Copyright Arbitrage

Copyright Arbitrage

Regulatory arbitrage—defined as the manipulation of regulatory treatment for the purpose of reducing regulatory costs or increasing statutory earnings—is often seen in heavily regulated industries. An increase in the regulatory nature of copyright, coupled with rapid technological advances and evolving consumer preferences, have led to an unprecedented proliferation of regulatory arbitrage in the area of copyright law. This Article offers a new scholarly account of the phenomenon herein referred to as “copyright arbitrage.”

In some cases, copyright arbitrage may work to expose and/or correct for an extant gap or inefficiency in the regulatory regime. In other cases, copyright arbitrage may contravene one or another of copyright’s foundational goals of incentivizing the creation of, and ensuring access to, copyrightable works. In either case, the existence of copyright arbitrage provides strong support for the classification (and clarification) of copyright as a complex regulatory regime in need of a strong regulatory apparatus.

This Article discusses several options available for identifying and curbing problematic copyright arbitrage. First, courts can take a purposive, substantive approach to interpretations of the Copyright Act. Second, Congress can empower a regulatory agency with rulemaking and enforcement authority. Finally, antitrust law can help to curb the anticompetitive effects of copyright arbitrage resulting from legislative capture.


Babyface, Bon Jovi, Cher, Chicago, Katy Perry, and Lady Gaga—this may sound like a Grammy lineup, but it actually constitutes only a fraction of the 186 recording artists who signed a June 2016 petition asking Congress to fix the law that they claim allows YouTube to shirk proper payment for the music it uses.[1] Meanwhile, a slew of artists ranging from Frankie Valli & the Four Seasons to Guns N’ Roses, have filed “staggering”[2] lawsuits—explicitly rejecting a settlement offer of $43.45 million—against Spotify alleging that it failed to pay statutory royalties under Section 115 of the Copyright Act of 1976 (Copyright Act).[3] In both scenarios, the platforms in question are in technical compliance with the letter of the law, while simultaneously reducing, or altogether denying, royalties paid to copyright owners.

At fault for this convoluted state of affairs is regulatory arbitrage, or the reduction of regulatory costs via manipulation of regulatory treatment. All regulatory regimes are vulnerable to manipulation; it arises in diverse areas of the law, ranging from tax[4] to immigration[5] to bankruptcy[6] and even criminal law.[7] As a specialized type of regulatory manipulation, regulatory arbitrage has been described as “exploit[ing] the gap between the economic substance of a transaction and its legal or regulatory treatment, taking advantage of the legal system’s intrinsically limited ability to attach formal labels that track the economics of the transactions with sufficient precision,”[8] and also as a means to “reduce costs or capture profit opportunities created by differential regulations or laws.”[9] Regulatory arbitrage allows regulated entities to “invest in close economic substitutes but with lower regulatory costs.”[10]

Regulatory arbitrage is conventionally viewed as the inevitable result of inherent ambiguity and incompleteness of legal rules. Sometimes referred to as “avoision,”[11] such regulatory manipulation allows for “the avoidance of laws in ways that evade the law’s intent or purpose but do not actually constitute unlawful behavior.”[12] Most commonly seen in finance and tax, regulatory arbitrage has also transpired in copyright since at least the 1920s. The jukebox was arguably a transparent attempt to exploit an exception to the public performance right, which intended to exempt operators of penny arcades (but that used the language “coin-operated machines”).[13] Recently, technological advances—especially those relating to duplication and distribution—and increasing consumer demand for digital content have sparked a rapid proliferation of the phenomenon. This Article offers a new scholarly account of copyright arbitrage.

The scant literature discussing regulatory manipulation in copyright to date has focused primarily on technological avoision, or attempts to code or design around regulation.[14] In his work on code as a technological mechanism of avoision, Tim Wu describes the evolution of peer-to-peer (P2P) network design as a reaction to Napster’s legal demise.[15] Similarly, Dan Burk portrays the development of mutagenic crops—i.e., crops whose genome has been altered by chemical or radiation treatments—as a means of emulating the same desirable characteristics as GMO crops (such as pest resistance), but without falling under arduous GMO regulation.[16] Where technological avoision aims to avoid illegality (as in the case of P2P networks), or to dodge prohibition (as in the case of mutagenic crops), regulatory arbitrage straightforwardly aims to lower an entity’s cost of doing business. It is the latter that constitutes copyright arbitrage.

Copyright is a notoriously complicated statutory regime that regulates industries, such as music, film, and publishing, whose business models have been dramatically impacted by technology. This makes the field ripe for regulatory arbitrage, which falls into two frequently overlapping categories: (1) ontological—i.e., the manipulation of a statutory definition so as to either avoid regulatory scrutiny, or to benefit from preferential statutory treatment; and (2) exploitative—i.e., the utilization of a regulation for purposes at odds with the statute’s purported legislative intent. As explained further in Part II, an example of ontological copyright arbitrage is Pandora’s purchase of a terrestrial radio station in order to take advantage of the lower statutory royalty rate afforded to mixed-use companies. An example of exploitative copyright arbitrage is a copyright owner’s issuance of an unfounded takedown notice.

As with economic arbitrage, regulatory arbitrage is not necessarily detrimental to social welfare. In some contexts, regulatory arbitrage may serve as an impetus for technological innovation, or may signal to Congress an extant imbalance in statutory treatment of similarly situated entities, potentially resulting in societal benefit on balance.[17] As described further in Part I, Burk describes examples of technological innovation as a potential serendipitous byproduct of attempted manipulation of a regulatory scheme.[18] In their work on “carrots and sticks” in innovation, Ian Ayres and Amy Kapczynski cite the development of more energy-efficient cars and lightbulbs as a direct result of regulatory avoision.[19] Likewise, in some cases copyright arbitrage works to expose and correct extant gaps or inefficiencies in the regulatory regime. For example, and as discussed further in Part II.B.1, Pandora exposed a significant regulatory gap when it purchased a local terrestrial radio station for no business purpose other than to lower its royalty rate—a rate nearly double that of similarly situated Internet radio services with a terrestrial component.

In other cases, however, copyright arbitrage may contravene one or another of copyright’s foundational goals of incentivizing the creation of, and ensuring access to, copyrightable works.[20] Depending on the circumstances, arbitrage in the copyright context may thwart the incentive theory of copyright[21] by reducing the financial benefit that copyright protection affords rights holders (when perpetrated by users of content), or may contravene users’ rights and shrink the public domain (when perpetrated by rightsholders). From a welfare perspective, this suggests that some of the instances of copyright arbitrage described herein arguably either: (1) improve both user access (to content that might otherwise have been unavailable or unaffordable) and platform profits (since regulatory avoision reduces costs), but leave copyright owners worse off (since they are paid less than they’d otherwise be paid in a world without arbitrage); or (2) increase copyright owners’ earnings (e.g., by inflating rates or restricting access), but leave users and platforms worse off (since access is lowered and/or rates are increased). In either case, copyright arbitrage upsets the intended balance struck by Congress.

The blame for the potential problems posed by copyright arbitrage rarely lies with the companies, content owners, or users that engage in it; in most cases, these parties are acting rationally and predictably. The problem lies instead with copyright regulation itself, and its inability to anticipate, and to keep up with, rapidly evolving technology and its distributional effects. In the context of economic (as opposed to regulatory) arbitrage, the policy implications of shifting market allocations take a backseat to market effects, which tend to be net-neutral.[22] Copyright, on the other hand, is uniquely concerned with initial allocations, such that their manipulation may not always be net-neutral in effect. Unlike economic arbitrage, then, copyright arbitrage cannot always be accurately characterized as mere wealth transfer. Instead, copyright arbitrage may result in wealth reduction for the very parties (be they creators or users) that the legislature intended be rewarded and protected by the regime. When this is the case, it stands in direct contravention of copyright’s constitutional mandate to “promote the progress of Science and useful Arts.”[23]

That copyright arbitrage frustrates the production or dissemination of creative expression assumes current copyright protections are set at optimal levels of strength and duration; otherwise, arbitrage may simply correct an over- or under-protective copyright regime. To put it mildly, this is by no means a foregone conclusion, neither in the literature nor in practice. Indeed, the consensus among many intellectual property (IP) scholars is that the current copyright term—life of the author plus seventy years (or, in the case of works made for hire, ninety-five years from the date of distribution, or 120 years from the date of creation, whichever comes first)—is too long.[24] Future work will expand upon nascent empirical work in this area by considering the commercial half-life of information goods.[25] In the meantime, this Article will suggest that where copyright arbitrage is determined to be problematic, the potential mitigation approaches proposed in Part IV hold even in the likely event that protections are sub-optimal.

In sum, where copyright arbitrage works to expose a statutory rate as miscalculated, or as potentially discouraging or thwarting innovative and/or creative efforts, it may be best to let such arbitrage lie, so that it can serve as a signal to legislators and rate-setting bodies that adjustment is needed. In these cases, copyright arbitrage may expose extant inefficiencies. Alternately, in cases where copyright arbitrage may lead to market inefficiency, raise distributive justice concerns, impact consumer expectations, or have anticompetitive consequences, this Article suggests that the best means of mitigating copyright arbitrage’s negative impact might be a more proactive judiciary, coupled with an empowered administrative body, and backed up by antitrust law.

This Article’s first contribution is descriptive: using a series of recent developments from the copyright industries as examples, it describes a phenomenon that to date has not received much scholarly attention—that of copyright arbitrage. Part I begins with the basics of regulatory arbitrage theory, and considers whether, when, and how it applies in the copyright context. Part II presents several recent examples of copyright arbitrage. Using these examples, Part III discusses the potential implications of copyright arbitrage in the relevant markets.

This Article also offers a normative contribution: Part IV calls for recognition of copyright as a regulatory system in need of a regulatory apparatus. In cases where copyright arbitrage is determined to be problematic, it offers some thoughts on how administrative agencies, Congress, and the courts might mitigate its negative impacts. Specifically, the Article commends a three-pronged approach to identifying, and curbing, problematic copyright arbitrage. First, courts can take a purposive, substantive approach to interpretations of the Copyright Act. Second, Congress can empower a regulatory agency with rulemaking and enforcement authority. Finally, antitrust law can help to curb the anticompetitive effects of copyright arbitrage resulting from legislative capture.

I. Regulatory Arbitrage

When faced with regulation, a regulated entity can either comply or not. Compliance can be conditional—e.g., provisional compliance simultaneous with a regulatory challenge, in court or via lobbyist—or unconditional. Noncompliance likewise comes in various forms: a regulated entity can disobey, in whole or in part, and pay the penalty (monetary or otherwise), or it can engage in various regulatory manipulations. Where the compliance decision is essentially a cost-benefit analysis, a regulated entity may engage in regulatory arbitrage. This framing helps to clarify regulatory arbitrage as a legal form of regulatory noncompliance. This Section lays out the theoretical underpinnings of regulatory arbitrage, the conditions that encourage it, and the factors that limit it.

A. The Theory of Regulatory Arbitrage

Economic arbitrage—i.e., the practice of taking advantage of a difference in price between markets—can be beneficial, or harmful, or can simply amount to net-neutral wealth transfer.[26] Similarly, regulatory arbitrage may be problematic, or may work to correct legislative imbalances, or both (since any policy change thus effected is likely driven by private actors and not by Congress).

The circumstances in which regulatory arbitrage may prove beneficial vary. For example, where a regulation is driven by special interests in the first place, regulatory arbitrage may help to mitigate the overall reduction in social welfare.[27] Occasional arbitrage may also come to be accepted, and even relied upon, in certain regulatory settings. The Internal Revenue Service’s (IRS) so-called “Wall Street Rule,” for example, refers to the agency’s position that it will not challenge a transaction “if there is a long-standing and generally accepted understanding of [the] expected tax treatment.”[28] In a phenomenon known as “planning drift,” continued manipulation of this sort can actually influence public policy over time.[29]

The so-called “sharing economy”—Uber, Airbnb, and the like—is often cited as a form of regulatory arbitrage that takes advantage of divergent regulatory frameworks. These “regulatory entrepreneurs”—companies for whom “changing the law [is] a material part of [their] business plan”[30]—are frequently credited with exposing outdated or needless regulation. Uber is still a profit-maximizing entity, however, and need not take societal benefit into consideration when pushing for regulatory change. This can lead to regulatory capture.

Finally, where similarly situated entities are engaging in regulatory arbitrage, it might be irrational from a wealth maximization perspective for a company not to do the same. For example, most corporate statutes restrict a firm’s ability to declare and pay dividends that exceed the stated capital of the firm.[31] To avoid this regulation and the costly liability that comes with it, corporate boards will often set an exceedingly small par value per share (or, where allowed, no par value at all).[32]

On the other hand, some regulatory arbitrage can be problematic. This is because at its core, regulatory arbitrage is distortionary behavior that can thwart regulatory intent and disadvantage actors who play by the rules. For example, large pharmaceutical companies’ exploitation of Federal Drug Administration loopholes has allowed those companies to block generic competitors by preventing them from conducting studies required for approval, thereby delaying those companies’ entry into the market.[33] This delay denies consumers the option of lower-cost drugs. The Fair Access for Safe and Timely Generics Act of 2017 (FAST) was recently introduced in hope of reducing the anticompetitive exploitation of these loopholes.[34]

Regulatory arbitrage is typically characterized as either ontological, technological, or both.[35] In addition, and as discussed further in this Article, all regulatory arbitrage is arguably exploitative, in the sense that it seeks to manipulate—for better or for worse—some loophole in the regulatory regime. For this reason, the examples discussed in Part II.B are classified as either exploitative, or both exploitative and ontological. Ontological regulatory arbitrage allows an entity to either switch to, or away from, a particular label, depending on whether it aims to avoid, or to bring itself under, a particular regulation. An example is when a business decides to incorporate offshore to avoid paying domestic taxes.

Technological regulatory arbitrage entails coding, designing, or otherwise innovating around a regulation so as to avoid its purview altogether. For example, Wu describes the intentional design choices of early P2P platforms Gnutella and KaZaa as “founded on concepts of radical decentralization” and “clearly designed to avoid the copyright lawsuit that had befallen Napster.”[36] To be clear, the platforms’ design-around did not eliminate, or even mitigate, the copyright infringement concern; nonetheless, “[t]o date, neither GnutellaNet nor its main application designers have been sued, despite the substantial volume of infringement they facilitate,”[37] demonstrating that technical compliance with the law may excuse failure to comply with its spirit.

