copyright cases dealing with original paintings and manuscripts. In those cases, the transfer of a copy entailed the transfer of the copyright. In the "magic words" software cases, retention of the copyright entailed retention of personal property rights in copies distributed to third parties. Fundamentally, both sets of cases treat intellectual and personal property interests as bound together.

But while the courts in the common law copyright cases explicitly acknowledged the link between intellectual and personal property interests, the software cases appear to result from a failure to distinguish between the work and the copy. Both the first sale doctrine and § 117 are concerned with the "owner of a . . . copy." 57 But when the MAI and Harmony courts talked about ownership of "the software," they drew no distinction between owning the copyrighted work and owning a particular copy of it. 58 MAI may well have licensed certain uses of its copyrighted code. But that fact is entirely consistent with MAI transferring ownership of particular copies of that code. 59 By treating sales and licenses as mutually exclusive categories of transactions, the courts in those cases tied intellectual and personal property rights together.

Allowing copyright holders to assert control over copies is just as corrosive to the incentive structure of the copyright marketplace as treating the transfer of a copy as an assignment of copyright. Rather than undermining authorial incentives, the approach taken in the software cases reduces the motivation of consumers to participate in lawful copyright markets. Consumers are inclined to pay for copies because those copies impart value. The rights associated with copy ownership offer assurances of autonomy, persis- tence, and simplicity. If consumers are not treated as the owners of content they purchase, the value proposition presented by lawful copyright markets is diminished. If consumers are not owners, but licensees subject to the changing whims of copyright holders who can restrict otherwise lawful uses or deny access altogether, more consumers will simply opt out of the lawful copyright marketplace. 60

Despite the outlier cases discussed above, courts have historically distinguished between the rights of copyright holders and the rights of copy. In 1741, Pope v. Curl held that ownership of letters written by Alexander Pope did not entitle the copy owner to reprint and publish their contents. 61 More than a century later, the Supreme Court reached the same conclusion in Stephens v. Cady. 62 There the Court held that ownership of a copperplate, acquired at a sheriff's auction, did not entitle its owner to reproduce the map engraved on it. 63 As the Court explained, "the property acquired by the sale in the engraved plate, and the copy-right of the map secured to the author under the act of Congress, are altogether different and independent of each other, and have no necessary connection." 64 Copyright interests are "detached from the manuscript, or any other physical existence." 65 Revisiting the question a few years later, the Court explained that the owner of the copperplate is entitled to make a great many uses of his personal property, so long as they do not conflict with the interests of the copyright holder:

He may make any other use of the plate of which it is susceptible. He may keep it till the limited time, during which the exclusive right exists, shall have expired, and then use it to print maps. He may sell it to another, who has the right to print and publish, but he can no more use that right of property than he can use a press, or paper, which belongs to a third person. 66 But as the foregoing discussion suggests, some courts remained reluctant to distinguish between personal and intellectual property rights. Frustrated by the lingering inconsistency, Congress attempted to provide some clarity by explicitly rejecting the rule embodied in Pushman. 67 Section 202 of the Copyright Act provides:

Ownership of a copyright, or of any of the exclusive rights under a copyright, is distinct from ownership of any material object in which the work is embodied. Transfer of ownership of any material object, including the copy or phonorecord in which the work is first fixed, does not of itself convey any rights in the copyrighted work embodied in the object; nor, in the absence of an agreement, does transfer of ownership of a copyright or of any exclusive rights under a copyright convey property rights in any material object. 68 Recognizing that intellectual and personal property interests are separable, however, provides only a partial explanation of the relationship between them. The proposition that they can be separately transferred reveals little about the interplay between the rights of copy and copyright owners. Courts seem to take for granted that because these rights are distinct, they are entirely unrelated. Stephens v. Cady, for example, repeatedly stresses that the copyright in the map and the property rights in the copperplate are "wholly independent," "disconnected," and "altogether different." 69 To the extent these statements suggest that the substantive scope of personal and intellectual property rights are unrelated, they underestimate the extent to which those rights are intertwined.

Although distinct, the rights of copy owners and copyright holders share a rather porous and variable boundary. Their scope is defined, in part, in reference to each other. At the intersection of these two sets of rights, as one expands, the other necessarily contracts. So what it means to own a copy depends on what rights are reserved for the copyright holder, and vice versa. If the courts or Congress, for example, decide that ownership of a copy entails the right to publicly perform the work it contains, the rights of copy owners would be enlarged and those of copyright holders diminished. Rather than disconnected, these rights are deeply and inescapably interdependent. In this sense, consumers and copyright holders draw from a shared pool of potential rights in defining their respective property interests.

By deciding that personal and intellectual property interests are distinct, the courts and Congress are confronted with the separate but related question of appropriately balancing those respective rights. Whether they are aware of it, and whether they conceptualize it in these terms, courts and Congress cannot avoid apportioning rights between owners of copies and owners of copyrights. Any line separating the two reflects a policy choice on the question of apportionment. But that choice is not static. It is one that has been adjusted and calibrated over time in response to changing technological and market conditions.

The primary tool for fine-tuning this balance is the principle of copyright exhaustion. That principle holds that copyright owners cede the right to make some otherwise protected uses of their works to copy owners after a lawful sale or other transfer of ownership of a copy. 70 The first sale doctrine, which permits copy owners to sell or otherwise dispose of their copies despite the copyright holder's exclusive right of distribution, is the most important exhaustion rule. 71 Although its roots can be traced much earlier, 72 the Supreme Court first explicitly embraced the first sale doctrine in Bobbs-Merrill Co. v. Straus. 73 The Court held that after a publisher sold copies of its work, its right to control the terms under which those copies were distributed was exhausted. 74 Although the Court characterized its decision as an exercise in statutory interpretation of the scope of the exclusive right to "vend," Bobbs-Merrill is better understood as an example of the sort of common law adjudication that has historically shaped much of copyright law. 75 In any case, by deciding that the rights of copy owners to resell their books trumped the rights of copyright holders to control downstream distribution, the Court staked out a boundary between personal and intellectual property. And its role in determining apportionment was inescapable. Had the Court held that publishers can restrict the terms of subsequent distribution of copies, it would have shifted the boundary in a way that expanded intellectual property at the expense of personal property. But it could not avoid the tradeoff between the two.

Both the courts and Congress have relied on the exhaustion doctrine to alter the balance between copy owners and copyright holders. A number of early courts held that copy ownership justified the creation of reproductions and derivative works to the extent necessary to realize the value of a copy. 76 More recently courts have been asked to apply the exhaustion principle to the transfer of digital copies, so far with less favorable results for copy owners. 77 Congress has enacted exhaustion rules that favor copy owners-permitting the public display of works and the creation of archival and essential step copies of software-and others that favor copyright holders-restricting the rental of phonorecords and copies of computer software. 78 Within this framework, policymakers enjoy considerable flexibility in determining which rights are controlled by copyright holders and which are enjoyed by copy owners. Whether undertaken by the courts or Congress, this fine-tuning of the balance between personal and intellectual property has been premised on a functioning exhaustion doctrine. Exhaustion, in turn, is contingent on a meaningful notion of copy ownership. But as the next Part describes, copy ownership is increasingly at risk of becoming a relic of the past.

When the Supreme Court decided Bobbs-Merrill in 1908, the question of copy ownership was a simple one. The Court understood that a publisher's attempt to restrain subsequent alienation of books through a printed notice did not undermine the personal property interests of book owners. Despite the publisher's attempt to limit resale, those who bought books owned them. 80 A century later, the question of whether unilateral restrictions announced by copyright holders are sufficient to subvert copy ownership is one plagued by confusion and uncertainty.

In large part, that uncertainty emerged from disputes over computer software. This Part begins by describing the courts' treatment of copy ownership in the software cases. Although unable to settle on a single approach for analyzing copy ownership, many courts have endorsed efforts by copyright holders to characterize software transactions as licenses rather than sales. We offer two sets of explanations for the growing acceptance among courts of "licensing" particular copies of a work. First, some historical and functional characteristics unique to software may have convinced courts that it is particularly well-suited to a licensing regime. Second, and more generally, the conception of property as an infinitely malleable bundle of rights has left courts susceptible to strategic efforts by rightsholders to define and parcel out interests in copies as they see fit. Because this broader shift in thinking about property extends well beyond software, and because software is increasingly indistinguishable from the rest of copyrightable subject matter, copy ownership and exhaustion are at risk.

Disputes over copy ownership arise in two related contexts in software copyright law. Sections 109 and 117 of the Copyright Act both offer copy owners defenses for otherwise infringing acts. Again, the first sale doctrine allows for resale or other distribution of a copy by its owner. 81 Section offers targeted protection to owners of copies of computer programs, permitting them to make copies or adaptations essential to the use or backup of the program. 82 Under both of these exhaustion rules, the core question is the same-whether the defendant is the owner of a copy of the software program. 83 Courts have struggled to answer this seemingly straightforward question, adopting a range of inconsistent approaches often without clearly articulating any underlying guiding principle. 84 As discussed above, the least nuanced approach allows copyright holders to avoid the consequences of exhaustion simply by unilaterally declaring that they are licensing, rather than selling, their products. Early cases like MAI typify this approach. But the Ninth Circuit has deployed similar reasoning as recently as 2006, where the court explained that "if the copyright owner makes it clear that she or he is granting only a license to the copy of software and imposes significant restrictions on the purchaser's ability to redistribute or transfer that copy, the purchaser is considered a licensee, not an owner, of the software." 85 Other courts probe the declarations of copyright holders somewhat more meaningfully, looking to the specific terms of the purported license agreement. Courts have held that where an agreement specifies a fixed duration, requires ongoing payments, or the return or destruction of the copies, they are licensed rather than sold. 86 More controversially, some courts have viewed provisions that restrict the manner in which the software can be used as undermining copy ownership. Terms that limit the number of users or specify hardware requirements, for example, have convinced some courts that a defendant did not own their particular tangible copy of the program. 87 Still other courts have turned to the Uniform Commercial Code to decide whether defendants hold title to their copy of a work. 88 The basic rule under section 2-401 is that a seller's reservation of title to goods shipped or delivered to a buyer is, at best, considered a reservation of security interests. 89 So where goods are shipped or delivered, and full payment is rendered, the seller has no claim of ownership. 90 83 More recently, the Supreme Court resolved a dispute over the other central question in first sale cases, whether or not the copy was lawfully made. In Kirtsaeng v. John Wiley & Sons, Inc., the Court rejected the argument put forth by publishers that only copies made within the territorial border of the United States were "lawfully made" for first sale purposes. 133 S. Ct. 1351, 1355-56 (2013).