Burk offers the case of Chrysler’s PT Cruiser vehicle as an example that fits into both the ontological and technological arbitrage categories: the PT Cruiser’s design—characterized by the unique distance of its wheels both from each other and from the sides of the vehicle—was less an attempt to evoke the styling of the 1930s, and more an attempt to classify itself as a “light truck” instead of a “passenger car,” thereby avoiding costly fuel efficiency regulations.[38] The bet paid off: at 1.3 million units sold, the PT Cruiser was declared a “massive sales success,” with sales “beyond anything Chrysler could have imagined.”[39]

Where this Article describes regulatory arbitrage as exploitative, it means to say that instead of merely manipulating its categorization—i.e., ontological arbitrage—an entity might affirmatively utilize a statutory regime for ends at odds with a statute’s purported legislative intent. An example is the so-called “double-dip” lease, in which a company in one jurisdiction buys something—say, a building—and leases it to a company in another jurisdiction, such that each is treated as the “owner” in their respective jurisdiction, with each enjoying depreciation on the asset.[40] In this example, the regulated entities are actively taking advantage of the depreciation rules, but doing so in a way that diverges from the legislative intent behind those rules in order to reduce their regulatory costs.

Coase and transaction cost economics tell us that a rational entity will structure its dealings in such a way as to minimize transaction costs.[41] When a firm or individual engages in regulatory arbitrage, however, it may actually assume higher transaction costs in exchange for lower regulatory costs. As Fleischer explains, “So long as the regulatory savings outweigh the increase in transaction costs, such planning is perfectly rational.”[42] Owing to regulatory cost savings, both ontological and exploitative arbitrage result in overall cost savings, even if transaction costs are higher. The next two Sections discuss other conditions for, and constraints on, the occurrence of regulatory arbitrage, both generally and in copyright specifically.

B. Conditions Promoting Regulatory Arbitrage

In his work on regulatory arbitrage, Victor Fleischer borrows from financial arbitrage theory to establish a taxonomy of inconsistencies under which regulatory arbitrage is likely to take place. This Section will consider each in turn as they apply in the copyright context.

  1. Regulatory-Regime Inconsistency

The first condition under which regulatory arbitrage is likely to arise is “regulatory-regime inconsistency,” which occurs when the same transaction receives different regulatory treatment under different regulatory regimes.[43] This doctrinal inconsistency may result from outdated laws, or from divergent policy goals from one industry, or jurisdiction, to the next.[44] The classic example of arbitrage stemming from regulatory-regime inconsistency is the decision by a company to incorporate in a state like Delaware, with its more favorable tax treatment (among other advantages). As a regime, copyright is infamous for its inability to keep up with rapidly changing technology, and for its heavily lobbied rules affording different treatment—including differing statutory rates—to otherwise similarly situated parties. These issues make copyright particularly vulnerable to this brand of arbitrage.

  1. Economic Substance Inconsistency

The second type of inconsistency—“economic substance inconsistency”—can take place within a single jurisdiction and involves changing the formal structure of a business deal without affecting the underlying transaction.[45] This type of regulatory arbitrage can range from the simple to the elaborate. A common example is the total-return swap, which Fleischer describes as a means of avoiding dividend tax by running earnings through an investment bank[46]: same result, but different regulatory treatment. With its fungible, and often ambiguous categories for both licensors and licensees, copyright law is also susceptible to this type of inconsistency.

  1. Time Inconsistency

The third and final inconsistency condition outlined by Fleischer is that of “time inconsistency.” It derives from the differential treatment of a transaction over time.[47] Where a regulated entity knows that a new law is coming, for example, it can work to delay a transaction so as to benefit from favorable treatment under the new law.[48] The call for a “next great Copyright Act,”[49] coupled with the fact that copyright has not undergone comprehensive revision since 1976, make it highly vulnerable to time-based arbitrage. The grandfathering of existing music services under the Digital Millennium Copyright Act (DMCA) is a prime example of this vulnerability. The DMCA establishes four categories of digital audio services which can be broken down broadly into preexisting and other. Preexisting services are subject to the rate standard set forth in Section 801(b)(1) of the Copyright Act—a four-factor standard that notably does not consider the market rate. Other digital audio services, both subscription and nonsubscription, are subject to rates set by the Copyright Royalty Board (CRB), which intended to emulate fair market value by considering what a willing buyer and willing seller would agree to in a hypothetical free market. This allows SiriusXM, a preexisting service, to pay a lower effective rate than comparably situated latecomers.[50]

Fleischer emphasizes that successful regulatory arbitrage does not require economically or strategically identical transactions; close substitutes are sufficient.[51] In other words, two possible deal structures—one facing high regulatory costs and the other minimizing or avoiding them—need merely be similar enough. So long as the regulatory savings under the substitute transaction outweigh any additional transaction costs that may be incurred, a rational entity will engage in regulatory arbitrage. Regulatory-regime, economic substance, and time inconsistencies abound in the copyright context, thereby encouraging regulatory arbitrage in the space.

C. Constraints on Regulatory Arbitrage

Several factors may constrain how and when entities might elect to engage in regulatory arbitrage, even where the requisite conditions exist. In addition to the need for an economically and strategically close substitute transaction, Fleischer adds the following potential constraints on the ability to successfully engage in regulatory arbitrage: legal, transaction costs, professional, ethical, and political.[52]

  1. Legal Constraints

Some regulatory regimes have explicit rules intended to discourage arbitrage opportunities anticipated, or later discovered, by lawmakers. One example is proposed Section 710 of the tax code, which contains a provision treating general partners’ debt-financed investments as if they carried interest.[53] This proposal represents Congress’s attempt to curb the regulatory arbitrage move of avoiding carried interest by borrowing from limited investors.[54]

Other regulatory regimes have broader rules aimed not at a specific maneuver, but rather at an entire category of suspect transactions, or aimed at curbing all manner of regulatory avoidance more generally. Some examples include the passive loss rules in tax (aimed not at any specific deal structure, but rather at all tax shelter activity),[55] and the common law constraints on tax avoidance: substance-over-form, economic substance doctrine (also known as the “sham transaction doctrine”), business purpose doctrine, and the step transaction doctrine.[56] Notably, copyright does not currently have any explicit rules aimed at curbing regulatory arbitrage.

  1. Transaction Costs

What Fleischer calls “transaction costs” might be more accurately called “frictions,” as they refer not only to Coasean transaction costs, but also to risk aversion, search costs, and information asymmetry.[57] The costs of regulatory avoidance, then, go beyond the typical transaction costs necessary to conduct a transaction, such as attorneys’ fees, to also include costs associated with financial reporting, adverse selection, agency costs, and moral hazard.[58] Where an entity nonetheless engages in regulatory arbitrage, we can assume that the regulatory cost savings outweigh the sum total of various frictional costs.

Given the disparity in size and sophistication between different licensors and licensees of copyrighted content, a likely source of friction in copyright arbitrage is a difference in bargaining power. Prior work on private ordering in copyright has detailed this concern as one of adverse selection: larger, more powerful parties opt out of a regulatory regime, leaving smaller, weaker parties behind.[59]

In 2012, for example, singer-songwriter Taylor Swift and media conglomerate Clear Channel struck a deal in which Clear Channel agreed to pay Swift a terrestrial performance royalty (an obligation that does not exist under current copyright law) in exchange for a lower-than-statutory digital performance royalty.[60] In addition to opting out of Section 114’s statutory royalty scheme, this agreement also released the parties from the statutory obligation to pay third parties (in this case, session musicians and backup singers).[61] In other words, a potentially good deal for Swift, but not so great for the smaller musicians whose royalty expectations were retroactively eliminated.

Similarly, in 2014, two of the country’s three major music publishing companies withdrew their digital performance rights from the American Society of Composers, Authors, and Publishers (ASCAP), the performance rights organization that had previously administered those rights.[62] This left smaller publishing companies and individual songwriters (who lack the capacity to withdraw) holding the bag for the majority of ASCAP’s overhead (derived almost entirely from the analog side of the business).

In both of these instances, prohibitive transaction costs excluded the adversely affected parties from the substitute transaction.

  1. Reputational Concerns

The final three constraint categories outlined by Fleischer are professional, ethical, and political. He describes professional constraints as “institutional constraints that follow from being a member of the legal profession and a partner at a law firm”; ethical constraints as “personal moral obligations specific to lawyers as individuals, separate from any professional or institutional pressures”; and political constraints as “pressures not to proceed with the planning strategy separate from any legal, professional, ethical, or moral concern.”[63]

In the copyright context, these three constraint categories can be described jointly as “reputational concerns.” While copyright lawyers are subject to the professional and ethical rules of their respective jurisdictions, and those of the state and federal courts in which they practice, all players in the copyright space—from content owners to platforms to collective rights organizations—engage in continuous, repeat interactions, making reputation an important consideration in the regulatory arbitrage calculus.

In a world with only three major music publishers, and only three major performance rights organizations, reputational concerns can play a significant role in constraining copyright arbitrage. The role may be retaliatory: licensors wishing to show their displeasure with a licensee’s move to lower its royalties vis-à-vis a certain use of content through arbitrage might, for example—assuming they steer clear of collusion and other antitrust concerns—decide to charge the company more for a different use of that same content. The role might also be preemptive: licensors unhappy with an arbitrage move that takes advantage of a time delay in copyright registrations might begin filing those registrations several months before the actual release of new content.

Repeat interaction may also contribute to the role of political constraint in the copyright context. All proposed copyright legislation and amendments come before the U.S. House Judiciary Committee on Courts, Intellectual Property, and the Internet. As many of the committee members serve repeat terms, it is in the best interest of parties who may be seeking to advance a particular piece of legislation not to engage in transactions potentially viewed as abusive or exploitative.

As the examples in the next Section demonstrate, a copyright entity seen as acting in bad faith may also have difficulty in securing cooperation from its counterparts, which may be able to work together to frustrate the bad actors’ attempts at regulatory manipulation. Despite these potential constraints, instances of copyright arbitrage continue to proliferate, suggesting they alone are not sufficient to curb problematic copyright arbitrage. The next Section describes some recent examples of copyright arbitrage.

II. Copyright Arbitrage

Examples of copyright arbitrage date back to at least the jukebox.[64] The “video showcasing” cases in the 1980s also demonstrate the phenomenon. Those cases involved video cassette rental stores operating in-store “showcases”— small booths in which a few patrons could sit and watch a video cassette copy of a film. The question in those cases was whether these screenings constituted a “public performance” as defined by Section 101 of the Copyright Act. The Third Circuit determined that they did, and that the showcases were just a ploy to avoid paying public performance royalties.[65]

A few years later, the Ninth Circuit considered a strikingly similar business model—that of a hotel’s front desk renting videocassettes to guests to view in their rooms. The court distinguished this from the booths in Redd Horne and Aveco by noting that “[w]hile the hotel may indeed be ‘open to the public,’ a guest’s hotel room, once rented, is not.”[66] This allowed hotel owners to design around the public performance right.

The activity of copyright arbitrage has proliferated rapidly over the last few years. This Section first considers the impetus behind the recent proliferation of copyright arbitrage and then turns to six recent examples.

A. The Impetus for Contemporary Copyright Arbitrage

One explanation for the recent increase in copyright arbitrage is that the copyright industries, more so than industries affiliated with most other fields of law, have been dramatically impacted by rapid technological development, in ways both positive and negative. As just one such example, the last few years have seen dramatic advancements in cell phone technology (especially cameras) and mobile applications that enable everything from music streaming to live program recording. The availability of these technologies impacts consumer preferences, challenges old business models, and provides an impetus for new ones.

Importantly, the rapid pace with which these technologies evolve makes it difficult for copyright law to adequately address—or, in some cases, to address at all—the myriad legal concerns they present. This is a well-worn theme in copyright and entertainment: beginning with Sony[67]in which the Supreme Court held time-shifting of television programming via Sony’s Betamax machine to be fair use—and continuing with Cartoon Network[68]in which the Second Circuit held similarly with respect to an early DVR system—on through to Napster[69]—where the Ninth Circuit found the Internet’s first stab at P2P to amount to nothing short of mass copyright infringement—and beyond, technological developments have always challenged the boundaries of copyright law.

In particular, the shift from analog to digital technologies continues to challenge the definition and import of key concepts in copyright law.[70] For example, prior to the digital age, copies were of utmost importance because it cost a lot to make and sell them. A printing press, for example, had to invest a substantial sum in typesetting and assembling a book. No such investment is required to convert a file to a PDF and email it. In addition, a file backed up to the cloud may make “copies” of the underlying content, yet not constitute infringement.[71] Similarly, in the days of in-person, synchronous performance, the question of whether a performance was public did not push legal boundaries the way on-demand streaming might.[72]

Another explanation for the recent, rapid proliferation of copyright arbitrage is the disruptive impact technology has had on the business models of the information goods industries—i.e., movies, television, music and books. The last two decades have brought remarkable changes in the manner, ease, and speed with which users can access copyrighted materials, both legally and illegally. Social media platforms have changed the way we learn about, share, and create content. In turn, these developments have informed and influenced the preferences of consumers to which platforms are beholden, often in the absence of legal guidance. For example, Facebook initially introduced its “share” button without restriction as to what sort of content—including copyrighted content—could be posted and re-posted.[73]

Finally, the heavily lobbied and piecemeal nature of copyright law—described by one commentator as a “swollen, barnacle-encrusted collection of incomprehensible prose”[74]—has led to multiple instances of disparate treatment of similarly situated entities.[75] The perceived inequities of such a regime also encourage gaming.

For all of these reasons, copyright regulation—especially in the content-licensing context—is notoriously complex, and subject to constant revision and uneven application. As a result, many players, including both licensors and licensees, have gone from trying to influence copyright’s labyrinthine rate-setting procedure, to avoiding statutory licenses altogether. Prior work has discussed this move toward private ordering in copyright as potentially enhancing efficiency, while at the same time introducing both adverse selection and distributive justice concerns.[76]

As a subversive form of private ordering, copyright arbitrage shares these advantages and disadvantages. Part III will discuss these possible implications in detail. First, the next Section will consider examples of copyright arbitrage of two different, frequently overlapping types: ontological (which encompasses most technological arbitrage in this space) and exploitative (which applies across the board to all instances).

B. Examples of Copyright Arbitrage

This Section discusses recent examples of copyright arbitrage. Some of the examples in this Section are ontological, meaning they involve an entity seeking to fit, or not fit, a regulatory definition so as to either avoid regulatory scrutiny, or to benefit from preferential statutory treatment. All of the examples might fairly be described as exploitative, which is to say that in addition to, or instead of, an entity working to manipulate its categorization or label, it affirmatively utilizes a statutory regime for purposes at odds with the statute’s legislative intent. Several of the examples fit into both categories.

  1. Section 114 Arbitrage: Pandora’s Terrestrial Radio Station

Our first example is a clear exemplar of ontological arbitrage, wherein Internet radio service Pandora seeks to recategorize itself so as to save regulatory costs. ASCAP is the largest of three performance rights organizations (PROs) in the U.S. PROs like ASCAP serve as administrators and collection agents for public performance royalties, i.e., royalties incurred from the play of a song on analog or digital radio.

There is no compulsory license for the public performance of musical compositions; instead, the rates ASCAP charges are set by its membership, composed of both individual songwriters and music publishing companies.[77] Challenges to those rates can be brought in the U.S. District Court for the Southern District of New York (SDNY) in its capacity as “rate court.”[78] Challenges raised at the rate court take time, and there is no guarantee of success. Alternatively, a licensee like Pandora can engage in arbitrage in order to achieve a lower rate.

Pandora is a digital radio platform and a licensee of ASCAP. In 2015, ASCAP’s going rate for digital licensees like Pandora was approximately 4 percent of a licensee’s revenue.[79] The going rate for terrestrial radio stations, including those owned by digital radio properties such as iHeartMedia,[80] was less than half that at 1.7 percent. This reduced rate was the result of an agreement between ASCAP and the Radio Licensing Marketing Committee (RMLC), an organization that lobbies on behalf of commercial radio stations.[81] In 2015—in an effort to recategorize itself from “digital radio station,” with a rate of 4 percent of revenues, to “terrestrial radio station with digital component,” thereby reducing its royalty rate to 1.7 percent—Pandora purchased a small, terrestrial radio station, KXMZ-FM, in Rapid City, South Dakota, for $600,000.[82] The investment was a no-brainer for a company who was paying out nearly half of its revenue in total royalties.[83]

The applicable statute doesn’t prohibit Internet radio stations from expanding into the terrestrial space, but it also doesn’t contemplate them doing so for the express and stated purpose of manipulating their royalty rate. In an op-ed penned by Pandora’s then-assistant general counsel Chris Harrison, Pandora made little attempt to characterize the transaction as anything other than ontological copyright arbitrage:

[W]e have encountered many attempts by the incumbent industry players to undermine Pandora’s mission to connect millions of fans with the music and artists they love. These organizations seek to impose unprecedented royalty increases that are neither reasonable nor competitive. . . . Terrestrial broadcasters and their Internet properties were given preferential treatment via a January 2012 agreement between the Radio Licensing Marketing Committee (RMLC) and ASCAP and BMI. To put this in perspective, at least 16 of the top 20 Internet radio services that compete with Pandora operate under the RMLC license that has not been made available to Pandora. . . . So, today we are also announcing the purchase of KXMZ-FM, a terrestrial radio station broadcasting out of Rapid City, South Dakota. This acquisition allows us to qualify for the same RMLC license under the same terms as our competitors.[84]

In other words, Pandora views its purchase of the terrestrial radio station as an appropriate response to the disparate treatment of Internet and terrestrial radio under the Digital Performance Right in Sound Recordings Act of 1995 (DPRSRA).[85] In introducing a performance right for digital transmissions, the DPRSRA effectively established three pay tiers: (1) terrestrial broadcast retransmissions are exempt from payment; (2) noninteractive Internet radio platforms are subject to the statutory license, with rates set (and adjusted at regular intervals) by the CRB; and (3) interactive Internet radio platforms must negotiate directly with the content owners.[86] This gives a distinct competitive advantage to existing terrestrial radio stations whose programming might later be rebroadcast online, while making it comparatively more expensive to operate a competing Internet radio service. From this perspective, Pandora could argue that it is merely leveling the playing field, and that in challenging the terrestrial broadcaster oligopoly, its arbitrage is actually a social benefit, or at least a rational response to preexisting arbitrage. The veracity of such an argument depends on whether we think there is social value in treating Internet and terrestrial radio differently, or if the dichotomy is solely the result of industry capture.

  1. Section 115 Arbitrage: Mass “Address Unknown” Notices of Intent

Our second example describes an instance of exploitative arbitrage in which several licensees follow the letter of the law in a manner that may yield them regulatory savings not contemplated by the statute. Traditionally referred to as the license for cover songs—the purpose for which it was originally intended—Section 115 of the Copyright Act establishes a compulsory license for the use of musical compositions.[87] The DPRSRA extended this license to digital distributions with the addition of Section 115(c)(3).[88] Importantly for this example, Section 115(c)(3) allows digital streaming services to utilize the compulsory license for use of the musical compositions underlying the sound recordings played by their consumers.[89] The statutory rate and terms are set, and adjusted at five-year intervals, by the CRB—a body of three appointed judges who serve staggered six-year terms—with input from interested parties.[90]

In order to avail oneself of the compulsory license, Section 115(b) requires a prospective licensee to first file a Notice of Intent (NOI) with the copyright owner.[91] Traditionally, owners of copyrights on musical compositions engage the licensing agency Harry Fox, or one of a couple other smaller agencies, to manage their catalog, and to collect and administer their statutory royalties under Section 115.[92] Likewise, prospective licensees will typically turn to Harry Fox, or a similar agency, in order to license repertoire that the agency manages. In such a case, the agency then issues an NOI to respective content owners on behalf of the statutory licensee.

In either case, where no copyright owner or address can be identified from the registration records held by the Copyright Office, the prospective licensee (or its agent) can instead file an “address unknown” NOI with the Copyright Office, which maintains a list of such notices online.[93] The purported goal of this list is to afford copyright owners the opportunity to identify themselves and to claim their royalties, while at the same time allowing for uninterrupted use of the occasional orphan work by licensees.[94] The logistics of identifying and claiming an “address unknown” NOI have been questioned.[95] Notably, the statute doesn’t have any kind of “reasonable” or “good faith” search requirement; a prospective licensee can limit their search to the Copyright Office’s public records only. Importantly, statutory royalties under Section 115 accrue only after a copyright owner has been identified.[96] A prospective licensee who files an “address unknown” NOI therefore enjoys a period of gratis use up until the point that the content owner identifies and claims their content. A licensee who takes advantage of this gratis use period for content that isn’t actually “unknown” would arguably be engaging in exploitative arbitrage.

Beginning in April 2016, Google, Pandora, Spotify, and Amazon, among others, began serving “address unknown” NOIs on the Copyright Office in unprecedented numbers. Specifically, these companies filed more than twenty-five million NOIs in the eight months between April 2016 and January 2017.[97] That’s an average rate of roughly three million filings per month, suggesting an effort to shift the burden to rightsholders to assert their ownership rather than good-faith failures to determine who owns what, while allowing licensees to use the content for free unless and until it is claimed. Licensees maintain that it’s not unusual for them to receive incomplete data from publishing companies, and that in many cases this lack of information explains the need for filing “address unknown” NOIs. In a 2015 press release announcing their intention to build a better system than Harry Fox’s, Spotify notes:

One of the most difficult challenges is the lack of accurate data as to who owns the rights to a specific track, especially when it comes to songwriter and publisher rights. In many cases, the ownership of the rights are [sic] not even finalized when a record is released; in many other cases, rights are held by multiple parties, rights change hands, and rightsholders remain entirely unclear.[98]

Incomplete publisher information has always been a challenge, however, and might not fully explain the recent rapid proliferation of these notices.

Indeed, if all of the filings were legitimate, this might simply suggest a problem of scale. Unfortunately, not all of the filings appear to be made in good faith. For example, a recent search of the Copyright Office’s Section 115 “address unknown” database turned up a notice posted by Spotify for the song “The Window Up Above,” written by iconic country music artist George Jones.[99] As recorded by Jones, the track peaked at number two on the Billboard Hot Country chart in 1961, where it sat for thirty-four weeks.[100] In 1975, Mickey Gilley’s recording of the same song was the number one hit single on the Billboard Hot Country chart for fifteen weeks.[101] In other words, this is not an unknown song by an unknown artist uploading tracks from her parents’ basement. There are currently eighty-four George Jones albums available for streaming on Spotify,[102] suggesting Spotify might otherwise be able to identify the rights holder for this song.

Some commentators have labeled this behavior an exploitative misuse of Section 115’s “address unknown” NOI protocol:

Whether their motivation is avoiding liability, avoiding royalties, or both, this means that Amazon, Google, Pandora and others pay no statutory royalties on any of the song copyrights in their millions of “address unknown” NOIs until the song copyright owner becomes “identifiable” in the “public records” of the U.S. Copyright Office.[103]

How, exactly, can the filing of “address unknown” NOIs delay (and ultimately reduce) royalties paid by these companies? The answer lies in the language of Section 115, which instead of requiring a good-faith search of all available sources—Harry Fox, Google, etc.—directs a prospective licensee to the Copyright Office’s public records in order to identify the copyright owner.[104] Importantly, royalties do not begin to accrue unless and until the copyright owner claims the content. Thus, these commentators worry that the filing of mass NOIs is a pay-stall tactic.

Exacerbating the problem, they suggest, is the fact that the Copyright Office is understaffed and overworked, with a tremendous backlog of registrations waiting to be processed. According to the Copyright Office’s own registration guidelines, the processing time for new registrations (i.e., registrations relating to newly released content) currently ranges from an average of seven to sixteen months.[105] Older releases may also be affected, either because they were registered before 1978 (and so aren’t yet digitized and publicly searchable), or because they were never registered at all. Even a song whose title varies slightly from version to version can be affected—for example, there may be a registration for Prince’s song “Purple Rain,” but if the prospective licensee searches the database for “Purple Rain [Live],” it won’t turn up the track.

In any of these cases, an “address unknown” NOI could be exploited to avoid both infringement liability and royalty payment, even for properly registered works. In the meantime, SoundExchange—a PRO for digital performance royalties—has launched a specialized search engine to help artists find out if any of their works are among the sixty million reported NOI filings with the Copyright Office.[106]

Section 115’s “address unknown” NOI protocol was designed to permit continuity of use for the occasional orphan work, and as a means of enabling rights holders for those works a second chance to make themselves known. It was not intended as a means of punishing content owners for an understaffed Copyright Office.[107] If these accusations are accurate, this seems a clear example of problematic copyright arbitrage. While the backlog and staffing problems at the Copyright Office cannot be attributed to Google et al., the alleged bad faith involved in taking advantage of the circumstances may be. In such a case, this would suggest problematic copyright arbitrage.

Another explanation for an increase in the number of “address unknown” NOIs filed might not be indicative of problematic arbitrage at all. Instead, evolving business models and consumer interest in streaming services may have exacerbated extant backlog caused by incomplete publisher data. There is some evidence to suggest that some licensees, such as Spotify, have established a so-called “pending and unmatched” or “P&U” account for these unmatched mechanical royalties:

Due to a combination of the song-by-song mechanical licensing regime in the U.S. and a lack of reliable publishing information, many songs that are streamed on Spotify cannot be matched with publishers to earn royalties, and the money accrues as P&U. . . . Third parties like RightsFlow have stepped in to alleviate the problem, but even then only 75% of P&U royalties are matched. Israelite [President and CEO of the NMPA], confident that Spotify is acting in good faith, affirmed that the NMPA is undergoing talks with Spotify to reach a settlement so that the large pool of money can be distributed among NMPA member publishers. Israelite indicated that there will be a claims process in which member publishers can manually claim their ownership of the unmatched songs and receive their portion of the P&U money.[108]

If true, this explanation is less indicative of exploitative arbitrage and more indicative of a good faith attempt to accrue and distribute mechanical royalties at some point in the future.

  1. Section 109(a) Arbitrage

a. ReDigi’s Second-Hand Digital Music Marketplace

Part II.A defined regulatory arbitrage as technical compliance with the letter of the law, if not the spirit. In the case of ReDigi, the “technical compliance” piece is yet to be finally determined.[109] A colorable argument has been made that the statute at least doesn’t explicitly prohibit the business model in question; more tenuous is the contention that the business model complies with the alleged spirit of the law.[110] This example considers a business model that purports to allow for the sale of “used” MP3s, without an effective means of avoiding the making, nor retention, of copies in violation of Section 106.[111]

Launched in October 2011, ReDigi markets itself as “the world’s first and only online marketplace for digital used music.”[112] Specifically, ReDigi offers its users a platform to “sell their legally acquired digital music files, and buy used digital music from others at a fraction of the price currently available on iTunes.”[113] In other words, ReDigi aspires to be a virtual record store, in which consumers swap MP3 files instead of vinyl records.

The statutory basis on which ReDigi relies is Section 109 of the Copyright Act, commonly known as “the first sale doctrine.”[114] Section 109 provides that “the owner of a particular copy or phonorecord lawfully made under this title, or any person authorized by such owner, is entitled, without the authority of the copyright owner, to sell or otherwise dispose of the possession of that copy or phonorecord.”[115] It is the first sale doctrine, for example, that allows a student to sell her used textbook back to the university book store at the end of a semester. The bookstore acquires the textbook; the student leaves with cash (but no textbook).

In other words, the conventional underpinning of the first sale doctrine is that the person who sells the good no longer has access to it. Unlike vinyl records or CDs, digital music files can be reproduced without deterioration of quality and shared without loss of the original copy. ReDigi seeks to ameliorate this “complication” by requiring its users to first download a proprietary software called “Media Manager” to their computers.[116] Once installed, Media Manager analyzes the digital music files on a user’s computer to determine which are legally acquired, i.e. eligible for sale.[117] Once the eligible files have been identified, the user can upload them to ReDigi’s “Cloud Locker”—a remote server located in Arizona—for sale on the platform’s website.[118]

ReDigi asserts that once a file has been uploaded from a user’s computer to its Cloud Locker, Media Manager deletes the file completely from the user’s computer so that the data does not exist in both places at any one time.[119] There are two potential problems with this. First, the process necessarily involves the making of an unauthorized copy of a file from the user’s computer in order to upload it onto the Cloud Locker server.[120] In the absence of fair use, this infringes a rightsholder’s exclusive right of reproduction.[121] Second, Media Manager can only scan for and delete files located on the computer to which the software is downloaded. It cannot search external drives or other hardware to which a copy of the same file could additionally be stored, such that it relies on the honesty (and/or, perhaps even more worrisome, competency) of the individual user.[122]

Capitol Records was the first to bring suit against ReDigi for mass copyright infringement.[123] At the district court, Judge Sullivan agreed with the record label, finding direct and secondary liability on the part of ReDigi, and granting summary judgement to Capitol in March of 2013.[124] The Second Circuit affirmed in 2018.[125]

Putting aside alternate interpretations of Section 109, and the plethora of fair use issues that this case presents,[126] at heart, ReDigi’s motivation can be described as copyright arbitrage: it is cheaper for ReDigi to offer its users a percentage of the sale price for tracks sold on the platform than it is for ReDigi to license those same tracks for sale from copyright owners. Under this characterization, ReDigi’s business model exploits the uncertainty of Section 109’s application to digital goods, or, it applies form over substance. With regard to its violation of the reproduction right—which is incontrovertibly the case—ReDigi’s argument flips to substance over form (assuming it could ensure the “substance,” i.e., the erasure of the original copy).