84 For a thorough taxonomy of these cases and their competing approaches, see Finally, some courts have proven willing to look beyond both the licensing terms and formal title transfer to interrogate the nature of the economic transaction between the rightsholder and the copy owner. Transactions characterized by single payments and perpetual rights of possession and use have been deemed sales regardless of the rightsholder's contentions to the contrary. 91 According to the Second Circuit, it would be "anomalous for a user whose degree of ownership of a copy is so complete that he may lawfully use it and keep it forever, or if so disposed, throw it in the trash" to be treated as a nonowner. 92 Rather than focus on formal title or self-serving license terms, "courts should inquire into whether the party exercises sufficient incidents of ownership over a copy of the program to be sensibly considered the owner of the copy." 93 Without a consistent rule, or at least clear competing theories, the copy ownership cases are difficult to reconcile. Their seemingly ad hoc outcomes offer consumers little certainty or predictability about their rights in the copies they acquire. Despite the uncertainty, there has been an identifiable shift in judicial thinking on the question of copy ownership over the last century. For most of the twentieth century courts would have rejected out of hand the contention that a printed notice attached to a chattel could prevent a transfer of ownership. But today courts entertain and often accept that very characterization, particularly in cases involving software.

One reason courts may be more tolerant of limitations on consumer ownership in this context is the sense, however vague, that software is somehow different-that the rules that govern ownership of other copies and chattels generally-do not apply to the thoroughly modern stuff that is computer software.

The Ninth Circuit offers the clearest illustration of this sort of software exceptionalism. On June 7, 2010, Judges Canby, Callahan, and Ikuta heard oral arguments in two cases, both of which turned on the question of copy ownership. 94 Both involved the allegedly unlawful resale of copies of protected works in violation of the terms of license agreements that purported to preclude copy ownership by the defendants. And in both cases, the copies at issue were lawfully made, the license agreements imposed similar restrictions, and the works were embodied in the same medium, plastic discs. But the court reached very different conclusions relying on two seemingly inconsistent approaches. One case was about software; the other was not.

UMG v. Augusto involved the resale of promotional CDs given away to music reviewers and other industry insiders. 95 The defendant purchased used CDs from local record stores and resold them on eBay. 96 He argued that as the owner of the discs, his resale was protected by the first sale doctrine. But the record label insisted that Augusto was not the owner of those discs because of a notice printed on the discs declaring that they were the property of the label. 97 Despite the label's insistence, the Ninth Circuit held that title to the discs transferred to their recipients upon delivery and, eventually, to Augusto, entitling him to invoke the first sale doctrine. 98 Vernor v. Autodesk centered on the resale of discs containing not music, but software. 99 Autodesk, like UMG, argued that the notice accompanying its software discs meant that customers who paid thousands of dollars for a copy of its software did not own those plastic discs, but merely "licensed" them. 100 Rather than apply the rule that governed Augusto, however, the court looked to its prior inconsistent caselaw to distill a three-part test that asks: first, "whether the copyright owner specifies that a user is granted a license," second, "whether the copyright owner significantly restricts the user's ability to transfer the software," and third, "whether the copyright owner imposes notable use restrictions [on the intangible copyrighted work]." 101 Since Autodesk's terms contained the necessary language, the court concluded that Autodesk owned the discs in Vernor's possession. 102 The court applied two very different frameworks in these cases. In Augusto, it focused on the practical reality of the transaction and the form of media upon which the work was fixed, regardless of the purported licensing terms. But in Vernor, it was concerned almost exclusively with the text announcing restrictions and reservations in the licensing agreement. What is more, if we apply the Vernor test to the facts of Augusto, the cases are nearly indistinguishable. Each prong of the Vernor test is satisfied: UMG characterized the transaction as a license; it prohibited recipients from transferring the discs to others; and it confined them to "personal" use of the discs. 103 95 Augusto, 628 F.3d at 1177. 96 Id. at 1178. 97 Id. at 1179. That notice provided:

This CD is the property of the record company and is licensed to the intended recipient for personal use only. Acceptance of this CD shall constitute an agreement to comply with the terms of the license. Resale or transfer of possession is not allowed and may be punishable under federal and state laws. This inconsistency strongly suggests that the Ninth Circuit has created two parallel regimes for distinguishing licenses from sales. In software cases, it applies a test that turns on factors entirely within the control of the rightsholder. In nonsoftware cases, it adopts a more consumer-friendly approach that turns on the nature and consequences of the transaction. If these divergent approaches are motivated by software exceptionalism, the court's basic intuition is easy to forgive. Software is indeed different from other sorts of copyrighted works in important ways. The question, however, is whether those differences justify a shift in our thinking about the relationship between intellectual and personal property. Below, we identify three characteristics that may influence courts to treat software as a special case, none of which ultimately distinguish software from other types of copyrighted works from the perspective of copy ownership.

First, software has always been an idiosyncratic component of the copyright system. For decades, there was considerable confusion and disagreement as to whether software should be included in the copyright system at all. Congress explicitly extended protection in 1980. 104 But from the outset, applying that mandate has required courts to make significant accommodations. Perhaps most fundamentally, copyright law excludes any "procedure, process, system, [or] method of operation," a limitation difficult to square with the deeply functional nature of computer code. 105 Code is not written for its aesthetic beauty; it is written to do work. In this sense, the mere inclusion of software within copyrightable subject matter makes it a special case. 106 Second, the functional nature of software means that particular copies are of little intrinsic value. Unlike books, paintings, or even musical compositions, the protected expression of a computer program is not apparent to the owner of a copy. 107 Assuming the owner of a book is familiar with the language in which it is written, or the owner of sheet music can read musical notation, the mere viewing of a copy reveals its expressive content. The underlying work announces itself on the surface of the copy. Not so for software. You can stare at a CD-ROM or floppy disk all day and learn nothing about the program it contains. In part, this reflects the fact that software is typically distributed in object code-computer-readable ones and zerosrather than human-readable source code. 108 Of course, the same is true of movies and music distributed on digital media. Where software differs from enforce its restrictions. music and movies is the access to the work users are granted once those ones and zeros are interpreted by the appropriate machine. If you pop a DVD into your laptop, it displays the movie as plainly as a book displays words on the page. But if you put a copy of AutoCAD into your laptop, you will not see the object code on your screen, much less the source code. Instead, you will see a set of images rendered by your operating system interpreting instructions from the program. 109 Those images are distinct from the underlying protected software, which remains hidden even from a copy owner. And because copies of software cannot be used or enjoyed directly, they are of little use in isolation. Their value is only realized within a computer's operating environment, where running the program requires the creation of additional copies of the work, which may or may not implicate the exclusive rights of the copyright holder. 110 Third, licensing has been a component of software transactions since mass markets for programs emerged. Book publishers, record labels, and others all attempted to impose licensing restrictions on copies of their works in the past. 111 But consumers and courts saw those efforts as anomalous departures from existing transactional forms and established expectations. By introducing licensing during its formative stages, the software industry helped shape those expectations. 112 To many, licensing is part of the very nature of the "thing" that is software. 113 Early efforts to license software reflected the uncertainty surrounding available copyright protection and 109 The relative inaccessibility of the underlying expression is one reason courts have been receptive to intermediate copying in the software context. Sega Enters. Ltd. v. Accolade, Inc., 977 F.2d 1510, 1525 (9th Cir. 1992) (noting that programs "are typically distributed for public use in object code form, embedded in a silicon chip or on a floppy disk," and "[f]or that reason, humans often cannot gain access to the unprotected ideas and functional concepts contained in object code without disassembling that code-i.e., making copies"). 110 See Perzanowski, supra note 48, at 1104-05. 111 See Bobbs-Merrill Co. v. Straus, 210 U.S. 339, 350 (1908) (refusing to enforce book publisher's restrictions on resale price); RCA Mfg. Co. v. Whiteman, 114 F.2d 86, 87-88 (2d Cir. 1940) (refusing to enforce a restriction deeming phonographs "Not Licensed for Radio Broadcast" because "the records themselves could not be clogged with a servitude"); see also ALEX SAYF CUMMINGS, DEMOCRACY OF SOUND 28 (2013) (describing a 1906 congressional hearing on restricting the lending of sheet music and an industry lobbyist enthusiastically agreeing with the notion that "the property itself does not carry the right to use it" (internal citation omitted)); Pamela Samuelson, Legally Speaking: Does Information Really Have to Be Licensed?, COMMUNICATIONS OF THE ACM, Sept. 1998, at 15, 18 (noting that Victrola records carried a license purporting to restrict use of the record to one machine and "to deny authority to retransfer one's copy of the recording"); Cory Doctorow, Thomas Edison's Crappy, Price-Fixing EULA, BOING BOING (Jan. 13, 2009, 10:28 PM), http://boingbo-ing.net/2009/01/13/thomas-edisons-crapp.html (describing printed notice on wax cylinder phonograph purporting to set a minimum retail price of 35 cents). 112 Consumer expectations are malleable things. In large part, the debate over digital exhaustion is as much about shaping consumer expectations as it is about defining legal doctrine. 113 See Robert W. Gomulkiewicz, The License Is the Product: Comments on the Promise of Article 2B for Software and Information Licensing, 13 BERKELEY TECH. L.J. 891, 896 (1998) ("For concerns over confidential trade secret information. 114 A key feature of those early licenses was "the provision that the developer retains title to-that is, licenses and does not sell-the individual copy of the program itself." 115 The software industry understood the power of licensing to undermine exhaustion rules, control secondary markets, and limit competition. And its licenses were in part a strategic effort to do just that. 116 But by that point, the conception of software as a product licensed rather than sold had already established a strong foothold.