Another characterization might describe ReDigi as engaging in “regulatory entrepreneurship”—i.e., a philosophy in which “changing the law [is] a material part of [a company’s] business plan.”[127] Perhaps the service aims to highlight the differences between physical and digital copies—including the ease with which the latter may be reproduced and distributed, as well as an increase in durability—in an effort to encourage the law to treat them differently. While this is arguably a form of innovation, it is not without its downsides. For one thing, ReDigi—as with any profit-seeking actor—is looking to maximize its outcome, not that of society. This is not to say that ReDigi’s efforts might not have the opposite effect; namely, affording resale rights to owners of digital files might encourage greater consumption of digital goods. Such an outcome could still be characterized as rent-seeking on the part of ReDigi and could result in a quiet change of law without popular debate or consensus. Such regulatory capture might be avoided by redirecting copyright arbitrage efforts into a regulatory process.

A broader question here is whether “purchasers” of digital files really “own” them. In their work on the subject, Aaron Perzanowski and Jason Schultz explore how notions of ownership have changed in the digital context.[128] The next example describes vendors of digital content advertising content for sale, where the fine print describes instead a conditional license, subject to revocation by the copyright holder.

c. “Buy Now” Buttons for Licenses

Upon locating an eBook to purchase on, consumers frequently encounter a button labeled “Buy Now.” A similar button labeled simply “Buy” is featured on the iTunes store for music downloads. If either of these buttons led to an actual purchase of content, Section 109(a) might apply, allowing the purchaser to freely sell, lend, or give away her copy.[129] In both of these cases, however, terms of use and end-user license agreements—automatically agreed to upon “purchase”—restrict all manner of transfer, making these transactions at best licenses.[130] Amazon Kindle’s terms of service, for example, state: “Unless specifically indicated otherwise, you may not sell, rent, lease, distribute, broadcast, sublicense, or otherwise assign any rights to the Kindle Content or any portion of it to any third party . . . .”[131]

In their study on the impact of “Buy Now”-style marketing, Aaron Perzanowski and Chris Jay Hoofnagle conclude that “[n]ot only are consumers misled [by the “Buy Now” buttons], they are misled about ownership rights that are important to them. A sizable percentage of consumers express a desire for those rights and many say they are willing to pay more to preserve them.”[132] By utilizing a “Buy Now” message for what actually amounts to a license, these platforms and content owners are able to command a higher price for the content—a price that consumers would not be willing to pay if they understood what they were actually getting. As with ReDigi, this too is an example of ontological arbitrage of Section 109(a) in which platforms and content owners attempt to misclassify licenses as purchases in order to increase profits.

  1. Section 110(11) Arbitrage: VidAngel’s Filtered Streaming Service

The next example of exploitative arbitrage involves a creative interpretation of statutory language concerning what is, and is not, an “authorized copy.” VidAngel is a streaming film service marketed to consumers wishing to filter out objectionable content. The company purchases multiple copies of physical DVDs of various copyrighted movie titles, then uses a software called AnyDVD HD to rip the movies from the DVDs to digital files in order to break them into filterable segments. Consumers then “purchase” a physical DVD from VidAngel for $20. VidAngel retains the physical copy on behalf of the consumer and streams the movie to the consumer’s television with whichever filters (for language, violence, nudity, etc.) they preselect. After viewing, the consumer “returns” the DVD to VidAngel for a refund of the $20 “sales price” less $1–$2 per night (depending on whether standard or high-definition play is selected).[133]

In some cases, VidAngel offered movies for streaming that were otherwise only available via DVD, or for which the corresponding movie studios had licensed exclusively elsewhere.[134] In addition, VidAngel consumers were initially able to forego selecting a filter altogether—or to choose an irrelevant filter, such as “filter out all Star Wars characters” for a non-Star Wars movie—in order to watch films unfiltered. VidAngel eventually corrected for this by requiring at least one relevant filter.[135]

Led by Disney, several movie studios brought suit against VidAngel for violation of their exclusive rights under the Copyright Act,[136] as well as for circumvention of technological measures under the DMCA.[137] VidAngel raised fair use[138] as a defense, and also argued that its actions were exempt from copyright infringement under the Family Movie Act (FMA) of 2005.[139] The FMA holds, in relevant part, that:

[T]he making imperceptible, by or at the direction of a member of a private household, of limited portions of audio or video content of a motion picture, during a performance in or transmitted to that household for private home viewing, from an authorized copy of the motion picture, or the creation or provision of a computer program or other technology that enables such making imperceptible and that is designed and marketed to be used, at the direction of a member of a private household, for such making imperceptible, if no fixed copy of the altered version of the motion picture is created by such computer program or other technology.[140]

The district court rejected VidAngel’s fair use argument and also found that the FMA didn’t apply since it was only applicable to filtered transmissions that come from an authorized copy of the film.[141] On appeal, the Ninth Circuit affirmed.[142] The applicability of the FMA to VidAngel’s business model turns on whether they are using an “authorized copy.” VidAngel argued that because the company started with a legitimately purchased DVD, the filtering was exempt under Section 110(11). The studios disagreed, noting that the copy ultimately streamed to consumers was ripped, and therefore not “authorized.”

In other words, “VidAngel’s interpretation would create a giant loophole in copyright law, sanctioning infringement so long as it filters some content and a copy of the work was lawfully purchased at some point.”[143] VidAngel’s exploitation of the FMA’s “authorized content” language would allow it to serve content to its consumers for a fraction of the price that a streaming license would cost, thereby amounting to copyright arbitrage. In this case in particular, streaming licenses are legal and readily available, albeit more costly, such that VidAngel cannot make a colorable argument that it is engaging in regulatory entrepreneurship.

  1. Section 106(4) Arbitrage

a. Aereo’s “Private” Broadcasts

This example of copyright arbitrage begins as an instance of exploitative arbitrage and evolves into an attempt at ontological arbitrage. Ultimately successful at neither, television streaming service Aereo offered its subscribers the ability to watch television programs in almost-real time. Aereo’s system was made up of thousands of small antennas—each dedicated to a single subscriber—which would, through a transcoder, transmit content selected by the subscriber over the Internet. A subscriber-specific folder on Aereo’s hard drive would then stream the show to the subscriber with only a few seconds delay.[144]

Broadcasters sued Aereo for copyright infringement, alleging violation of Section 106(4) of the Copyright Act, which gives copyright owners the “exclusive righ[t]” to “perform the copyrighted work publicly.”[145] Aereo maintained that it merely provided equipment to subscribers, who in turn performed the copyrighted work, and that in any case, each subscriber was assigned his own antenna and personal folder, so that any and all performances were private. As in the ReDigi example, Aereo sought to exploit ambiguity in the statute—specifically, its definition of “public performance.”

In considering whether Aereo’s system “performed publicly,” the Supreme Court looked to the Transmit Clause, which holds that a work is performed publicly if (1) it is performed at “a place open to the public or at any place where a substantial number of persons outside of a normal circle of a family and its social acquaintances is gathered”; or if (2) it is transmitted or otherwise performed “to the public, by means of any device or process, whether the member of the public capable of receiving the performance or display receive it in the same place or in separate places and at the same time or at different times.”[146]

Ultimately, the Supreme Court likened Aereo’s system to that of cable companies’ and held that Aereo, like cable companies, publicly performs.[147] Beaten but not yet defeated, Aereo next attempted to use the Court’s own language recognizing “the many similarities between Aereo and cable companies”[148] to argue that it should be allowed to continue operating under Section 111, the compulsory license for cable companies.[149] In other words, the company pivoted to an ontological arbitrage approach, and sought to fit itself under the statutory definition of “cable company” in order to take advantage of statutorily dictated royalties, a considerably more cost-effective option than negotiating a rate with disgruntled broadcasters.[150] At the same time, Aereo specifically sought not to be labeled a “cable company” for the purposes of oversight by the Federal Communications Commission (FCC)—a label which would require direct negotiation with, and payment to, broadcasters. This result is, after all, precisely what Aereo’s business model intended to avoid.[151]

Unfortunately for Aereo, a subsequent letter from the Copyright Office held that Section 111 doesn’t cover “internet retransmissions of broadcast content,” thereby kicking the case over to the FCC.[152] The FCC has been considering how best to define multichannel video programming distributors (MVPDs)—the FCC equivalent of a “cable system”—since 2012 (i.e., before Aereo even came along). As of this writing, the FCC has sought comment, but has not yet issued a ruling, on an MVPD definition.[153]

If Aereo is found to fit that definition, it would seem to contradict their entire purpose for being. Still, Aereo has said that it would accept such a finding, presumably because their real goal is a rework of the entire regulatory regime.[154] The potential problem with this is the same as with any so-called regulatory entrepreneur: Aereo’s legislative agenda is concerned with maximizing its profits and minimizing its costs, not with net social welfare.

Interestingly, the Aereo decision hasn’t stopped copycat services from trying their hand at live “re-broadcasting.” In early 2018, nonprofit Sports Fans Coalition launched a service called that livestreams fifteen local channels to all five of New York City’s boroughs.[155] The company claims to be different because (a) it doesn’t charge; and (b) it serves an otherwise unserved population: namely, New Yorkers unable to receive a free, over-the-air broadcast signal.[156]

b. Downloads, Samples & Public Performances

This example involves ontological arbitrage around the definition of “perform” in the Copyright Act. In 2007, ASCAP attempted to collect public performance royalties on downloads of MP3s, on downloads of movies and television episodes containing copyrighted music, and on thirty-second song samples offered to prospective consumers prior to purchase on iTunes’s music service.[157] To understand the significance of this attempt, it is helpful to look to the Copyright Act itself. As discussed in the Aereo example, a copyright owner’s exclusive right to public performance comes from Section 106(4), which says that “the owner of copyright under this title has the exclusive rights . . . to perform the copyrighted work publicly. . . .”[158] Section 101 defines “perform” as “recite, render, play, dance, or act it, either directly or by means of any device or process” and defines “publicly” as:

(1) to perform or display it at a place open to the public or at any place where a substantial number of persons outside of a normal circle of a family and its social acquaintances is gathered; or (2) to transmit or otherwise communicate a performance or display of the work to a place specified by clause (1) or to the public, by means of any device or process, whether the members of the public capable of receiving the performance or display receive it in the same place or in separate places and at the same time or at different times.[159]

What songwriters and music publishers sought with their motion, then, was the classification of downloads—be they MP3s, or audiovisual episodes containing copyrighted music—as “public performances” for the purposes of earning royalties on them under Section 106(4). At the time, the NMPA explained the effort as premised on equivalence:

If you watch a TV show on broadcast, cable or satellite TV there is a performance fee collected . . . [b]ut if that same TV show is downloaded over iTunes, there’s not. We’re arguing that the law needs to be clarified that regardless of the method by which a consumer watches the show there is a performance right.[160]

On its face, this characterization seems both reasonable and defensible, but it doesn’t tell the full story. As the Digital Media Association (DiMA)—a trade organization representing platforms that use music, such as iTunes and Pandora—countered at the time: “Songwriters are getting paid. They’re paid sync rights and (mechanical) rights. They aren’t getting paid for the public performance in a download because there is no public performance in a download.”[161]

The question—whether a download triggers the 106(4) exclusive right of public performance, thereby (and importantly) removing the use of the musical composition from the compulsory license—went before the SDNY in its capacity as rate court. Finding a download more akin to a reproduction than a performance, the court held:

Although we acknowledge that the term “perform” should be broadly construed, . . . we can conceive of no construction that extends it to the copying of a digital file from one computer to another in the absence of any perceptible rendition. Rather, the downloading of a music file is more accurately characterized as a method of reproducing that file.[162]

The court noted that its interpretation is consistent not only with case law,[163] and a statement from the Copyright Office,[164] but also with legislative history: “House Report 94, which speaks in terms of the transmission and communication of performances, confirms the requirement of a ‘performance’ [rather than just transmission of data] to trigger the copyright owner’s right of exclusive performance under the Act.”[165] In sum, Judge Connor determined: “We can discern no basis for ASCAP’s sweeping construction of § 101.”[166] As such, the court rejected ASCAP’s attempt at ontological arbitrage of Section 106(4). The Second Circuit affirmed.[167]

  1. Section 512 Arbitrage

a. YouTube’s Alleged “Value Gap”

Because YouTube qualifies for a statutory “safe harbor” under the copyright laws, and as a result effectively pays considerably less than its competitors for the same copyrighted content, the company has been accused of taking advantage of a so-called “value gap.” Up to this point, the examples discussed have all involved a deliberate decision on the part of a platform or content owner to either lower costs or raise revenues via regulatory manipulation. This example also involves regulatory avoision resulting in lower costs for a platform, but lacks the element of intentional planning. Nonetheless, it is included here as an example of a problematic disparity caused by an outpaced regulatory regime without a strong regulatory apparatus.

The Digital Millennium Copyright Act was enacted in 1998.[168] It added, among other things, Section 512 to the Copyright Act.[169] Section 512 establishes a “safe harbor” for online service providers (OSPs) who comply with specified requirements. This safe harbor allows OSPs, like YouTube, to avoid copyright liability for infringing content that is uploaded to a provider’s platform by a user (user-generated content or UGC) without the knowledge of the OSP; or, upon receipt of a notice informing the OSP of the infringing content, is promptly removed.[170] The safe harbor was arguably intended to allow OSPs to operate, and to grow, without fear of crippling liability for copyright infringement effected by their users.[171]

OSPs whose business models prominently feature user-generated content—such as YouTube—rely specifically on Section 512(c), which exempts “for infringement of copyright by reason of the storage at the direction of a user of material that resides on a system or network controlled or operated by or for the service provider . . . .”[172] For the purposes of 512(c), Section 512(k)(1)(B) defines “service provider” as “a provider of online services or network access, or the operator of facilities therefor . . . .”[173] To date, the courts have determined that YouTube meets this definition.[174] Competing digital streaming services, such as Spotify and Apple Music, do not provide a platform for hosting user-generated content. As such, they do not meet the Section 512(k)(1)(B) definition of “service provider” and so do not qualify for the Section 512(c) safe harbor.