Each of these three characteristics of software-its outlier status in copyright law, the limited value of any particular copy, and the early introduction of licensing-might help to explain why courts in software cases tend to be more solicitous of aggressive licensing regimes. But they offer little justification for treating software differently than other works for purposes of copy ownership. 117 First, the imperfect fit of software within the copyright system suggests we should be particularly careful about strategies that would enable copyright holders to leverage their statutory rights in a way that harms consumers. Second, Congress squarely addressed the limited value of the copy when it privileged the creation of necessary step copies and adaptations in order to ensure that copies retained meaningful value in § 117, a provision premised on the continued recognition of consumer ownership. Third, the fact that software licenses are commonplace tells us very little about the wisdom of enforcing them in a way that undermines consumer ownership.

But these three characteristics are not the whole story. As the next Section explains, the tendency to embrace "licensing" particular copies of a work has its roots in a deeper uncertainty about the nature of property rights.

In the century separating Bobbs-Merrill from the contemporary debate over copy ownership, our understanding of property has undergone a dramatic shift, from the certainty of Blackstonian absolutism to the rather more equivocal bundle of rights. 118 Despite the merits of a more nuanced relational theory of property, that shift enabled a blurring of the distinction most software products, the license is the product."); Michael J. 117 Nor is there any obvious legal basis for such a distinction. Id. at 300 ("[T]here is no evidence in the statute or in the logic and history of copyright law that supports permitting owners of copyrights in computer programs to have the power to 'license' copies in ways that publishers of books and phonorecords cannot."). 118 See Arthur Linton Corbin, Taxation of Seats on the Stock Exchange, 31 YALE L.J. 429, 429 (1922) ("Our concept of property has shifted . . . [and] ceased to describe any res, or object of sense, at all, and has become merely a bundle of legal relations."). between property and contract, which in turn bolstered efforts to redefine copy ownership through licensing terms.

Early twentieth-century thinking was dominated by a tradition rooted in natural rights and encapsulated in Blackstone's oft-quoted-if oft-misunderstood-description of property as "that sole and despotic dominion which one man claims and exercises over the external things of the world, in total exclusion of the right of any other individual in the universe."foot_6 Within that framework, property rights were rights in things held by a single owner with a nearly absolute right to exclude. 120 Beginning with Hohfeld's taxonomy of jural relations, however, our picture of property became more complex. By breaking the notion of an integrated property right into constituent rights, privileges, and powers, Hohfeld took the first steps towards deconstructing the property monolith. Moreover, he advanced the idea that property rights were not rights in things, but rights against people. 121 Under this view, the central difference between rights in personam (paucital rights) and rights in rem (multital rights) is the size of the class of individuals against whom those rights apply. Paucital rights, like those arising under contract, apply to specified individuals, while multital rights, like those arising under a property regime, apply to a much larger, less specific class. 122 A property right then does not describe the relationship between an owner and a thing, but the aggregation of relationships between individuals. 123 Decades later, Honoré's eleven incidents of ownership further disaggregated our notion of property into discrete component parts. 124 And according to Honoré, no particular incident or combination of incidents is necessary for ownership; any one or more may be absent. 125 Taken together, these insights form the core of the "bundle of rights" conception of property that largely defined property orthodoxy in recent decades. 126 As an alternative to the rigid formalism of earlier eras, the bundle of rights approach has much to commend it. But its critics charge that it results in a notion of property that is infinitely malleable, 127 lacking any core or essence, 128 conceptually unmoored, 129 fragmented, 130 and ultimately meaningless. 131 If property rights are nothing more than a collection of relationships between individuals, infinitely divisible and alienable, and if we lack any articulable criteria for which discrete rights render one an owner or even what sorts of things are proper subjects of property rights, property lacks any distinctive character that sets it apart from other modes of allocating resources. According to this view, the "shadowy" bundle of rights model marks a transition "from a world in which property was a central idea . . . to one in which it is no longer a coherent or crucial category in our conceptual scheme. The concept of property and the institution of property have disintegrated." 132 As a result, "property ceases to be an important category in legal and political theory." 133 Capitalizing on this perceived breakdown, some economists offer a reassuringly simple take on the role of property. From this perspective, property rights do little more than establish default entitlements to be allocated through private bargaining. 134 same flexibility and granularity we see in the realm of privately negotiated agreements. 135 And if so, property rights become indistinguishable from contractual ones. 136 Once property is viewed from this angle, the use of standardized contractual terms to "license" a tangible copy transferred to a consumer looks less like an anomaly inconsistent with the hundreds of years of common law rejecting equitable servitudes on chattels and more like an example of property formalism yielding to market efficiency.

Copyright law, by congressional design, lends itself to the bundle of rights characterization, and perhaps by extension, to the substitution of property rights with contractual ones. 137 The rights of a copyright holder are quite explicitly an enumerated bundle. The Copyright Act defines that bundle as comprising the exclusive rights to: reproduce, create derivative works, distribute, publicly perform, and publicly display the work. 138 And those exclusive rights are infinitely divisible. As the Act makes clear, "[a]ny of the exclusive rights comprised in a copyright, including any subdivision of any of the rights," can be "transferred" and "owned separately." 139 As a result, copyright law encourages the creation of narrow, idiosyncratic subdivisions of rights constrained by time, geography, medium, or any other condition devised by the initial rightsholder. 140 Given the malleability of this bundle, copyright's notion of ownership of the work is correspondingly fuzzy. If assignment or exclusive license of narrow subdivisions of any exclusive right is a "transfer of ownership" for copyright purposes, a single work has a potentially infinite class of owners. And if copyright holders are free to disassemble and parcel their rights into any configuration they choose, granting some rights to use the work while withholding others, it is easier to understand how courts might buy into the myth of licensing particular copies of the work. Why should rightsholders be denied the flexibility to distribute copies to the public while retaining ownership of them if the function of property rights is merely to set the stage for private bargaining? 141 Ultimately this shift in the property paradigm toward contractualism, 142 combined with the tendency of courts to see software as somehow special, helps explain the erosion of consumer ownership of copies of software. But as explained below, the licensing regime constructed for software is unlikely to remain quarantined.

Courts have endorsed attempts to characterize purchases as licenses in the software context, but have proven far less open to that characterization when it comes to more traditional copyrighted works like movies, music, and books. 143 This fact might suggest that the effects of the software licensing paradigm will be contained. But there are two reasons to think this licensing model will have ripple effects across the copyright economy. First, despite the caselaw, assertions that consumers do not own the copies they buy are not limited to software transactions. Second, the gap between software and more traditional works is narrowing.

The copyright marketplace is rife with examples of rightsholders and their intermediaries insisting that consumers do not actually own the copies they buy. 144 Amazon's Kindle Store is among the largest sellers of e-books in the United States. According to Amazon's Terms of Use, consumers who discover a book they would like to read and hit the "Buy now with 1-click®" button do not own the copy they download. 145 As those terms explain, "[u]nless specifically indicated otherwise, you may not sell, rent, lease, [or] distribute . . . any rights to the Kindle Content." 146 This sort of restraint on alienation is clearly inconsistent with copyright law's understanding of copy "[t]he concept of license . . . belongs fundamentally to property, not contract" but "[n]evertheless, the contract theory of license persists"). 142 The breakdown of traditional contract law, from enforcing bargained-for exchanges to enforcing adhesive boilerplate terms, bears some responsibility here as well. ownership and likely at odds with what consumers reasonably understand the word "buy" to mean. 147 Apple's iTunes Store, the largest music retailer in the world, is somewhat more conflicted in how it characterizes transactions with consumers. 148 After describing those transactions as "purchases" and noting that "[a]ll sales . . . are final," Apple insists that consumers agree not to "rent, lease, loan, sell, [or] distribute" their purchases. 149 Amazon's competing MP3 store offers similar terms. Although Amazon's store "allow[s] you to purchase . . . digital versions of audio recordings," your payment merely "grant[s] you a nonexclusive, non-transferable right to use [the] . . . Music Content . . . only for your personal, non-commercial purposes." 150 However, "you may not redistribute, . . . sell, . . . rent, share, lend, . . . or otherwise transfer or use Purchased Music." 151 These efforts are not confined to digital downloads. Beachbody LLC, the makers of the popular P90X home workout routine, insists that its customers do not own the DVDs they purchase from the company's website. 152 According to the Beachbody Terms of Use, "[y]ou may not, without the express written permission of Beachbody or the respective copyright owner . . . sell [or] resell . . . services or products obtained through [its] Sites." 153 Beachbody has aggressively targeted individual consumers who resold legitimate copies of its DVDs on eBay, threatening litigation and 147 Aside from the question of copy ownership for exhaustion purposes, to the extent consumers understand the terms "buy" and "purchase" to indicate the ability to alienate their purchases, and that understanding is material to their decisionmaking, digital retailers could face liability for false advertising or deceptive trade practices. See 15 U.S.C. § 1125(a)(1)(B) (2012) (prohibiting the use of "any word, term, name, symbol, or device, or any combination thereof, or any false designation of origin, false or misleading description of fact, or false or misleading representation of fact, which . . . in commercial advertising or promotion, misrepresents the nature, characteristics, qualities, or geographic origin of his or her or another person's goods, services, or commercial activities"); id. § 45(a) (prohibiting "unfair or deceptive acts or practices in or affecting commerce"). Consumers would be particularly surprised to find that Amazon could disable access to the books they purchased. Joel Johnson, You Don't Own Your Kindle Books, Amazon Reminds Customer, NBC NEWS (Oct. 24, 2012, 10:43 AM), http://www.nbcnews.com/technology/you-dont-own-your-kindle-books-amazon-reminds-customer-1C6626211. 148 demanding exorbitant compensation. 154 It is easy to understand why Beachbody would want to prevent customers from reselling their workout videos after their New Year's resolve runs out. As reasonable as three easy payments of $39.95 may be, 155 used DVDs on the secondary market would introduce unwanted downward price pressure. What is harder to see is how this restraint on alienation can be squared with the principle of exhaustion or consumer property interests more generally. 156 These examples illustrate that, despite the caselaw, the practices of rightsholders and the experiences of consumers reflect a trend to deny copy ownership and restrict resale that extends beyond the software market. In part, this trend reflects the shrinking gap between software and other classes of copyrighted works. In its early days, software was a thing unto itself in the copyright landscape. But as traditional forms of expression-books, music, visual art-become more interactive and leverage the new avenues for creativity and the market penetration technology offers, they are becoming indistinguishable from software.