Herein lies the rub. Without the benefit of the safe harbor, non-UGC hosting platforms like Spotify must engage in direct negotiation with content owners for use of their songs on the service.[175] These negotiations have resulted in an average market rate for on-demand streams of roughly $7 per 1,000 plays.[176] This is seven times more than YouTube’s average streaming payment of approximately $1 per 1,000 plays.[177] This differential arguably stems from the existence of the safe harbor, which suggests a suboptimal (from the perspective of content owners) baseline for licensing negotiations: zero. This is because takedown of infringing content under Section 512 is temporary; both YouTube and content owners know that users will likely continue uploading and re-uploading content to the platform such that the content will be available whether YouTube pays for it or not.[178]

To be sure, YouTube relies on the Section 512 safe harbor for user-uploaded content beyond music streaming, and its offerings are likely driven more by consumer demand than by strategy. As music streaming has gained in popularity on the platform, however, the backdrop of the safe harbor allows YouTube to forego the per-stream royalty typical of digital streaming licensees, and to instead offer licensors a share of advertising revenue.[179] Those shares are typically substantively lower than going-rate per-plays for digital streams. YouTube recognizes as much, disclosing in its latest 10-K: “We rely on statutory safe harbors, as set forth in the Digital Millennium Copyright Act in the United States and the E-Commerce Directive in Europe, against copyright liability for various linking, caching, and hosting activities. Any legislation or court rulings impacting these safe harbors may adversely impact us.”[180]

When advertising sales decline (or advertising rates decrease), content owners earn less money, but the use of their content does not decrease, which further reduces the effective per-play rate. In addition to allowing YouTube to potentially underpay for content, this model also arguably punishes content owners for a platform’s underperforming business model.[181]

While a “market price” may be difficult to discern,[182] the loophole at play in the YouTube example is not: Congress arguably intended Section 512 to foster the growth of Internet-based companies and services.[183] Congress did not intend to afford certain platforms a competitive advantage over other similarly situated platforms,[184] nor to deprive content owners of the ability to negotiate a fair rate for use of their content. It might alternately be argued that Section 512’s notice and takedown procedure is functioning precisely as intended, and that the problem lies with the parties’ inability to predict the massive scale that it would eventually face. In either case, the use of Section 512 to lower royalty rates, however inadvertent, amounts to copyright arbitrage. The scale of this arbitrage is significant: YouTube accounts for nearly half of all music streaming,[185] yet constitutes less than 4 percent of total streaming revenue earned by content owners.[186]

This is not to suggest that YouTube is not a valuable platform providing a societal benefit, including as a popular distribution platform for content owners themselves. Indeed, another way to view YouTube’s lower rate is as a voluntary agreement to pay a price above zero (the effective “rate” under Section 512(c)). Rather, this Section is a call for a transparent and fairly-reached rate (if one is to be paid)—be it via private bargaining or administrative ratemaking—that is not coerced under the shadow of inevitable infringement by OSP users effectively immunized by the statutory safe harbor.

b. Improper Takedown Notices

In our final example, we see some content owners exploiting Section 512(c)’s notice and takedown procedure by issuing inaccurate, outdated, or unfounded takedown notices.[187] Some of these notices are generated automatically, without regard to fair use and other exemption analyses (in arguable contravention of Section 512(f)).[188] In some cases, takedown notices are even misused to remove negative reviews of a work,[189] or in an attempt to stop a so-called “suck site” on which consumers can post complaints about a company or product.[190]

A recent quantitative study found that roughly 30 percent of takedown notices issued by content owners are “problematic.”[191] Specifically, the researchers identified roughly one out of every twenty-five notices in which the targeted content did not match the allegedly infringed work, “suggesting that 4.5 million requests in the entire six-month data set were fundamentally flawed.”[192] Nearly 20 percent of the takedown notices analyzed raised fundamental questions about whether content owners had even accurately identified the allegedly infringed or infringing works, and suggested a failure to consider fair use.[193]

By removing non-infringing content, even temporarily, content owners misuse the Section 512 takedown mechanism. This misuse can lower costs for content owners by eliminating the need to hire expensive attorneys to conduct fair use analyses, and may also allow some content owners to remove competing or negative content—an ability clearly not contemplated by the statute. On the other hand, content owners can argue that the alternative—an individual fair use determination for every piece of removed content—is cost prohibitive given the staggering volume of infringing uploads.

All of the examples of copyright arbitrage detailed in this Section share a few key features: first, they all involve companies and/or content owners operating in an industry facing rapid technological advancement. This makes it difficult for the law to keep up. Second, each involves an entity subject to at best ambiguous, and at worst uneven, application of existing regulations. A perceived lack of “fairness” among similarly situated players goes a long way toward encouraging gamesmanship in the form of arbitrage, not least of all for its ability to overcome some of the reputational barriers to arbitrage discussed in Part II.C.3. Finally, many of the companies in these examples are—each in their own way—striving to fill a consumer demand not contemplated by the regulatory regime and not otherwise filled in the market. The next Section turns to discuss the implications of these conditions.

III. Implications

Copyright arbitrage may have broad-ranging implications for creators, content owners, users, and platforms alike. Different forms of copyright arbitrage may improve or decrease efficiency in the market, may have intended or incidental anticompetitive and distributional effects, may influence and/or reinforce consumer expectations (for better or worse), and may serve to identify and, in some cases, even correct imbalances and inequities. This Section will discuss these possibilities in turn.

One view of regulatory arbitrage is as a means of driving innovation: a company seeking to lower its costs by recategorizing itself, for example, may devise a new and improved way of doing something. In their work on innovation “sticks”—i.e., a threatened regulatory penalty for failure to innovate—Ian Ayres and Amy Kapczynski present the remarkable impact of Corporate Average Fuel Economy (CAFE) legislation in the emissions context as one such illustration.[194] Enacted by Congress in 1975, the CAFE standards set a fuel-economy standard in miles per gallon that auto manufacturers must meet in a given year, or face civil penalties.[195] A comprehensive review of the program conducted in 2002 found that since the standard’s passage, fuel economy had nearly doubled for passenger vehicles and had increased by 50 percent for light trucks.[196]

Where special interests drive the passage of a particular statute in the first place, regulatory arbitrage may even work to ameliorate the resulting reduction in social welfare.[197] This is because “[a]side from rent-seeking costs, political economists have noted that the content of lobbyist-influenced legislation is likely to be inefficient as well.”[198] One relatively recent example from copyright is the Copyright Term Extension Act of 1998 (CTEA), sometimes referred to as the “Sonny Bono Act” for the singer/congressman who supported it. The CTEA extended the term of copyright from life of the author plus fifty years to life of the author plus seventy years. For works for hire, the term was extended to 120 years from the date of creation or ninety-five years from the date of publication, whichever is shorter. Some characterized the CTEA as “a windfall to the families and corporations that owned these lucrative copyrights,” notably, Walt Disney and its Mickey Mouse franchise.[199] On one view, the extension sought by the CTEA simply aimed to bring the U.S. into compliance with the Berne Convention, where life plus seventy years was already the standard. Another view saw the retroactive extension as taking from the public domain without offering anything in return.[200]

Regulatory arbitrage that seeks to counter the CTEA’s social costs might therefore be socially beneficial. One example is the Internet Archive’s Sonny Bono Memorial Collection.[201] The collection utilizes an obscure section of the Copyright Act—Section 108(h)—that allows libraries to scan and make available works published between 1923 and 1941 so long as they are not currently being sold.[202] The purpose of the archive was to counteract what was perceived as overreaching on the part of CTEA advocates:

[T]he silver lining of the unfortunate Eldred v. Ashcroft decision was the response from people to do something, to actively begin to limit the power of the copyright monopoly through action that promoted open access and CC licensing. . . . As a result, the academy and the general public has [sic] rediscovered the value of the public domain.[203]

In other cases, regulatory arbitrage in the copyright context appears to simply contravene copyright’s purported goals of incentivizing creation and contributing to the public domain. This may be true even where the arbitrage is intended to combat extant inequalities. In contrast to the net-neutral wealth transfer common in economic arbitrage—Company A pays more, while Company B pays less—some copyright arbitrage involves wealth transfer away from the very creators or users that copyright purports to protect. While it is of course true that arbitrage doesn’t last forever—entities will engage in the same manipulations until the advantage afforded the original arbitrageur is depleted—the intervening period can last long enough to distort incentives and lead to market inefficiency, anticompetitive behavior, distributive justice consequences, and inexorably altered consumer expectations. This Section discusses the potential implications of copyright arbitrage, both problematic and otherwise, in turn.

A. Incentives & Inefficiency

Problematic copyright arbitrage, like some other forms of regulatory arbitrage, can prove privately beneficial but socially wasteful. Some of the examples described in Part II involve, at their core, manipulation of existing regulations to secure a lower (or higher) price than might otherwise be obtained in the market. Depending on whether the statutory rate strikes the right balance in the first place, such activity may distort incentives and lead to market inefficiencies.

YouTube’s utilization of the Section 512 safe harbor, for example, allows the company to “negotiate” with copyright owners in the shadow of a statutory protection that effectively allows for gratis use of content uploaded by its users. Both parties come into the negotiation knowing that the content is going to be available on the service in some form or another regardless of the price ultimately reached. This backdrop arguably places content owners in a disadvantaged bargaining position vis-à-vis YouTube, and may result in a lower rate than they might otherwise have obtained if content owners could ultimately deny YouTube access to their content in the absence of acceptable terms. In addition, as YouTube’s market share continues to increase,[204] it necessarily displaces higher-paying services (i.e., any service not protected by the safe harbor), thereby decreasing the total royalties paid out to rights holders.

The incentive theory of copyright would suggest that this can lead to disincentivization of creators and intermediaries, who in turn may produce less new content, thereby decreasing overall consumer welfare.[205] Glynn Lunney has argued that this may not always be the case. In his work looking at the impact of decreased music industry revenues on creative production, he concludes that in the face of diminishing financial incentives, song production has actually increased.[206] One explanation might be diversification on the part of increasingly risk-averse record labels; the conclusion remains the same in any case.

The availability of copyright arbitrage may also disincentivize an otherwise failing company from improving its business model. Pandora’s losses, for example, doubled from $170 million in 2015 to $343 million in 2016,[207] despite the lowering of its royalty rate via purchase of a terrestrial radio station. In another context, a business facing these odds might fail, making room for a more efficient and profitable business to take its place. Instead, copyright arbitrage effectively subsidizes these services by allowing them to use more content than they are able to pay “market” rate for.

Among other things, this can allow nonviable platforms to present themselves to investors and shareholders as more successful than they actually are. In its 2016 annual report, Spotify reported $3.3 billion in revenue—up 52 percent over the year before—and over 140 million subscribers, fifty million of which pay for a monthly plan (the remaining ninety million pay nothing).[208] But the company’s revenue growth has in fact slowed considerably, while its losses continue to grow. In 2016, Spotify’s net loss tallied $600 million, up from $257 million the year before.[209] The utilization of mass “address unknown” NOIs by Spotify allows it to pay fewer royalties, thereby presenting a rosier picture, at least temporarily.

In Pandora’s case, even the (arguably artificial) depression of its royalty rate via purchase of the terrestrial radio station wasn’t enough to push it into profitability.[210] To the suggestion that it ought to charge its users more (than nothing), the company would likely respond that it can’t simply do away with its free tier at this point. While it’s true that users can opt for the reasonably priced premium (i.e. ad-free) tier, Pandora Plus, for only $4.99 per month, or even Pandora Premium (which allows for on-demand listening à la Spotify) for $9.99 per month,[211] shockingly few do: of Pandora’s 76 million active listeners, less than five million were paid subscribers.[212] Pandora set the market rate at “free” since its founding in 2000, thereby holding its competitors to the same price point and setting the market rate—and, importantly, consumer expectations—at an unsustainable level.

In the case of large quantities of “address unknown” NOIs filed with the Copyright Office, licensees who aren’t accruing unmatched P&Us may effectively buy themselves a “free pass” for whatever period of time it takes either for the Copyright Office to catch up on its backlog and get the content registered and online, or for the content owners themselves to cull through literally millions of notices in hopes of identifying their content. Whatever the efficient level of remuneration for these content owners might be, we might reasonably assume that it is greater than zero.

The increased volume of “address unknown” NOIs being filed also adds to the workload of an already-overburdened Copyright Office, which now has to sort through and post millions of NOIs.[213] That staff time could be put toward the processing of copyright registrations, and the preparation of comments and recommendations, for example. Likewise, content owners are burdened with searching through millions of filings in order to claim content that might belong to them. This is time taken away from the creation and distribution of artistic works. On the other hand, the filing of baseless takedown notices by content owners drains OSP resources that could be put to better use improving user interface and app functionality, for example.

In addition, copyright arbitrage can inefficiently affect the price of both the regulated product, and the economic substitute. If ReDigi’s business model were approved, for example, the company could expect considerable growth in its consumer base. Unlike a used book, a “used” MP3 is exactly as good as a “brand new” MP3, so a rational consumer would be expected to buy from ReDigi over iTunes. This spike in consumer demand might allow for several possibilities: first, ReDigi might raise its prices, so long as they remain lower than iTunes’s. Over time, this might allow for a competitor to enter, pricing itself just below ReDigi, and so on, until a market price is eventually established. Because many users are likely to retain their music, the digital copies sold are essentially costless to them, making the “market” price the average cost per track for the digital music reseller (plus any opportunity cost).