The line between programs and data has always been a largely artificial one. 157 But today that distinction is increasingly blurred. 158 156 Whether Beachbody would prevail if it had the nerve to attempt to enforce its license through litigation is unclear. A court applying Vernor would be confronted with a transaction that Beachbody asserts is a license and that imposes significant restrictions on the use and transfer of its products. But the Ninth Circuit would likely analyze this case under the more pro-consumer Augusto framework since Beachbody markets in traditional expressive content, not software. These competing potential outcomes only underscore the uncertainty that consumers encounter. The Copyright Act's definition of a computer program, "a set of statements or instructions to be used directly or indirectly in a computer in order to bring about a certain result," applies equally well to digitally encoded data. 17 U.S.C. § 101 (2012). 158 Though more apparent today, this blurring has been in the offing for some time. Michael J. Madison, Legal-Ware: Contract and Copyright in the Digital Age, 67 FORDHAM L. REV. 1025, 1041-42 (1998) (noting that all digital information looks a lot like software). 159 Matt Miller, Heavy Rain and the Birth of Interactive Film, GAME INFORMER (Feb. 10, 2010, 11:00AM), http://www.gameinformer.com/blogs/editors/b/gimiller_blog/ archive/2010/02/10/heavy-rain-and-the-birth-of-interactive-film.aspx (describing the responsiveness to user input impossible with the printed page. 160 And artists like Jay-Z, 161 Lady Gaga, 162 and Bjork 163 have all released new music embedded in smartphone applications. In short, traditional expressive content and software are merging.

That blurring renders efforts to maintain the current bifurcated approach to licensing untenable. It depends on an identifiable line separating software from everything else copyright protects. If current legal and technological trends continue, the notion of personal property in copies will recede as more and more of copyright's subject matter evolves to resemble software. 164 As described in the next Part, that problem is compounded by the shifting nature of the copy itself.

The copy, needless to say, has been an essential concept in the law of copyright for centuries, 165 and part of copyright law in the United States since its inception. 166 The term's use is somewhat counterintuitive since "copy" refers not only to infringing and noninfringing reproductions of a video game Heavy Rain as "interactive storytelling" that is "more film than it is game"); Greg Tito, Heavy Rain Is Not a Game, THE ESCAPIST (Dec. 18, 2009, 12:37 PM), http://www .escapistmagazine.com/news/view/96957-Heavy-Rain-Is-Not-a-Game (describing Heavy Rain as "not a game" but an "interactive movie"). 160 See Avi Itzkovitch, Interactive eBook Apps: The Reinvention of Reading and Interactivity, UX MAG. (Apr. 12, 2012), http://uxmag.com/articles/interactive-ebook-apps-the-reinven-tion-of-reading-and-interactivity. 161 Evan Minsker, Jay-Z Announces New LP Magna Carta Holy Grail, PITCHFORK (June 16, 2013, 9:39 PM), http://pitchfork.com/news/51187-jay-z-announces-new-lp-magna-carta-holy-grail. 162 Alice Vincent, Lady Gaga Announces ARTPOP, Her New Album App Release, TELEGRAPH (July 12, 2013, 11:01 AM), http://www.telegraph.co.uk/culture/music/music-news/10175 597/Lady-Gaga-announces-ARTPOP-her-new-album-app-release.html. 163 Bjork, Damon Albarn, Snoop Dogg for Manchester International Festival, NME (Mar. 17, 2011), http://www.nme.com/news/bjork/55508. 164 The idea that copyright law's treatment of consumer ownership in the context of its core subject matter-expressive media content like book, music, and films-should owe its origins to an approach that gained its first foothold in cases involving computer software is particularly perverse, given the serious misgivings many voiced about protecting software through copyright law at the outset. work,foot_7 but also to the initial fixation of the work. 168 So Mary Shelley's original handwritten manuscript of Frankenstein is as much of a copy as a Xeroxed paperback. 169 Central to contemporary copyright thinking is the notion that copies and their statutory companions, phonorecords,foot_10 are material objects distinct from the intangible works that the copyright grant is meant to encourage.

Much like the once clear but increasingly opaque notion of ownership, the copy is now a concept plagued by uncertainty. Changes in storage and distribution technologies alongside shifting media consumption patterns have profoundly altered the way in which we interact with copyrighted works. The tangible copy, once the primary means of distribution, has been displaced by cloud storage, streaming, and "software as a service." Copies were once finite, stable, and valuable. But the unitary copy-the hardcover, the LP, the film reel-has been largely displaced. Today's marketplace is characterized by ubiquitous, temporary instantiations of works that have diminished independent value. Copyright law has struggled to assimilate these developments. This Part first describes those changes and then turns to the efforts of the courts to make sense of the shifting role of the copy and the implications for the distinction between the copy and the work.

Dramatic improvements in remote computational capacity and storage, along with increasingly fast and pervasive data connectivity, allow consumers to acquire, store, and access their media through the much-ballyhooed cloud, that assemblage of remote servers housing our contacts, family photos, and media collections. Consumers can choose between cloud storage services from Amazon, Apple, Dropbox, Google, and Microsoft, among others, at nearly no cost. But the cloud has moved beyond mere storage. Amazon's MP3 store enables consumers to buy, save, and play their purchases directly from its Cloud Player without ever downloading a permanent file to their laptop or mobile device, let alone handling a plastic disc. 171 Software-as-aservice offerings, like Adobe Creative Cloud-the new home for Photoshop and other editing and layout programs-or the Google Apps suite of office productivity tools, prove that the functionality that once required the distribution of copies can be achieved by remotely accessing data and processing power. 172 These technological developments have been accompanied and partly driven by changing consumer preferences. As evidenced by the fact that close to half of internet traffic is attributable to Netflix, many consumers would rather access a library of streaming video titles than purchase or even rent tangible copies. 173 The popularity of Spotify, Pandora, and other streaming music services suggest the same might be true for music. 174 This transition from distribution to performance is evident in the video game market as well. OnLive launched a cloud-based gaming platform in 2010. 175 And Sony has recently announced Playstation Now, a cross-platform service that will allow subscribers to rent and stream games to consoles and mobile devices. 176 Taken together, these changes offer great promise to consumers. But they also threaten to destabilize our understanding of the copy and its place in the copyright system.

Four cases-one from the early twentieth century and three of more recent vintage-illustrate the ongoing dialogue between technological innovation and the legal understanding of the copy.

In the late 1800s, the advent of perforated player piano rolls prompted both the courts and Congress to reconsider what constituted a copy. In White-Smith v. Apollo, the Supreme Court was asked to decide whether perforated rolls, which in conjunction with a player piano performed copyrighted musical compositions, were infringing copies. 177 The Court, following a number of earlier opinions addressing both piano rolls and phonographs, held that the perforated rolls were not "a written or printed record of . . . intelligible notation." 179 If "even those skilled in the making of these rolls are unable to read them as musical compositions," they could not be copies. 180 To hold otherwise, according to the Court, would be to force "a strained and artificial meaning" on the copy. 181 Congress, revising the Copyright Act just a year later, had an opportunity to weigh in on White-Smith's vision of the copy. 182 However, it took a decidedly unclear position, ultimately maintaining the general White-Smith interpretation but extending protection to piano rolls. 183 White-Smith demonstrates that judicial and legislative struggles to understand the idea of the copy are a function of the day's technology. Two more recent cases illustrate how that same basic challenge confronts courts faced with contemporary technological advances.

If the question in White-Smith was whether embodiments of a work were too inscrutable to count as copies, the question in Cartoon Network LP v. CSC Holdings, Inc. was whether they were too temporary. 184 Defendant Cablevision was sued for offering consumers access to a Remote Storage Digital Video Recorder system (RS-DVR) that functioned similarly to a set-top home DVR but stored programs centrally at a Cablevision data center. 185 177 White-Smith Music Publ'g Co. v. Apollo Co., 209 U.S. 1, 8-9 (1908). 178 See Stern v. Rosey, 17 App. D.C. 562, 564-65 (1901) ("We can not regard the reproduction, through the agency of the phonograph, of the sounds of musical instruments playing the music composed and published by the [complainants], as the copy or publication of the same within the meaning of the act. . . . Conveying no meaning, then, to the eye of even an expert musician and wholly incapable of use save in and as a part of a machine specially adapted to make them give up the records which they contain, these prepared wax [ed] cylinders can neither substitute the copyrighted sheets of music nor serve any purpose which is within their scope."); see also Kennedy v. McTammany, 33 F. 584, (C.C.D. Mass. 1888) ("I cannot convince myself that these perforated [sheets] of paper are copies of sheet music[ ] within the meaning of the copyright law. They are not made to be addressed to the eye as sheet music, but they form [a] part of a machine."). 179 White-Smith, 209 U.S. at 17 (internal quotation marks omitted). 180 Id. at 18. 181 Id. at 17.

The RS-DVR system divided the data stream containing Cablevision's television programming into separate streams for each television channel using a device called the Broadband Multimedia Router (BMR). 186 In this process, the BMR briefly loaded the data stream into its buffer memory, for as much as 1.2 seconds. 187 The BMR then relayed the separated data streams to a server containing hard drives on which recorded programs were stored. 188 The server buffered one tenth of a second of video for each channel at any time. 189 If a customer initiated a recording of a program, the server moved data for the selected channel to another buffer, after which data were written to the hard drive for later viewing. 190 Upon playback by a customer, the data passed through a streaming buffer, bringing the total number of buffer instantiations to four, not counting the more durable copies on the hard drive. 191 Cartoon Network alleged that each of these buffers resulted in infringing copies of their television programs. The case turned on whether the buffer data were fixed in a tangible medium and therefore copies of the Plaintiffs' television programs. Unlike earlier courts, the Second Circuit recognized that fixation entails two distinct requirements. First, the work must be sufficiently embodied to be perceived, reproduced, or communicated. 192 And second, that embodiment must persist for more than a transitory duration. 193 Because "[n]o bit of data remains in any buffer for more than a fleeting 1.2 seconds . . . . [and] each bit of data here is rapidly and automatically overwritten as soon as it is processed," the court was satisfied that the "works in this case are embodied in the buffer for only a 'transitory' period" and thus not copies. 194 Although we applaud the Second Circuit's careful statutory and factual analysis, 195 it highlights an important conceptual difficulty for copyright law. If, as the court rightly concluded, buffer data are not copies, what exactly are they? They are not the intangible work, but seem to occupy some interstitial space within the copy/work dichotomy. The uncertain status of embodiments like these suggests a growing difficulty in identifying where copies end and where non-copies begin. And if copyright law cannot tell us what embodiments even count as copies, copy ownership faces a significant challenge.

In Capitol v. ReDigi, we saw yet another challenge to our understanding of the copy-or in this case, the phonorecord. 196 There, ReDigi attempted to create a secondary marketplace for digital media by offering consumers a platform for the resale of pre-owned digital music. 197 If a consumer who bought a copy of "The Sign" by Swedish pop group Ace of Base from the iTunes Music Store in a fit of 1990s nostalgia later regretted that purchase, ReDigi would allow her to recapture some of her investment by transferring her interest in the song to another equally nostalgic buyer.

According to ReDigi, its software first verifies that the music was legitimately purchased. 198 Next, it uploads the seller's file to ReDigi's Cloud Locker, where the file is stored until purchased by another user. 199 The technical design of ReDigi's upload process introduces an important wrinkle to the copyright analysis. ReDigi argues that the upload does not create a new copy, it merely migrates the file from one location to another. 200 As each packet of information that comprises the file is uploaded to the cloud, it is deleted from the user's local drive, virally transporting the file. 201 And "at the end of the process, the digital music file is located in the Cloud Locker and not on the user's computer." 202 Finally, the software deletes any additional copies of the file on the user's computer and devices. 203 In this way, ReDigi's platform is intended to mimic the consequences of an analog sale, a transfer of ownership that terminates the rights of one party and establishes the rights of another.

Capitol Records sued ReDigi, alleging that the process described above is nothing more than an unauthorized reproduction of its protected works. 204 The question for the court, in essence, was whether the copy on Redigi's server should be understood as new and distinct from the one that previously resided on the seller's hard drive, or instead, as ReDigi contended, the identity of that copy persisted as it moved from one storage location to the next.

ReDigi analogized its service to an artist who used emulsions to lift images from greeting cards and apply them to ceramic plaques. 205 But the court rejected this analogy, holding that the migration process created a new material object and thus, a new phonorecord. 206 In doing so, the court drew a sharp distinction between the copyrighted sound recording-the workand the "appropriate segment of the hard disk"-the phonorecord. 207 The court characterized this conclusion as not only demanded by the Copyright Act, but dictated by the limits of technology:

This understanding is, of course, confirmed by the laws of physics. It is simply impossible that the same "material object" can be transferred over the Internet. . . . Because the reproduction right is necessarily implicated when a copyrighted work is embodied in a new material object, and because digital music files must be embodied in a new material object following their transfer over the Internet, the Court determines that the embodiment of a digital music file on a new hard disk is a reproduction within the meaning of the Copyright Act.

This finding holds regardless of whether one or multiple copies of the file exist. . . . Simply put, it is the creation of a new material object and not an additional material object that defines the reproduction right. 208 But far from confirming that the rules imposed by law and those imposed by physics are aligned, the ReDigi opinion underscores the mismatch between our current technological capability and copyright law's preoccupation with the unitary copy. Digital consumers are awash in a sea of copies. Copies flit into and out of existence. They are created, used, discarded, and created yet again. Through memory virtualization, they can even be distributed across many different storage media, often across great distances. 209 Cloud computing frees us from the burden of the unitary copy. Consumers no longer need to lug their digital media collections with them everywhere they go; they simply need a data connection. Even in the precloud environment, the ReDigi court's reasoning overlooks the continual creation of copies necessary for computers to operate. The notion that we can identify the particular copy that a consumer purchased from iTunes is almost hard to square with the reality of networked distribution. Yet the ReDigi court felt a need to pin the tail on the proverbial donkey and identify particular phonorecords within the ReDigi system. Finally, in UsedSoft GmbH v. Oracle, the European Court of Justice approached the question of the identity of the copy very differently. There, Oracle sued UsedSoft for allowing users to purchase second-hand software licenses. 211 Oracle claimed that when these consumers subsequently downloaded the software from its servers, along with patches and updates, they illegally reproduced the code. 212 The court rejected this theory, holding that the doctrine of exhaustion applied not only to the initial download, but to all subsequent updates and patches, even if downloaded by a subsequent user. 213 The court reasoned that the sale was not tied to a particular download or copy, but rather extended to any "functional equivalent" so long as payment was initially made to Oracle. 214 In other words, the court was less concerned with particular copies-that is, particular downloads to particular material objects-than it was with the rights of subsequent transferees to access the work. If exhaustion applied, it did so regardless of the identity of any particular copy. 215 Together, these cases signal a growing uncertainty about the definition of the copy and the legal status of those instantiations of a work that occupies that middle ground between the tangible unitary copy and the intangible work. Historically, copyright law has conceptualized consumer rights as situated in the particular copy they own. As the role of the unitary copy is diminished, the exhaustion principle and the values it serves are at risk. As the next Part explores, disaggregating rights from copies may offer a clearer picture of the values at stake in copyright exhaustion.

Without a stable understanding of the copy or a meaningful notion of copy ownership, copyright law's exhaustion doctrine-as it has been historically understood and applied by courts-faces a crisis. Although some rightsholders would celebrate the end of first sale and other exhaustion rules, our prognosis is not quite so dire. Rather than give up altogether on the consumer rights that flow from copy ownership, this Part identifies the functions copy ownership has served in order to reformulate the exhaustion principle in a way that preserves its crucial role in the copyright system.

There is nothing magical about the tangible copy or inherently transformational about the label "owner." The copyright system has relied on copy ownership as one of many tools for delineating the scope of copyright holders' control over their works and balancing that power against the interests of the public in using and deriving value from those creations. Here we consider why. What is it about the relationship between the rightsholder, the work, and the consumer that justifies extending rights to copy owners that would otherwise fall within the exclusivity of the copyright holder?

We argue that copy ownership serves as a proxy for a cluster of related considerations that together justify distinguishing copy owners from the public at large as a matter of copyright policy. These core functions of copy ownership-the contribution to incentives for authors, the inducement of consumer participation in lawful markets for protected works, and the reduction of information costs-explain why copyright law confers special status on some users of protected works. But they do not dictate how. With a clearer understanding of the reasons we privilege copy owners in place, we detail an approach to copyright exhaustion that legislators and courts can use to more transparently balance the competing interests of consumers and creators.

The creation of economic incentives for creative production is among the core constitutional and instrumental objectives of copyright law in the United States. 216 Because creative works are public goods, copyright theory worries that, absent some legal intervention to limit competition and create artificial scarcity, creators will underinvest in new works. 217 Through legal exclusivity, "copyright supplies the economic incentive to create and disseminate ideas." 218 Although the descriptive accuracy of this incentive story is an open question, 219 copyright law remains deeply committed to the notion that exclusive rights are intended to ensure that authors are compensated for their work. 220 For much of our recent history, the financial and popular success of copyrighted works has been a function of the sale of copies. There are exceptions of course. Box office returns and ticket sales for live performances are significant revenue streams for some creators. And streaming, subscriptions, and rentals offer increasingly important nonsale revenue. But as a general rule, authors who sell more copies make more money. As a result, copy ownership provides a strong indication that compensation sufficient to motivate an author has been paid. The exclusive right of initial distribution gives copyright holders the power to set their desired price for sales to the public. Once that price has been paid, we should assume that the copyright holder has reaped what it considers a fair reward and continued exclusivity can no longer be justified. 221 Perhaps not surprisingly, a number of courts have recognized the centrality of this compensatory justification for exhaustion rules. 222 An exhaustion rule triggered by copy ownership is one that strikes a balance between the rights of authors and the rights of the public, a balance calibrated to preserve creative incentives. Exhaustion applies only to lawfully made copies disposed of at an initial price set by the rightsholder. 223 This ensures that the rightsholder receives a measure of compensation sufficient to justify its investment. In this way, copy ownership functions as a limit on 220 Whether copyright law in practice achieves that goal is debatable. See Julie E. Cohen, Copyright as Property in the Post-Industrial Economy: A Research Agenda, 2011 WIS. L. REV. 141, 143 (noting that "[t]he incentives-for-authors formulation of copyright's purpose is so deeply ingrained in our discourse and our thought processes that it is astonishingly hard to avoid invoking it" but arguing that in practice "the purpose of copyright is to enable the provision of capital and organization so that creative work may be exploited"); Jessica Litman, Real Copyright Reform, 96 IOWA L. REV. 1, 8-9 (2010) (noting that that "[t]he law's congruence with the [incentive] theory, at least in many fields of authorship, is more aspirational than real"). 221 Brilliance Audio, Inc. v. Haights Cross Commc'ns, Inc., 474 F.3d 365, 373-74 (6th Cir. 2007) ("Once a copyright holder has consented to distribution of a copy of that work, this monopoly is no longer needed because the owner has received the desired compensation for that copy."). 222 Platt & Munk Co. v. Republic Graphics, Inc., 315 F.2d 847, 854 (2d Cir. 1963) (noting that "the ultimate question embodied in the 'first sale' doctrine" is "whether or not there has been such a disposition of the article that it may fairly be said that the patentee [or copyright proprietor] has received his reward for the use of the article" (quoting United States v. Masonite Corp., 316 U.S. 265, 278 (1942)) (internal quotation marks omitted)); Parfums Givenchy, Inc. v. C & C Beauty Sales, Inc., 832 F. Supp. 1378, 1389 (C.D. Cal. 1993) ("[T]he distribution right and the first sale doctrine rest on the principle that the copyright owner is entitled to realize no more and no less than the full value of each copy or phonorecord upon its disposition."); Burke & Van Heusen, Inc. v. Arrow Drug, Inc., 233 F. Supp. 881, 884 (E.D. Pa. 1964) ("[T]he ultimate question under the 'first sale' doctrine is whether or not there has been such a disposition of the copyrighted article that it may fairly be said that the copyright proprietor has received his reward for its use."). 223 See Kirtsaeng v. John Wiley & Sons, Inc., 133 S. Ct. 1351, 1354-55 (2013) (describing the relation between the "first sale doctrine" and exhaustion).

do their job, consumers must be willing to pay supracompetitive prices for protected works despite their widespread availability at near-zero marginal cost. The struggle facing the copyright system is convincing consumers that a lawful copy is more desirable than an unlawful one. The most obvious way to do that is through the stick of infringement liability.