If all of this results in decreased sales for iTunes,[214] it can respond in one of several ways: it can lower its prices (and with them, its profit margin). This may lead Apple to seek to renegotiate with content owners over royalties. Alternately, iTunes may raise its prices to compensate for the loss in volume. Or it might lobby Congress to amend Section 109 to exclude digital products (in hopes of eliminating ReDigi as a competitor). The company might even decide to join ReDigi in the “digital resale” market. All of these scenarios lead to the same end: less money for copyright owners. The incentive-access paradigm contemplates a copyright regime that aims to encourage creation while balancing access for users. To the extent we believe this balance is optimally calibrated, ReDigi’s style of copyright arbitrage is not an effective means of accomplishing this balance. Even where we feel less confident about the current regime’s balance calibration, copyright arbitrage that works to reduce incentives to creators is unlikely to prove efficient.[215]

B. Anticompetitive Effects

By virtue of the fact that not all companies are able or willing to engage in it, copyright arbitrage has the potential to function in an anticompetitive capacity. Indeed, competitive advantage is often an end for which arbitrage is the means. For example, YouTube’s subscription music service competes head-on with Spotify’s, yet Spotify does not qualify for the safe harbor.[216] Instead, it has to engage in direct negotiation with record labels, resulting in an average streaming rate of $7 per 1,000 streams, or seven times more than the streaming rate of $1 per 1,000 streams paid by YouTube.[217] This differential is sometimes referred to as a “value gap.”[218]

The legislative history of the DMCA describes the purpose of the safe harbor as “encourag[ing] responsible behavior and protect[ing] important intellectual property rights.”[219] The value gap arguably reflects a disconnect between the safe harbor’s original intent and YouTube’s current utilization thereof. In other words, “[t]he safe harbor was intended to protect passive Internet platforms with no knowledge of what its users are doing, not active music distributors like YouTube.”[220]

In addition to securing a competitive advantage over competitors who do not qualify for the safe harbor, copyright arbitrage affords YouTube an unfair bargaining advantage vis-à-vis content owners. The Section 512 safe harbor effectively sets the baseline for licensing negotiations to “free.” The Recording Industry Association of America’s (RIAA) CEO and Chairman, Cary Sherman, described these negotiations in a recent interview:

The way the negotiation goes is something like this: “Look. This is all we can afford to pay you,” YouTube says. “We hope that you’ll find that reasonable. But that’s the best we can do. And if you don’t want to give us a license, okay. You know that your music is still going to be up on the service anyway. So send us notices, and we’ll take ‘em down as fast we can, and we know they’ll keep coming back up. We’ll do what we can. It’s your decision as to whether you want to take our deal, or whether you just want to keep sending us takedown notices.”

That’s not a real negotiation. That’s like saying, “That’s a real nice song you got there. Be a shame if anything happened to it. . . .”

That’s what it’s like to negotiate, when somebody can claim the benefit of an expansive safe harbor. They’re taking the benefit of a safe harbor that was intended for people who were passive, neutral intermediaries. People like Verizon, where the content is just passing through their system. They’re not making money off of distributing content. YouTube does.[221]

In other words, critics of YouTube’s utilization of the Section 512 safe harbor contend that “[i]t isn’t a level playing field . . . because ultimately you’re negotiating with a party who is going to have your content no matter what.”[222]

In its defense, YouTube points to its proprietary Content ID system, a voluntary agreement between YouTube and certain qualified content owners. Content ID works by identifying potentially infringing uploads by comparing them to a database of songs verified by participating copyright owners. Those owners can then elect to either take the infringing content down (or prevent its upload, as the case may be), or to “claim,” or monetize, the advertising revenues associated with it.[223] YouTube reports that most content owners opt to claim infringing content, rather than taking it down.[224] The result of claiming is a per stream rate based on YouTube’s going ad rates—rates nearly seven times lower than the per-stream rates paid by YouTube’s competitors.[225]

Alternately, YouTube points to the licensing deals that it has in place with some content owners. While the terms of these licensing deals are not publicly available, rightsholders have suggested that these agreements were reached “under very difficult circumstances.”[226] According to Warner Music’s CEO Steve Cooper, “There’s no getting around the fact that, even if YouTube doesn’t have licenses, our music will still be available but not monetized at all.”[227] Interestingly, elimination of the safe harbor might serve only to reverse the advantage by allowing content owners to demand above-market royalties against a backdrop of statutory damages for infringement.[228]

Furthermore, successful copyright arbitrage—i.e., arbitrage that results in cost savings while surviving superficial regulatory scrutiny—by one company puts pressure on competitors to employ the same practices, or risk being noncompetitive. If everyone was willing and able to do so, this would eventually erode the competitive advantage gained by the original company. In practice, however, this is usually not the case. The same characteristics that normally advantage larger companies—e.g., capitalization and diversification—also support these companies’ superior ability to engage in copyright arbitrage.

Small competitors of Pandora, for example, are unlikely to be able to afford to purchase a terrestrial radio station, leaving them to pay the higher royalty rates assigned to pure digital services. A new, upstart competitor to Spotify is less likely to have the staffing (or funding) required to prepare and file millions of “address unknown” NOIs, leaving it to forego a meaningful period of free content use, and to instead pay whatever rate it can negotiate directly with content owners (a rate necessarily higher than zero).

This may sound like any other competitive advantage afforded a larger company with greater market power. When viewed through the lens of copyright arbitrage, however, these are not merely examples of a larger, better-funded company having an advantage over a smaller newcomer. In a functioning market, the aspiring competitor need simply distinguish itself—perhaps through some innovation or other—in order to compete, regardless of size disparity. Unfortunately, most content markets do not work this way. The reason for this is two-fold: first, there is the existence of the mini-monopoly that copyright gives to the rights holder of a work. Second, today’s content industries operate largely as oligopolies. The inability to reach an agreement with even one music publisher, for example, is very high-stakes: it would deprive a prospective licensor of roughly one-third (possibly more) of the content that its competitors can offer.[229] Similarly, a first-mover advantage augmented by arbitrage can set barriers to entry that foreclose, or at least strongly discourage, competition. The network effects[230] common to copyright-related industries only serve to exacerbate this concern. To be sure, these concerns exist regardless of copyright arbitrage, but are invariably augmented by it.

C. Distributive Justice

Like all regulatory arbitrage, copyright arbitrage is a sophisticated activity. Companies who can afford better (paid) lawyers and financial advisors are more readily able to uncover and exploit arbitrage opportunities. These companies are also better able to deal with any potential blowback associated with the attempt. This enhanced aptitude for arbitrage serves to further widen the disparity between larger and smaller companies.

This is not merely a fairness critique. Prior work has shown how this disparity in the copyright context affects not only smaller firms, but also consumers and society at large.[231] For example, circumvention of Section 114 of the Copyright Act—the statutory license for sound recordings—includes circumvention of Section 114(g)(2), which mandates direct payment to artists.[232] This is an especially important entitlement for sessions musicians and backup vocalists who often rely upon this statutory payment as their sole compensation. Unlike monies collected under Section 114, monies collected under privately negotiated deals designed to circumvent this Section are not subject to this statutory distribution, and so are not required to make any payment directly to artists. The loss of this income could have a significant short-term impact on artists, as well as a long-term negative impact on creation and societal wellbeing.[233]

YouTube currently enjoys the lion’s share of all streaming consumers, yet it pays the least of all the streaming services. First and most obviously, this means less money for creators and intermediaries—in direct contravention of the incentivization goals of copyright. This reduction in income could have significant, lasting impact on creative output to the detriment of consumers. Second, it introduces the risk that if and when Congress turns to replacing the safe harbor with a statutory license, for example, the only “market rate” it will have to look to may be misrepresentative.[234] In addition, proliferation of copyright arbitrage among larger industry players will place smaller companies and new entrants at a marked disadvantage by raising the barriers to entry and potentially reducing consumer choice and access. Without competition from new entrants, the potential for collusive behavior among oligopolists increases.

D. Consumer Expectations

Copyright arbitrage’s impact on consumer expectations is subtle, yet dramatic. In the first instance, copyright arbitrage can artificially depress (or raise) licensing rates for some not insignificant period of time. During that time, licensors or licensees (as the case may be) often sustain a loss, or at least a substantive reduction in revenues, while the other party is able to either present as viable a business model that otherwise might not be (in the case of licensees), or enjoy supracompetitive rates (in the case of licensors).

During this time, consumer expectations are set. For example, ReDigi’s operation from October 2011 to March 2013[235] gives users the understandable impression that resale of their digital goods is legal.[236] Indeed, the company touted itself as “the world’s first pre-owned digital marketplace,” setting a “market rate” of $0.49 per track.[237] Over time, this can lead to the cultural normalization of reselling digital goods, a practice which may very well turn out to be illegal. Likewise, both YouTube and Pandora offer consumers free, all-you-can-eat music consumption. These ad-supported services—available to consumers since 2005[238] and 1999[239] respectively—may well turn out to be financially unsustainable without the benefit of arbitrage.

The real problem with all of this is that it is very hard to put the genie back into the bottle. File-sharing platforms like Napster, and P2P platforms like Kazaa, normalized access to individual songs anytime, anywhere. The eventual condemnation and dissolution of these platforms did little to dampen consumer appetite for the product, however. The resultant consumer behaviors and expectations are often credited, for example, as the impetus behind the development of streaming technologies.[240] Some hail this as an exemplar of disruptive innovation;[241] others lament the permanent reduction in valuation experienced by the music industry.[242] This Article assumes both observations are true, but laments that to the extent streaming is propped up by arbitrage, it may only prove sustainable through continued arbitrage.

E. Correction, Information-Forcing & Signaling

The aforementioned implications assume problematic copyright arbitrage. As the examples show, however, even behavior that explicitly exploits a regulatory loophole may not necessarily result in a problem in need of a fix. It has been suggested that some sections of the Copyright Act might be intentionally vague,[243] such that some instances of copyright arbitrage might be valuable in correcting an imbalance, and/or in signaling an inefficiency to Congress.

As discussed in more detail in Part II, in the Pandora example, the company’s purchase of a terrestrial radio station for no business purpose other than rate manipulation arguably worked to challenge the extant terrestrial broadcaster oligopoly. Similarly, ReDigi’s behavior might be characterized as regulatory entrepreneurship, wherein they highlight Section 109’s digital shortcomings to Congress in hopes of expanding the Section to cover digital goods. Even a blatant example of technological arbitrage such as Aereo’s might be alternately explained as providing a service to an underserved population. In all of these cases, copyright arbitrage may be viewed as providing a social benefit.

IV. Mitigation

Where copyright arbitrage is deemed undesirable,[244] its reduction or elimination lies with one or more of the judiciary, the legislature, an administrative body, or some combination thereof. Currently, “no agency exists with comprehensive and independent rulemaking authority in the area of copyright law.”[245] This leaves the problem of copyright arbitrage, where there is one, to Congress and the courts, neither of which are institutionally well positioned to respond nimbly and accurately on their own.

This Section begins by explaining the inefficacy of the current copyright regime when it comes to regulatory arbitrage as due in large part to a lack of responsive and efficient rulemaking. The next three sections discuss the pros and cons of several options available for identifying, and curbing, problematic arbitrage behavior: Subsection (b) urges courts to take a purposive, substantive approach in their interpretations of the Copyright Act. Subsection (c) calls for recognition of copyright as a complex regulatory system in need of a strong regulatory apparatus, and proposes that Congress empower a regulatory agency with rulemaking authority. It also considers what such a system might look like in practice and how it could work to mitigate the negative impact of copyright arbitrage. Finally, Subsection (d) acknowledges antitrust law as a backup for curbing copyright arbitrage resulting from legislative capture.

A. Legislative Response

Because copyright arbitrage is effectively noncompliance with the law, Congress is arguably in a position to curb it through legislation. Such legislation is a clean slate: copyright law does not currently have any explicit rules aimed at curbing regulatory arbitrage. As this Section demonstrates, however, Congress may not always be the ideal forum for correcting copyright arbitrage: first, Congress lacks the subject matter expertise and perspective necessary to identify and effectively mitigate regulatory manipulations in the copyright space. Second, the legislative process may be too slow and too cumbersome to respond effectively in the face of constantly evolving arbitrage efforts. Finally, but no less importantly, Congress is prone to legislative capture, especially in the copyright context.

  1. Subject Matter Expertise

When faced with a regulation, a regulated entity can either (i) comply; (ii) disobey and pay the penalty (monetary or otherwise); (iii) challenge the regulation (in court or via lobbyist); or (iv) exploit a loophole/engage in arbitrage. Economic theory tells us that a rational entity will pick whichever option is least costly. So, when an entity engages in arbitrage, we can assume it has been deemed the least costly option. In order to funnel entities away from problematic arbitrage, then, Congress needs to make arbitrage more costly than the other options.

This analysis need not be done on a case-by-case basis, but likely should be conducted on a section-by-section basis. For example, if abuses of Section 115’s NOI requirement are a concern, artists might be allowed to opt out from, or abusive entities might be suspended from use of, the compulsory license.

An opt-out option is suboptimal because it undermines the transaction cost savings contemplated by the compulsory license. Indeed, the very basis for a liability rule—of which the compulsory license is an example—is to minimize transaction costs[246] and to allow for more efficient bulk licensing of content and scalability of service: “A compulsory licensing scheme represents a real shortcut: it eliminates the need for private institution-building—a costly and time-consuming process.”[247]

On the other hand, suspension of abusive entities is attractive for several reasons. First, the compulsory license as currently drafted does not require a prospective licensee to submit to a credit check nor to otherwise make a showing of creditworthiness. This leaves content owners obligated to do business with a partner who may or may not hold up their end of the bargain (i.e., comply with the statutory requirements, and pay the statutory rate). At present, there is no penalty (aside from an unlikely lawsuit) for failing to comply with Section 115’s explicit requirements (i.e., failure to provide notice of intent and/or failure to pay), much less for abuses of the license. A copyright owner seeking to curb such abuses has to engage an attorney and file a lawsuit for copyright infringement. This places the burden on the nonoffending party, which is often not in a position to carry it.[248]

The preceding analysis requires: first, a detailed understanding of NOI practice under Section 115; and second, an appreciation for the power dynamic between licensors and licensees under the compulsory license. Congress does not have, and is not likely to attain, either of these sensibilities.

Since the drafting of the 1909 Copyright Act, copyright has been understood as:

[S]o complicated most members of Congress cannot understand it . . . Copyright legislation . . . has never been accorded the congressional staff or resources available for legislation on politically sensitive issues like tax or military appropriations. It may be that the impression that members of Congress cannot or will not spare copyright sufficient time to gain a thorough understanding has been a self-fulfilling one.[249]

At the time of the Copyright Act’s contemplation in Congress, for example, Representative Paul C. Jones of Missouri remarked: “[T]he bill is so full of things that people do not understand about it, that cannot be understood, it should not be passed. . . . I have talked to members of the Committee on the Judiciary who admit they do not know what is in it.”[250]

As discussed further at Part IV.C, Congress could instead empower an administrative agency to publish a biannual or quarterly list of “deadbeat” licensees whose access to Section 115’s compulsory license has been suspended for a period of, say, six months. This would at least slow the deluge of “address unknown” NOIs being filed by services like Amazon and Google, thereby allowing the Copyright Office a chance to catch its breath, while forcing these licensees to reconsider their approach to content licensing.