But the carrot is equally important. A clear exhaustion principle that entitles consumers to use and alienate copies makes the value proposition presented by copyright compliance more attractive than it otherwise would be. Copy ownership and exhaustion offer consumers real value. They can resell the copy and recoup some of its value. 228 They can also preserve, loan, and modify their copy to suit their needs. 229 When lawfully available copies lack the freedoms consumers expect, they are less desirable. 230 And unlawful copies are necessarily more attractive in comparison. But when consumers can rely on getting something of value for their money-a set of rights that includes unrestricted personal use and the power to alienate their intereststhey may be more likely to opt into the copyright economy. In short, first sale and related rights steer consumers towards lawful markets, those that promise compensation to rightsholders. In doing so, consumer property rights help copyright law serve its dual objectives of incentivizing creative production and encouraging access to the resulting creative output.

On a broader level, exhaustion helps consumers navigate the marketplace for protected works by limiting information costs. Unlike contract law, which imposes no limits on creative permutations of rights and limitations transferred between two parties, exchanges of property rights are limited to an identifiable number of standard forms. 231 This commitment to a closed set of property forms, or the principle of numerus clausus, not only helps dis-tinguish property regimes from private contractual ordering, it is one of property's principle advantages. By favoring sales and discouraging idiosyncratic transfers of rights in copies, exhaustion rules limit information costs.

One justification for limiting property transactions to a small universe of standardized forms is a concern over restraints on alienation. 232 Unpredictable bundles of rights, potentially divided among numerous parties, increase transaction costs in a way that could discourage transfers of property and hinder more valuable uses. As between the parties to the transaction, however, the numerus clausus solution to this potential anticommons problem can frustrate the intentions of the parties. But that worry is only convincing to the extent we believe the terms of the transfer reflect an actual bilateral agreement. In the case of nonnegotiable, standard form agreements attached to mass marketed copies that purport to bind subsequent parties, there is less reason to be deferential to the freedom of contract. 233 Setting aside the potential impact of idiosyncratic configurations of property interests on parties to a transaction, the imposition of information cost externalities on unrelated third parties offers another strong reason to curtail the proliferation of new property arrangements. Even if clear-eyed parties with equal bargaining power agree to a bespoke bundle of rights, that choice imposes costs on other market participants to investigate the details of property transactions with much greater scrutiny. 234 Once on notice that property bundles deviating from the standard forms are accepted, those acquiring property interests, as well as those hoping to avoid infringing them, bear the burden of due diligence to uncover their potentially unique characteristics. 235 This concern is not new. 236 But it is a particularly pressing one today in markets for copyrighted works given the increasing complexity of license 232 See Michael A. Heller, The Boundaries of Private Property, 108 YALE L.J. 1163, 1199 (1999) (describing the century-long concern of alienation in regard to property); Carol M. Rose, What Government Can Do for Property (and Vice Versa), in THE FUNDAMENTAL INTERRELA-TIONSHIPS BETWEEN GOVERNMENT AND PROPERTY 209, 214-15 (Nicholas Mercuro & Warren J. Samuels eds., 1999). 233 NANCY S. KIM, WRAP CONTRACTS: FOUNDATIONS AND RAMIFICATIONS (2013); RADIN, supra note 142. 234 Merrill & Smith, supra note 231, at 8 ("When property rights are created, third parties must expend time and resources to determine the attributes of these rights, both to avoid violating them and to acquire them from present holders. The existence of unusual property rights increases the cost of processing information about all property rights. Those creating or transferring idiosyncratic property rights cannot always be expected to take these increases in measurement costs fully into account, making them a true externality. Standardization of property rights reduces these measurement costs."). 235 Merrill & Smith, supra note 135, at 777 ("[T]he adoption of novel forms of property has implications not only for the immediate parties to the transaction but also for third parties, who must incur additional costs of gathering information in order to avoid violating novel property rights or to decide whether to seek to acquire these rights."). 236 Keppell v. Bailey, 39 Eng. Rep. 1042, 1049 (Ch. 1834) ("[G]reat detriment would arise and much confusion of rights if parties were allowed to invent new modes of holding and enjoying real property, and to impress upon their lands and tenements a peculiar terms that purport to govern those transactions and the escalating information costs they impose. 237 The current iTunes terms are over 15,000 words, translating into sixteen pages of fine print. 238 If the Chief Justice of the Supreme Court cannot expend the effort to wade through those terms, we should not expect the average consumer to take on those costs either. 239 The numerus clausus principle avoids these problems by refusing to recognize certain transactional forms. 240 Assured that nonstandard "fancies" are not lurking in the marketplace, parties are relieved of the burden of investigating the particulars of each transaction or interest they encounter. 241 Instead, they need only gather enough information to identify a property interest as fitting within one of the few established forms. Traditionally, those accepted forms have been limited with respect to real property, and constrained even more so in when it comes to personal property. 243 But as the infinite divisibility of copyright interests illustrates, the numerus clausus principle is largely absent from intellectual property law. Exhaustion, perhaps because it mediates the border between intellectual and personal property, is one notable exception. 245 supreme-court-chief-justice-admits-he-doesn-t-read-online-eulasor-other-fine-print.shtml (noting that Chief Justice John Roberts doesn't read website terms of use but that he worries that "providing too much information defeats the purpose of disclosure, since no one reads it"). 240 See Merrill & Smith, supra note 231, at 13-14 (noting, for example, that courts have refused to recognize attempts "to create a kind of hybrid between a life estate and a fee simple" and instead see their task as "squeez[ing] the interest into one of the established categories"). 241 Id. at 45. Of course, information costs might be lowest in a world in which all consumers were fully aware that their purchases gave them no rights whatsoever. Aside from running headlong into the problem of consumer incentives discussed above, that approach would raise pure consumer protection concerns. Whether treated as unconscionable or void as against public policy grounds, we would expect courts to refuse to enforce fundamentally one-sided adhesive contracts. While we believe a strong pure consumer protection case can be made, our focus is limited to the ways in which exhaustion functions as a component of the copyright system. 242 copyright's expression of the numerus clausus principle. 246 Exhaustion is an effort to effectively limit the permissible forms of transfers of copies and prevent the proliferation of idiosyncratic interests and the negative consequences that flow from them.

The statutory basis for this reading of the Copyright Act begins with the right of distribution itself, which enumerates the ways in which copies can be distributed to the public. It provides that copyright holders have the exclusive right "to distribute copies or phonorecords of the copyrighted work to the public by sale or other transfer of ownership, or by rental, lease, or lending." 247 This language suggests a bifurcated universe of copy transfer. First, we have transfers of ownership, which entail a perpetual right to possess and use the copy: sales or gifts. Second, we have transfers of limited duration: rental, lease, or lending. If we take exhaustion and numerus clausus seriously, transactions must fall within one of these two categories. Certainly, many licenses are easy enough to characterize as rentals, leases, or lendings. A consumer is not the owner of a book she borrows from the library. Nor does a Netflix subscriber own the physical or digital copies of movies she watches. Copyright holders, even under our reading of the distribution right and the exhaustion doctrine, retain considerable flexibility to devise subscription, streaming, and other business models not premised on the sale of copies. But distribution by license is not among the standard forms of transfers of copies recognized by copyright law. 248 In this sense, the licensed copy is simply not a thing.

Taken together, these three functions of copy ownership-safeguarding authorial incentives, encouraging consumer participation in copyright markets, and reducing information costs-explain why copyright law privileges copy owners over consumers at large. Next, we consider how to preserve those basic functions within a marketplace that deemphasizes the copy and a legal system that has diluted the notion of consumer ownership.

Although exhaustion has been tied to the copy historically, none of the functions detailed above are dependent on any particular medium of expression or method of distribution. These functions, like any property right, help define and structure the relationships between parties. All three are intended to construct a functioning marketplace in which creators and consumers are encouraged to participate. But the success of that market does 246 Id. at 252 ("The first sale doctrine is a form of the numerus clausus principle because it effectively acts as a prohibition on nonpossessory property interests in particular copies of copyrighted works and patented inventions."). 247 17 U.S.C. § 106(3) (2012) (emphasis added). 248 Carver, supra note 44, at 1948 ("[T]here is simply no such thing as distribution of a copy by means of a license. Instead, every transaction falls into one of two categories: those that provide perpetual possession are sales or other transfers of ownership (e.g., gifts) and those that provide a limited period of possession are either a rental/lease or a lending, depending on whether consideration is required."). not depend on the transfer of unitary copies. Instead, it depends on the exchange of rights between consumers and creators and an appropriate legal mechanism for ensuring the right balance between them.

The copy, because of the technological limitations that prevailed for most of copyright's history, was a helpful tool for calibrating those incentives. Ownership of a copy served as a sort of talisman that reassured courts that a defendant's behavior was not a threat to the system. The copy served as judicially administrable shorthand for a more complicated calculus. But the copy no longer serves that role effectively and may soon fail to exist as part of the modern digital environment. Courts need a new way to think through conflicts between the rights of consumers and the rights of creators. This Part begins by arguing against the notion that the copy is an indispensable concept for a functioning exhaustion doctrine. It then outlines an approach to exhaustion that, although less intuitive, offers greater transparency and adaptability.

Within the exhaustion framework, the copy is a token. 249 It represents a set of rights that entitle copy owners to make otherwise prohibited uses of a work, not because the copy itself has any special inherent virtue, but because it signifies that those uses satisfy a broader set of policy considerations. We can take away the copy, and we can even dispense with the preoccupation with ownership, so long as we have some other means of verifying that consumers stand in this privileged relation to the work.

The distinction between rights to the copy and rights in the work, and the related conceptual division between personal and intellectual property, have been useful constructs in copyright law. But both have served as proxies for a more complicated calculus weighing the respective rights of creators and consumers. These heuristics helped courts and policymakers balance those competing interests without undertaking an individualized analysis of the impact on the incentives and costs associated with each contested use. But at the same time, those heuristics have obscured the policy considerations that underlie the contest between intellectual and personal property rights. 250 And because of shifts in technology and the marketplace, those proxies are an increasingly unreliable measure of the interests they once incorporated. Although the U.S. Copyright Office continues to see it as a 249 There are, of course, many systems that manage rights in intangibles through tangible proxies. For example, bearer bonds are a classic example of a physical token that merely signals ownership without actually constituting the object of value itself. "defining element" of exhaustion, the copy looks increasingly like a "mere relic." 251 Thinking about the copy as a proxy reveals how little ultimately separates personal and intellectual property rights. At bottom, the difference between personal and intellectual property is simply who prevails. When the interests of consumers are given more weight, we speak in terms of personal property. When the interests of copyright holders carry the day, we invoke intellectual property. But the rights themselves are not qualitatively different. The rights to distribute, reproduce, and display, for example, are equally at stake in both personal and intellectual property. Nor does insistence that personal property concerns copies while intellectual property protects the work meaningfully distinguish between these two categories. Tying the consumer's right to her particular copy has been a convenient and easily understood way of articulating the limits of the consumer's rights to exploit the underlying work. 252 But it does not change the fact that in the context of exhaustion, intellectual, and personal property regulate the same sets of behaviors and relationships. 253 Calling the statutory interests the Copyright Act establishes for creators and consumers "property" ultimately states a conclusion about the degree to which the law will vindicate them. 254 But that label alone does not help us resolve conflicts between those rights. 255 Proponents of broader and stronger intellectual property protection frequently leverage the rhetorical force of the property label. 256 By recognizing that both creators and con-251 U.S. COPYRIGHT OFFICE, DMCA SECTION 104 REPORT 86 (2001). 252 The copy did much the same for consumers. See Madison, supra note 158, at 1042 (explaining that physical copies give us cues about what we can do with them while digital works introduce ambiguity). 253 When consumers make uses that fall outside of the scope of § 106 rights altogether-when they use a book as a doorstop, for example-we can readily distinguish intellectual property rights from personal property rights since they address different classes of behavior. 254 See Arnold S. Weinrib, Information and Property, 38 U. TORONTO L.J. 117, 120 (1988) ("It also makes plain the conclusory nature of the term 'property': it is a legal characterization, a statement that the court has chosen to assign a particular form of protection to the interest in question." (footnote omitted)); see also Johnson v. M'Intosh, 21 U.S. (8 Wheat.) 543, 572 (1823) (stating that property claims "must be admitted to depend entirely on the law of the nation in which they lie"); JEREMY BENTHAM, THE THEORY OF LEGISLATION 113 (C. K. Ogden ed., 1950) ("Property and law are born together, and die together. Before laws were made there was no property; take away laws, and property ceases."). 255 See Carol M. Rose, Possession as the Origin of Property, 52 U. CHI. L. REV. 73, 85 (1985) ("[Property law] puts an imprimatur on a particular symbolic system and on the audience that uses this system. Audiences that do not understand or accept the symbols are out of luck. . . . sumers can lay equal claim to the property mantle, policymakers might be better equipped to thoughtfully consider the merits of proposals to expand intellectual property. 257 In effect, exhaustion rules allocate usage rights to a defined group of consumers to make particular uses of a work without the permission of the copyright holder. But that allocation is constrained in important respects. Gifts, lending, and resale are limited to the number of copies acquired. Public displays are limited to the location where the copy is housed. Archival copies cannot be retained after the transfer of the originally acquired copy. 260 In each case, these rights and their limitations are justified in reference to a set of underlying policy considerations rooted in costs and incentives.

Historically, copy ownership has served a dual role in this system. It has identified the class of consumers granted these usage rights. And it has helped to define and reinforce the limitations of those rights. For the reasons we have documented here, copy ownership can no longer effectively serve those functions. Copyright law needs a new way to identify this class of consumers. And it needs to formulate new limitations on consumer usage rights that are sensitive to the three policies we have described.

In our earlier work, we have argued that given the comparatively broad scope of common law copyright exhaustion, courts have the authority to drive the evolution of the doctrine. 261 But we have also recognized the potential value of legislative intervention. 262 Although we still maintain that courts are empowered to act alone, the ideal solution is one in which the legislature endorses a more flexible digital exhaustion rule, but leaves courts to engage in the necessary balancing of competing interests. In much the same way Congress endorsed the common law doctrine of fair use without 257 To revisit our critique in Part II, our objection is that courts in copyright disputes have too readily accepted the idea that bilateral agreements can substitute for and effectively displace property rules. Instead, they should have closely interrogated the values that informed the statutory allocation of rights between creators and consumers. That reassessment may well change the balance between the author and the audience, but it would not allow copyright holders to unilaterally determine what rights, if any, consumers acquire in their purchases. 258 constraining its future development, 263 Congress could legislate a flexible exhaustion framework that identifies important factors to consider, but trusts courts to do the heavy lifting.

Although criticized as unpredictable and inconsistent, fair use caseswhen clustered together on the basis of common fact patterns-are reasonably predictable. 264 Equally importantly, the current fair use framework has proven remarkably resilient and adaptable over time. Exhaustion could profitably follow this model of judicial decisionmaking, guided by congressional input.

Any framework for exhaustion needs to address two questions. First, which consumers are granted personal property rights? And second, what do those rights entitle them to do?

The first of these questions addresses the ongoing license versus sale debate. So far, judicial handling of this question has been marked by a host of inconsistent approaches. And the legislative approach is characterized by utter silence. The question is whether a particular consumer stands in relationship to the work sufficient to justify granting rights not available to the public at large. Copy ownership, as currently construed by courts in software cases, is no longer a useful benchmark for identifying the relationship between the consumer and the work that triggers exhaustion. But articulating the right standard is not an easy task. Although there is no simple test for identifying transactions that give rise to this relationship, we think there are three key factors courts should take into account:

1. The duration of consumer possession or access; 2. Whether the payment structure is one-time or ongoing; and 3. The characterization of the transaction communicated to the consumer, including whether it is referred to as a sale or purchase.

No single factor is dispositive. Alone, each of these factors seems incapable of capturing the appropriate mix of incentives and costs. The most obvious dividing line between sales and lesser transfers is perpetual possession. But as a standalone consideration, perpetual possession is problematic. In the digital marketplace, a transaction that is clearly structured and marketed as a rental or subscription might entail perpetual possession of a copy. When a user rents a movie from iTunes, for example, a full copy is delivered to her hard drive that she can retain in perpetuity. What distinguishes that file from a purchased movie title is not that it must be returned, but that a small bit of code renders the file unplayable after a designated period of time. An exhaustion rule premised on perpetual possession alone could foreclose this business model and interfere with creative incentives. Conversely, a transaction that is called a sale and requires a one-time payment for unlimited ongoing access to a work may not require possession in the traditional sense of the word at all. A consumer who "buys" an MP3 from Amazon and stores and accesses that file from Amazon's Cloud Player appears to satisfy the requirements for exhaustion despite never having done more than access a temporary, ephemeral data stream.

Perpetual rights to access or use the work are equally unsatisfying. First, perpetual use might be conditioned on an ongoing payment. If this sort of subscription or renewable lease option triggered exhaustion, it would undermine creative incentives, limit available distribution models, and frustrate consumer choice. Second, perpetual use is an easily evaded trigger. A rightsholder who structured a transaction as a 1,000-year rental could avoid the consequences of exhaustion. 265 Payment structure is another consideration that, although highly relevant, is insufficient to distinguish transactions that confer usage rights to consumers and those that do not. The iTunes rentals described above, while characterized by one-time payments, are not the sort of transactions that should result in exhaustion. Nor is a one-time payment necessary for exhaustion to apply. Consumers who purchased P90X workout videos are no less owners if they opted for three easy payments instead of a single lump sum.

The way a transaction is described in advertisements and other communications to the public also gives courts useful guidance. Of course, some courts have held that rightsholders can avoid exhaustion by calling a transaction a license 266 or by asserting restrictions on use or alienation in a license agreement. 267 But the courts' inquiry should be confined to the terms of a purported license agreement. Because those terms are nonnegotiable, they give copyright holders complete control over their content. And since so few of us actually read end user license terms, they provide little information about how the transaction looks through the consumer's eyes. A test that looks purely at the representations of the rightsholder is open to the same sort of licensing by fiat that plagues software markets. In many instances, license agreements structured to avoid exhaustion accompany public-facing marketing that induces consumers to click a shiny button that says "Buy Now" or "Purchase." To the extent a transaction is held out as a sale or purchase, it seems appropriate to take the copyright holder at its word. At the very least, 265 If the Supreme Court's literal reading of "limited times" in connection to copyright duration is any indication, any finite period would escape the label "perpetual possession" so long as it had a definite end. this approach would encourage more effective notice of the restrictions imposed on consumers. 268 Taken together, we think these factors-though inconclusive in isolation-provide courts an avenue for assessing the consumer's relationship to the work. And if applied consistently, they would offer consumers greater confidence about whether they are buying something they can later alienate, donate, repurpose, recycle, or customize. Rightsholders could reliably structure transactions to avoid exhaustion by conditioning ongoing access on future payments, by clearly communicating time-limited access, or by marketing or advertising the transaction without using concepts such as purchase or buy. This would allow the continued development of rental, streaming, and subscription business models while simultaneously clarifying the choice between those delivery channels and true sales.