  1. Timeliness

The speed with which technological innovations make their way to market and present arbitrage opportunities requires a quick and accurate fix. The traditional, time-intensive legislative process with its drawn-out hearings and solicitation of multiple reports and comments, does not lend itself to real-time correction of problematic arbitrage, especially not in the copyright context. Where copyright arbitrage is deemed desirable, on the other hand, Congress may be the perfect venue. Its slow, cumbersome processes allow time for the socially beneficial arbitrage to do its work.

Indeed, the Copyright Act that we currently operate under was just over twenty years in the making. This included negotiations between industry representatives, “repeated, lengthy subcommittee hearings,” “numerous executive sessions,” and “a flood of committee reports.”[251] Likewise, the DMCA, added in 1998 in hopes of bringing copyright law into the new, technological age, was effectively obsolete by the time it went into effect. A more effective and streamlined process might instead see Congress empower an administrative body with rulemaking authority to correct abuses as they are uncovered in real time.

  1. Capture

Lack of subject matter expertise and timeliness are not the only aspects of the legislative process that counsel against Congress as the optimal venue for curbing copyright arbitrage; the process by which copyright legislation has traditionally been drafted is also potentially problematic:

[M]ost of the statutory language [of the 1976 Copyright Act] was not drafted by members of Congress or their staffs at all. Instead, the language evolved through a process of negotiation among authors, publishers, and other parties with economic interests in the property rights the statute defines. In some cases, affected parties agreed upon language, which was then adopted by Congress, while disagreeing about what the language meant.[252]

While the involvement of stakeholders can compensate for Congress’s lack of subject matter expertise, Congress is notoriously prone to legislative capture, particularly in the copyright context. To quote Representative Robert Kastenmeier of Wisconsin, speaking of the current Copyright Act: “Really, all interests in this bill are, in one form or another, special interests.”[253] In her work on copyright reform, Jessica Litman recognizes “the political power of copyright lobbies, aided by members of Congress eager to be glamoured by famous entertainers and willing to be persuaded that the only fundamental problem with the United States economy is widespread piracy of American creations,” and so suggests a redistribution of rights from powerful intermediaries to creators.[254]

One powerful example of legislative capture in copyright is the CTEA. That legislation, spearheaded by Walt Disney, extended copyright’s term by twenty years. Of the act, Herbert Hovenkamp wrote:

It is hard to come up with any serious argument that retroactive extensions of old copyrights serve the constitutional purpose of promoting the progress of the useful arts. Those inventions and ideas have already been created. The Copyright Term Extension Act shows us Congress at its worst, passing legislation at the behest of powerful interest groups at society’s expense.[255]

William Patry, former counsel to the Subcommittee on Intellectual Property described the process this way:

Copyright interest groups hold fund raisers [sic] for members of Congress, write campaign songs, invite members of Congress (and their staff) to private movie screenings or sold-out concerts, and draft legislation they expect Congress to pass without any changes. In the 104th Congress, they are drafting the committee reports and haggling among themselves about what needs to be in the report. In my experience, some copyright lawyers and lobbyists actually resent members of Congress and staff interfering with what they view as their legislation and their committee report. With the 104th Congress we have, I believe, reached a point where legislative history must be ignored because not even the hands of congressional staff have touched committee reports.[256]

In some areas of the law, regulatory loopholes allowing for arbitrage might even be the intentional result of legislation championed by special interest groups who support it only because they deem it sufficiently weak or ambiguous. An example might be the recently enacted tax bill, with its various opportunities for pass-through income that is not scored by the Congressional Budget Office.[257] To the extent that this practice may carry over to the copyright context, the very lobbying that enables it makes Congress poorly positioned to curb it.

The nature of the legislative process as described herein, and its inability to keep up with technological advancements, is unlikely to change. Instead, where copyright arbitrage is deemed problematic, its negative impact might best be mitigated by a combination of (i) a more proactive judiciary, coupled with (ii) an empowered administrative body, and (iii) backed up by antitrust law.

A. Judicial Response

In the absence of congressional or administrative action directed at curbing problematic copyright arbitrage, courts stand as the first line of defense. We have seen judicial efforts at curbing potentially abusive practices, for example, in the ReDigi case.[258] Recall that the arbitrage at play in ReDigi was the potential misapplication of Section 109’s first sale doctrine to digital resale. Section 109’s application to digital goods is unclear, not least of all because it was drafted and adopted before the existence of digital content.

The district court in ReDigi temporarily halted the potential misuse of Section 109 without making a determination of its application to digital goods one way or the other. It did so by focusing on the requirement in Section 109 that the good being sold be “lawfully made under this title,” and found the copy necessarily made by ReDigi’s Media Manager software to be unlawful.[259] In so doing, it enjoined the arbitrage without ruling on its legality. This brand of indirect judicial response reserves to Congress the final interpretation and determination of the copyright laws, while reducing the damage arbitrage might bring about in the meantime. In other words, a temporary fix.

Historically, copyright jurisprudence—particularly from the Supreme Court—has focused heavily, if not explicitly, on the equity of copyright doctrine to justify lawmaking in this area.[260] Aereo is a recent example of this purposive, rather than overly literal, interpretative approach: there, the majority took the broad view in determining that since Aereo aimed to provide the same type of service that cable companies do, it should be treated like one (i.e., it should be found to perform publicly).[261] Rather than leave it to Congress to plug the loopholes, courts could adopt the Aereo majority’s interpretative approach. To be sure, purposive judicial decisions may fail to account for industry changes, or may simply get it wrong. When combined with the legislative and administrative suggestions in this Section, some of these concerns may be at least partially ameliorated.

There are other challenges to this suggestion too: to be sure, the judiciary could engage more purposively in cases of copyright arbitrage, but it often doesn’t. Perhaps this is because—as Part V.C details—copyright has become an increasingly regulatory body of law. Perhaps judges recognize that in many cases they are not institutionally well positioned to see the field in order to make adequate determinations. Or perhaps the difficulty in distinguishing “use” from “misuse” when it comes to the copyright statute has courts hesitant to overstep: “rather than resorting to open-ended considerations of utility or social welfare, to resolve indeterminate doctrinal puzzles, the Court’s copyright jurisprudence might [ ] be seen as consciously mediating extra-legal normative considerations through doctrinal ideas.”[262]

Unlike the other two branches of intellectual property law—patent and trademark—copyright does not currently have a specialized tribunal. The authorization of the U.S. Court of Appeals for the Federal Circuit as the only appellate-level federal court with the authority to hear patent appeals and administrative trademark decisions afforded the opportunity for that tribunal to develop an expertise in the subject matter, allowing it to handle complicated, highly technical disputes.[263] A tribunal specializing in copyright might afford similar benefits.

In 2017, a bill was introduced that would establish a specialized copyright tribunal. Should some version of the Copyright Alternatives in Small-Claims Enforcement (CASE) Act of 2017 pass,[264] we might see a more active role for the judiciary in copyright generally and in copyright arbitrage in particular. In the meantime, “we have seen a definite turn in IP enforcement away from common law adjudication of disputes and towards a regulatory state . . . .”[265]

The move away from a common law copyright regime toward an increasingly regulatory one has been recognized in the literature:

For many years, U.S. copyright law was based largely on a judicially administered, industry-neutral property rights regime. Congress was responsible for setting the property entitlement. The courts were responsible for defining and enforcing the entitlement. And the markets and private institutions were responsible for organizing the production of creative works in light of the property rights structure. The Copyright Office’s role was primarily ministerial, registering and tracking ownership of copyrighted works.

In recent years, however, Congress has been much more willing to intervene in the structure of copyright markets. The 1976 Act departed from the pure property rights view by introducing detailed, industry-specific exemptions and several complex compulsory licenses for certain industries. The Librarian of Congress was, for the first time, charged not only with registering copyrights, but also setting licensing rates, albeit in only a few industries. Since the 1976 Act, amendments to the Act have become increasingly more detailed and industry-specific, relying more on compulsory licenses and, in some cases, mandating adoption of certain technologies and banning others. The Librarian of Congress’s duties have similarly expanded beyond mere registration, encompassing not only ratemaking but also substantive rulemaking. Recently proposed legislation, as well as academic proposals for significantly revamping the copyright system, also exhibit similar qualities. The trend is such that this mode of “regulatory copyright” is now the dominant mode of copyright lawmaking.[266]

The next Section acknowledges the debate over copyright’s status as a property right versus a regulatory regime, and suggests that the existence of copyright arbitrage supports recognition of copyright as a regulatory regime in need of a regulatory apparatus.

C.  Administrative Response: A Proposal

Copyright’s status as either a property right or as a regulatory regime has been extensively debated in the literature.[267] The significance of copyright’s status as stemming from property law or from legislation is that the former is superintended by the courts, while the latter can be best dealt with through an administrative agency.

  1. Copyright as a Regulatory Regime

There is merit to treating copyrights like property rights: free market advocates, for example, would argue that it’s the only way to induce private ordering and with it, efficiency.[268] But even the 1909 Act (predecessor to the Copyright Act that we operate under today) had moved away from a property right-centric view with the introduction of compulsory licenses and industry-specific exemptions.[269] The 1976 Act took copyright still further in a regulatory direction by charging the Librarian of Congress with rate-setting responsibility (a task currently delegated to the CRB).

The last couple of decades in particular have seen a marked increase in the regulatory nature of copyright.[270] As Mark Lemley has observed, “Large swathes of the Copyright Act really are regulation: regulating price, setting compulsory licenses, determining what can be done, etc.”[271] Tim Wu has described copyright as:

[C]omprised of both authorship and communication regimes with often independent functions. The first regime is the familiar system, run by the courts, that grants exclusive [property] rights to encourage creativity. The second is a messier regulatory regime comprised mainly of the sections of copyright that have always perplexed copyright theorists and have never fit the central theme of author-incentives.[272]

In practice, of course, copyright is both, although arguably it does neither very well: copyright simultaneously lacks the clean, common law lines of the Sherman Act and is riddled with heavily lobbied niche rules and ratemaking. As we are unlikely at this point to wipe the slate clean and start over, this Article suggests that the increasingly regulatory nature of copyright—and especially the emergence of copyright arbitrage—strongly supports recognizing copyright as a regulatory regime in need of a regulatory apparatus.

While this move toward a regulatory regime allows for a great degree of tailoring and accommodation, it comes at the cost of flexibility and nimbleness. As Part IV.A demonstrates, Congress’s process may be too cumbersome to keep up with changes in technology and business models, much less to curb problematic arbitrage. Writing about the current Copyright Act, Melville Nimmer observed that “[w]here previously the statute had too little to say in many vital copyright areas, it may now be argued that it says too much. . . . [T]he flexibility and pristine simplicity of a corpus of judge-made copyright law . . . has now been replaced with a body of detailed rules reminiscent of the Internal Revenue Code.”[273]

As such, Congress should take a note from other areas of the law that suffer from regulatory arbitrage—such as tax—and empower an administrative body with rulemaking authority comparable to that of the IRS.

  1. Call for a Regulatory Apparatus

There are a few possible approaches to establishing an effective regulatory apparatus for copyright. These include a single-administrator approach, a full-service agency, or some combination of the two.

The IRS follows the single-administrator model. Under this model, the commissioner is a presidential appointee, employed at-will, under the oversight of the White House. This allows the commissioner to clarify extant statutes and to function as an enforcement arm. This model would work well in the Pandora example, where the statute just requires clarification of, say, what proportion of revenues must come from each source to qualify for the lower rate.

The FCC is an example of a full-service agency with multiple members and greater insulation from White House oversight. This model allows the agency to identify new breaches as they happen and to impose penalties. This model might be preferable in a case like ReDigi’s, where greater interpretative authority would be useful.

Owing to copyright’s current organizational structure, both of these models face some challenges in the copyright context. Currently, the Copyright Office sits within the Library of Congress, and its Register is appointed by the Librarian of Congress, herself a presidential appointee.[274] As such, the Copyright Office does not have independent rulemaking authority; instead, the Register can merely recommend regulations to the Librarian of Congress.[275] Without clear authority and a means of enforcing its guidance, any arbitrage mitigation attempts undertaken by the Copyright Office could lead to less, not greater, clarity.

Simply granting the Copyright Office rulemaking authority may not be sufficient to resolve this concern, however. There is a unique history of constitutional challenge to the Librarian’s appointment of officials tasked with administering the copyright laws. In a D.C. Circuit case from 2012, a disgruntled licensor challenged the constitutionality of the Librarian’s appointment of CRB judges as violating the Appointments Clause.[276] That Clause requires all principal officers of the United States be appointed by the President and confirmed by the Senate.[277] In that case, the court initially found that the CRB judges were indeed acting like principal officers, in violation of the Appointments Clause. To correct this constitutional breach, the court amended the statute to clarify that CRB judges could be appointed and dismissed at will by the Librarian, making them inferior officers and thereby removing the conflict.

Because the CRB is a department within the Library of Congress, as the Copyright Office is, and because the Librarian of Congress also appoints the Register of Copyrights (without presidential nomination nor Senate confirmation), the concern is that similar allegations could be brought against the Register’s authority by someone unsatisfied with a particular rulemaking or enforcement procedure. While it is beyond the scope of this Article to resolve the Article II challenge, one idea might be to move the Copyright Office to the Department of Commerce, where the United States Patent and Trademark Office sits. In this way, the Register of Copyrights could be presidentially appointed and senatorially confirmed. Bureaucratic difficulties aside, this arrangement may have the added benefit of more consistent treatment of all three branches of intellectual property.

Alternately, as suggested by Sandra Aistars in her work on reformation of the Copyright Office, creation of a separate administrative body focused solely on copyright may be the option most likely to resolve the constitutional question definitively.[278] Unfortunately, this administrative body would face some of the same challenges as the Copyright Office—namely, undue influence and capture.[279] As there is little reason to believe the extant Copyright Office would be less beholden to industry interests if given rulemaking authority, establishing a newly constituted regulatory body may, at the very least, improve credibility. Drawing from examples in tax law, the next Section details what such a regulatory body might look like, and how it might operate in practice.