After establishing a transaction that triggers exhaustion, courts need to determine whether a consumer's actions fall within the lawful scope of her personal property rights. A workable digital exhaustion doctrine cannot be limited to the distribution right alone. It should also permit limited acts of reproduction, and even the creation of derivative works, to the extent necessary to enable transfers across competing technology platforms. But purchasers of digital content cannot be given free rein. In assessing consumer behavior, courts should consider whether:

1. The purchaser fully parted with possession of or access to the work; 2. The use deprives rightsholders of a fair return; and 3. Whether the purchaser has materially altered the underlying expression.

The first of these factors isolates the central feature of a lawful secondary transaction. Whether permanently or temporarily, resale or lending entails the seller transferring possession or access to the buyer. For digital goods, no less than physical ones, such a transaction cannot result in an increase in the number of individuals who can simultaneously enjoy the work. Rather than try to legislate the myriad ways these transfers of possession and access might work in practice through statutory rules, this factor allows courts to focus on the end result of a transaction rather than its mechanics.

The second factor, much like the fourth fair use factor, considers the impact of the use on the incentives of rightsholders. Sensitive to concerns about creative incentives, courts often justify exhaustion on the grounds that the rightsholder, having completed a sale, has been fairly compensated. 269 268 Importantly, it would also enable consumers to meaningfully choose between streaming and sales models. So long as the unclear licensing terms and misleading advertising define digital "sales," the market is incapable of communicating consumer preferences effectively. 269 See Platt & Munk Co. v. Republic Graphics, Inc., 315 F.2d 847, 854 (2d Cir. 1963) (noting that "the ultimate question embodied in the 'first sale' doctrine is 'whether or not there has been such a disposition of the article that it may fairly be said that the patentee [or copyright proprietor] has received his reward for the use of the article'" (alteration in In the analog world, fair compensation was guaranteed by the rightsholder's ability to set the initial price and by the inherent limits of physical goods. As copyright holders point out, digital distribution alters the economics of secondary markets. 270 The physical world imposes limits not present in the digital one. You can only lend a favorite novel to so many friends before wear and tear, and the occasional spilled drink, take their toll. But digital goods are more durable and easier to reproduce. 271 However, digital goods face their own challenges; rapidly changing hardware and software trends often mean than digital works are often practically inaccessible long before their analog counterparts. 272 Perhaps more importantly, the transaction costs of resale and lending are greatly reduced. The physical book must be delivered and returned-or not, depending on your friends. The digital book can be transferred far more quickly and at much lower cost. Under these conditions, courts need to safeguard against secondary transactions that could undermine the primary markets for protected works.

Imagine an online community for e-book lending. Thousands or millions of users sync their e-book collections with this service, enabling users to search for books they want to read. When you borrow a book from user A, the book is transferred to your device, and no one else including user A can access that book by virtue of user A's right to lend it. So far, this does not sound much different from analog book borrowing. But imagine that the ebook lending platform knows whether the book is being actively read at any given moment. And if not, it can enable another user to "check out" that book from the lending library via a temporary transfer of user A's rights to ("[T]he distribution right and the first sale doctrine rest on the principle that the copyright owner is entitled to realize no more and no less than the full value of each copy or phonorecord upon its disposition."); Burke & Van Heusen, Inc. v. Arrow Drug, Inc., 233 F. Supp. 881, 884 (E.D. Pa. 1964) ("[T]he ultimate question under the 'first sale' doctrine is whether or not there has been such a disposition of the copyrighted article that it may fairly be said that the copyright proprietor has received his reward for its use."). 270 Of course, digital distribution also alters economics in ways that favor rightsholders. Most obviously, costs of production and transportation plummet for digital goods. 271 These concerns have led some to attempt to recreate the imperfections of the analog world for digital works. See Aging File System, U.S. Patent No. 20,110,282,838 (describing IBM's self-degrading file system); see also Evan Hess, Code-ifying Copyright: An Architectural Solution to Digitally Expanding the First Sale Doctrine, 81 FORDHAM L. REV. 1965, 2009 (2013) (suggesting a "regulatory regime that implemented an aging file system" in order to implement a digital first sale doctrine). We believe these backward looking efforts to preserve the downsides of pre-digital information storage are both counterproductive and unnecessary. 272 See Mulligan, supra note 144, at 280 ("Digital works have greater hurdles to preservation than analog or physical copies of works; in addition to preserving a copy and translating the language, 'digital translation' presents an additional problem."); R. Anthony Reese, The First Sale Doctrine in the Era of Digital Networks, 44 B.C. L. REV. 577, 633-39 (2003) (noting the difficulties of preserving readable digital content in an environment in which languages, software, hardware, and file formats undergo rapid change). the book. With a sufficiently large user base, this far more efficient lending system could guarantee that once sold, an e-book would never have a wasted cycle. Someone, somewhere on the planet, would be reading it every second of every day-without any fear of broken bindings or torn pages-all thanks to user A's purchase. Unless rightsholders radically increase the price for an e-book purchase, that sort of lending system would pose a genuine risk to copyright incentives.

This example illustrates why we cannot simply port the exhaustion rules of the analog world over to the digital marketplace. It does not prove, despite the claims of some rightsholders, that exhaustion is fundamentally inconsistent with digital distribution. Our second factor would enable courts to interrogate, assess, and if necessary, prevent, the harmful effect such a platform could have on rightsholder incentives.

Finally, the third factor prevents exhaustion from becoming a back door to transformative uses better considered under fair use. To be effective in enabling robust secondary markets, copyright exhaustion should permit some limited adaptation by consumers. Just as § 117 allows consumers to adapt their computer programs to work with new hardware and software, courts could endorse format and platform shifting to the degree necessary to alienate or use lawfully acquired content. For example, if an Amazon customer wants to purchase a used Apple iBook, either the buyer or the seller might need to create a new copy in the appropriate format. On the other hand, this factor would weigh against permitting remixing or other expressive changes based on the mere purchasing of the underlying raw material. 273 These two multifactor tests allow courts the ability to directly, openly, and transparently assess the impact of allegedly infringing consumer behavior through the lens of the three core functions of the exhaustion doctrine. Although this approach might not initially offer the intuitive appeal of one rooted in the inherent and familiar limits of the tangible copy, we think in time courts, consumers, and rightsholders could embrace our proposal. It will require all three to become more thoughtful and better attuned to the policy motivations underlying the longstanding exhaustion principle, rather than relying on unstated assumptions about the physical world. But we believe ultimately this more adaptable framework is better suited to the rapidly changing copyright marketplace. CONCLUSION What will copyright law look like in the post-copy world? We can imagine a number of possible copyright futures after the demise, or at least the decline, of the traditional copy. In one, the copyright system would be divided into rightsholders and licensees. The rights of alienation, modifica-273 Annotations and commentary would fall somewhere in the middle under this factor. A user who has taken extensive notes on her e-book has arguably created a derivative, but probably has not materially altered the underlying expression. tion, and preservation would be things of the past. In another, rightsholders may permit tightly controlled licensed resale markets in an effort to recreate the sense of ownership consumers have come to expect. Some of the rights enjoyed by consumers would persist, in a diluted form and subject to the permission of copyright holders. In a third possible future, we follow the framework we outlined here. 274 There, copyright exhaustion would be disentangled from particular copies and apply to consumer rights to intangible products. Each of these possible futures has very different consequences for the policy goals of copyright.

If, as a consequence of technological and legal shifts, copy ownership is eliminated and streaming services and tightly controlled licensing arrangements define the market, the experiences of consumers will change profoundly. More importantly, such a model would disrupt the incentives so central to a functioning copyright ecosystem. With regard to authorial incentives, this model has already produced mixed results. Spotify has been subject to several ongoing controversies over the size of its payments to artists and the metrics used to calculate them. 275 A number of artists have complained that these models provide insufficient incentives for musical creativity. 276 Whether such claims are ultimately true, they raise serious questions about authorial incentives.

In part, payments to artists are constrained by the price consumers are willing to pay for temporary access to a shifting library of content. And there is little reason to suspect consumers will be willing to pay more for a system that reduces them to passive recipients of content. They do not own anything. They do not store anything. They cannot resell, loan, or otherwise dispose of anything. Even their access is contingent on the licensing agreements between copyright holders and service providers, which can result in works disappearing and reappearing without any predictability. 277 That unpredictability drives up information costs for both artists exploring new means of distribution and consumers building a media collection.

A variant on this future attempts to replicate some of the benefits of copy ownership within the confines of a licensing regime. Amazon and Apple have both patented technologies that would allow for licensed secondary markets. 278 And both are tentatively experimenting with simulacra of lending and sharing. 279 Within those respective platforms, consumers could alienate their purchases not as a matter of property law, but rather as the result of copyright owner permission expressed through the vendor's terms of service. But these ersatz secondary markets are unsatisfactory on a number of levels. Privacy and anonymity would be threatened. The risk of consumer lock-in would increase since content can only be loaned or transferred within a particular technology platform, potentially threatening incentives for both authors and consumers. From an information cost perspective, allowing the details of alienability to turn on each provider's terms of service imposes considerable costs on consumers and rightsholders. And these commercial offerings provide consumers no reliable assurances about their future availability. 280 The approach we advocate is more in keeping with the policy concerns motivating exhaustion rules. In this future, copyright exhaustion would not be tied to a particular copy but would grant an identifiable subset of consumers' limited rights to use, modify, and alienate their interests in a work, regardless of its embodiment in any particular tangible form. This model offers the consumer incentives and low information costs of the traditional property-based approach. And if properly limited, it would maintain creative incentives by ensuring fair rewards for the sale of well-defined transferable interests in works to the public.

The contours of the copyright law, and by extension the respective rights of creators and consumers, have been shaped by the interactions of unchang-ing policy commitments and the contingencies of existing technology, markets, and practices. As that backdrop changes, copyright law should and must adapt. But it should do so with an eye to its underlying purposes and justifications. The transition to digital distribution of information demands a change to our approach to copyright exhaustion. But that transition should not be exploited as an opportunity to do away with a key component of copyright's balance between the creators and the public. The approach we outlined above helps ensure that the exhaustion principle and the policy commitments at its core outlive the paperback.