  1. Application

A look into the practices of the IRS, the largest bureau of the U.S. Treasury, is instructive in the ways that a newly created administrative agency might address copyright arbitrage. The IRS has established several approaches to combat tax arbitrage: the form versus substance doctrine, the business purpose doctrine, the economic substance test (also known as the “sham transaction doctrine”), and the step transaction doctrine.[280] The form versus substance doctrine allows the IRS to reclassify a transaction according to its substance, regardless of the form that it takes. The business purpose doctrine allows the IRS to challenge and disallow claimed business expenses for which it finds no legitimate business purpose (such as a nonexistent home office). A transaction satisfies the economic substance test if it has both substantive economic effects and a purpose other than reducing or avoiding tax. Finally, the step transaction test enables the IRS to collapse a series of separate transactions into one single transaction for the purposes of determining the proper tax consequences.[281]

Each of these doctrines involves an administrative agency redefining a transaction to bring it into line with regulatory intent. Extrapolating into the broader legal context, Leo Katz posits a common theory underlying each of these doctrines: the so-called “mismatch theory.”[282] According to the mismatch theory, the assumption made by these doctrines is that legal rules are unavoidably imperfect, and therefore “doomed to be over- and underinclusive with respect to their true rationales.”[283] Katz identifies two types of mismatches: limited foresight—the notion that “it is impossible for the rule maker to anticipate all situations in which the rule might have to be applied in the future”—and limited hindsight—the notion that “even where one is able to anticipate such situations in theory, it might be hard to determine when one is actually encountering them in practice.”[284] Ultimately, Katz agrees with the conventional view that mismatches are “largely irremediable,”[285] albeit for different reasons, which leads to the question: What do we do about them?

Like the IRS, the newly created administrative body for copyright might devise a system of review that allows for undesirable arbitrage transactions to be redefined and reclassified, and so brought back under the regulatory umbrella. Pandora’s purchase of the South Dakota radio station, for example, would likely fail under a copyright version of the sham transaction doctrine.[286]

In the tax context, the Treasury has also established both formal and informal procedures specifically addressing the arbitrage problem. One such formal mechanism is the annual “General Explanations of the Administration’s Revenue Proposals,” commonly known as the “Greenbook.”[287] Released each fiscal year, these reports from the Treasury to Congress detail the agency’s proposals relating to various tax issues, including transactions identified as arbitrage. The Greenbook for FY2017, for example, includes, among others, the subheading “Other Business Revenue Changes and Loophole Closers.”[288]

One recommendation under that subheading is titled “Conform Corporate Ownership Standards.”[289] The recommendation aims to “conform the control test under section 368 with the affiliation test under section 1504.”[290] The reasons given for this proposed change are instructive:

By carefully allocating voting power among the shares of a corporation, taxpayers can manipulate the control test in order to qualify or not qualify, as desired, a transaction as tax-free (for example, a transaction could be structured to avoid tax-free treatment to recognize a loss). In addition, the absence of a value component allows corporations to retain control of a corporation but to “sell” a significant amount of the value of the corporation tax-free. Congress amended the affiliation test in 1984 to address similar concerns regarding the manipulation of the vote and value of affiliated corporations. A uniform ownership test for corporate transactions will also reduce complexity currently caused by these inconsistent tests.[291]

It is easy to imagine a comparable section in a “Copyright Greenbook” intended to close the exploitation of Section 115’s “address unknown” NOI requirement:

By utilizing Section 115’s “address unknown” NOI for content whose registration has either not yet been processed, or not yet digitized, by the Copyright Office, prospective licensees can temporarily use that content without payment. In addition, the existence of this option reduces the bargaining power of content owners, who find themselves competing with free. Congress established the Section 115 license as a means of reducing transaction costs associated with the production of cover songs, and to guarantee payment to owners of the copyright on the musical composition in digital transmissions. A good faith efforts test for use of the “address unknown” NOI will foreclose misuse of the notice, while retaining the feature for its intended use.

In the category of less formal administrative action, the IRS also maintains a running list of “Recognized Abusive and Listed Transactions” on its public website.[292] This allows for more expedient recognition and foreclosure of abusive arbitrage transactions than the more formal annual report. While the listing of a specific transaction doesn’t make it illegal per se, listed transactions require the filing of special disclosures that effectively shut those practices down.

For example, a recently listed tax arbitrage transaction—“Revenue Ruling 2003-6, Abuses Associated with S Corp ESOPs”—reads, in relevant part, as follows:

The Internal Revenue Service and the Treasury Department understand that certain arrangements involving employee stock ownership plans (ESOPs) that hold employer securities in an S corporation are being used for the purpose of claiming eligibility for the delayed effective date of § 409(p) of the Internal Revenue Code . . . . This revenue ruling alerts taxpayers and their representatives that the tax benefits purportedly generated by these transactions are not allowable for federal income tax purposes. This revenue ruling also alerts taxpayers, their representatives, and organizers or sellers of these transactions to certain responsibilities that may arise from participating in these transactions. . . . Transactions that are the same as, or substantially similar to, the transaction described in this revenue ruling are identified as “listed transactions” . . . transactions that are the same as, or substantially similar to, the transaction described in this revenue ruling may already be subject to the disclosure requirements of § 6011, the tax shelter registration requirements of § 6111 or the list maintenance requirements of § 6112 . . . . Persons who are required to satisfy the registration requirement of § 6111 with respect to the transaction described in this revenue ruling and who fail to do so may be subject to the penalty under § 6707(a). . . . In addition, the Service may impose penalties on participants in this transaction or substantially similar transactions, or, as applicable, on persons who participate in the reporting of this transaction or substantially similar transactions, including the accuracy-related penalty under § 6662, and the return preparer penalty under § 6694.[293]

The listing of certain S Corp ESOP transactions does not prohibit them altogether—though it may flag them for Congress to do so—but it does single them out for special reporting and disclosure requirements. This effectively raises the cost of engaging in arbitrage: “It is bad news if a taxpayer’s transaction is on the list because it automatically triggers the registration and disclosure requirements, probably IRS examination, costly litigation, and a material risk that additional taxes and perhaps penalties will be owed.”[294] Listing abusive transactions can also work to clarify which types of transactions are acceptable: “The good news . . . [is that] if the transaction is not listed, the tax shelter investor may be in better shape.”[295] While not without its downsides, affording enforcement authority to the new regulatory agency might lend greater weight to such designations.

A comparable listing of potentially abusive copyright transactions might likewise discourage copyright arbitrage, while working to improve predictability around legitimate transactions. For example, the administrative copyright authority might list “Use of Section 109 for Digital Goods” to ward off business models like ReDigi’s, unless and until Congress acts to clarify application of the first sale doctrine to digital information goods:

This agency understands that certain platforms claiming to establish a marketplace for the purchase and resale of digital content rely upon Section 109 of the Copyright Act. This listing alerts platform owners that Section 109’s exemption from copyright infringement for first sale does not apply to digital goods at this time. Transactions that are the same as, or substantially similar to, the transaction described in this listing may be subject to the following disclosure requirements. . . . In addition, the Copyright Office and/or the Librarian of Congress, may impose penalties on participants in this transaction or substantially similar transactions . . . .

This guidance would confirm digital sales as outside the scope of Section 109 (unless and until Congress acts to clarify otherwise), and would also lend predictability to reliance on Section 109 for nondigital sales. While private litigants may challenge these interpretations, administrative deference would still improve predictability.

While the aforementioned administrative responses are a good starting point, they are not foolproof. If drafted too narrowly, parties may simply work around the listed transaction descriptions just as they do with existing regulations. Such work-arounds may cost them a bit more than the original arbitrage transaction, but still less than full compliance, and so they may still engage, albeit less efficiently and with more waste. In the tax context, this type of gamesmanship has been referred to as the “Tax Avoidance Game.”[296] Assuming the propensity for gamesmanship can be overcome, or at least mitigated, an administrative response coupled with enforcement authority shows the most promise in mitigating the negative impact of copyright arbitrage. This is especially true where antitrust law serves as a backdrop for curbing instances of copyright arbitrage occasioned by legislative capture, as described in the next section.

D. Antitrust

A final option for responding to problematic copyright arbitrage bears mention: antitrust. Whereas regulation may or may not promote competition—especially in cases of industry capture, as is often cited in copyright—antitrust law can work to curb the anticompetitive effects of copyright arbitrage. In their work on antitrust and regulation, Stacey Dogan and Mark Lemley define regulatory gaming as “private behavior that harnesses procompetitive or neutral regulations and uses them for exclusionary purposes.”[297] Noting that “the existence of regulation can sometimes exacerbate, rather than alleviate, the risk of exclusionary behavior in regulated markets,” they use the existence of regulatory gamesmanship to support the need for ongoing antitrust oversight in heavily regulated markets.[298]

Prior work has detailed the myriad challenges for antitrust in the copyright context.[299] In order to operate effectively, antitrust requires a workably competitive market, something arguably missing from the copyright industries. The extent of regulation in copyright further serves to undermine antitrust by insulating incumbent firms from antitrust review.[300] The IP context is particularly challenging for antitrust insofar as it grants a limited “monopoly” to copyrighted works. While these legally sanctioned monopolies are traditionally justified on a social utility basis,[301] they still make effective application of the antitrust laws difficult. Antitrust, then, can serve as a helpful addition to—but not a replacement for—the judicial and administrative responses detailed herein.


Rapidly developing technologies and changing consumer preferences have led to a heretofore unexplored phenomenon in copyright law: regulatory manipulation aimed at reducing regulatory costs via manipulation of regulatory treatment. Calling these manipulations what they are—copyright arbitrage—has real value, as it directs attention to their economic and societal impact. In a functioning market, these issues might best be dealt with by doing nothing. Copyright is not such a market. Identifying the types of copyright arbitrage seen to date, and the conditions for and constraints thereon, sets the stage for lawmakers, administrators, and the courts to implement mitigation strategies in those cases where the behavior is deemed problematic. The move away from common law and toward a regulatory regime suggests a three-pronged approach to curbing the proliferation of problematic copyright arbitrage: a more proactive, purposive judiciary, coupled with a regulatory body authorized to engage in rulemaking, and backed up by antitrust law.

This recommendation holds even when we recognize regulatory arbitrage as potentially encouraging socially beneficial experimentation. We can imagine, for example, a world in which ReDigi exposes the fact that application of the first sale doctrine to MP3s has little to no impact on the production or sale of digital tracks. A regulatory regime with rulemaking authority would not necessarily prohibit the development of such a business model, but rather could work to bring it into line with statutory protections.

To be clear, this conclusion assumes rules that best serve copyright’s delicate incentive-access balance. To the extent that copyright arbitrage may in fact be reacting to an imbalance, a rulemaking body would still be best positioned to consider a global amendment to the rules. Similarly, while mitigation of copyright arbitrage won’t eliminate all of the anticompetitive and antidistributive effects seen in the content industries, it can work to reduce the potential for negative impact.

This Article’s call for the establishment of a stronger regulatory apparatus in copyright also speaks to a more general question: When should Congress delegate its rulemaking authority to administrative agencies? The conditions that allow for copyright arbitrage suggest a few criteria that may support delegation: first, an administrative body with rulemaking authority makes sense in areas of the law that especially benefit from specialization and in-depth context, as well as industry-specific knowledge. Delegation may also be indicated in fields with very active lobbies, and the concomitant vulnerability to legislative capture. Finally, the call for delegating rulemaking authority may be particularly acute in areas of the law in which the degree of statutory intricacy renders judicial intervention at best challenging, and at worse, ineffective. As an exemplar of each of these criteria, copyright is ripe for a stronger regulatory apparatus.

Kristelia A. García: Associate Professor at the University of Colorado Law School and Director of the Content Initiative at the Silicon Flatirons Center for Law, Technology and Entrepreneurship. For their generous and helpful comments, the author thanks Michael Abramowicz, Eric Alston, Ian Ayres, Robert Brauneis, Ben Depoorter, Eric Gerding, Jim Gibson, Mark Lemley, Yvette Liebesman, Jessica Litman, Barak Orbach, Gideon Parchomovsky, Blake Reid, Matthew Sag, Andrew Schwartz, Sloan Speck, Harry Surden, Rebecca Tushnet, Phil Weiser, and the participants of the St. Louis University School of Law Faculty Workshop, Yale Law School’s ISP Speaker Series, and the University of Pennsylvania’s Third Annual Copyright Roundtable. For their generous support, the author thanks the Honorable Nancy F. Atlas IP Inn of Court and the Institute for Intellectual Property & Information Law (IPIL). A special thanks to Rachael Smith, and to Jane Thompson and the Colorado Law Library team for their excellent research assistance, and to Daniel Yablon, Martha Reiser, and the editors at the California Law Review without whom this Article would be far less readable. Accolades and compliments, and criticism couched as such, are welcome at


Copyright © 2019 California Law Review, Inc. California Law Review, Inc. (CLR) is a California nonprofit corporation. CLR and the authors are solely responsible for the content of their publications.

          [1].     Tim Ingham, Revealed: The 186 Artists Fighting the Youtube-Shielding DMCA, Music Bus. Worldwide (June 21, 2016), [].

          [2].     Eriq Gardner, Spotify Hit with Two Lawsuits Claiming “Staggering” Copyright Infringement, Hollywood Rep. (July 18, 2017), [].

          [3].     17 U.S.C. § 115 (2012).

          [4].     Perhaps no other area of law is as closely associated with regulatory manipulation as tax. Abuses are so prevalent that the Internal Revenue Service maintains a running list of so-called “Recognized Abusive and Listed Transactions.” Recognized Abusive and Listed Transactions, Internal Revenue Serv., []; see also infra Part IV.D. Some examples from this list include offshore deferred compensation arrangements, sale-in lease-out transactions, and basket option contracts.

          [5].     Historically, the most common regulatory manipulation in immigration law was something known as “bootstrapping,” wherein a foreign national comes to the United States on a visitor’s visa and  promptly commits some act of political activism—distribution of literature, say—that would make him a target in his home country, thereby enabling him to claim political asylum. See, e.g., Leo Katz, Why the Law Is So Perverse 76, 114 (2011).

          [6].     The classic example of regulatory manipulation in bankruptcy involves the shifting of one’s assets to protected categories, such as a home.

          [7].     Rather than walk up to her arch nemesis with a gun, a would-be murderer might instead intentionally provoke her foe into attacking first, so that she can later claim self-defense.

          [8].     Victor Fleischer, Regulatory Arbitrage, 89 Tex. L. Rev. 227, 229 (2010).

          [9].     Frank Partnoy, Financial Derivatives and the Costs of Regulatory Arbitrage, 22 J. Corp. L. 211, 227 (1997).

        [10].     Erik F. Gerding, Law, Bubbles, and Financial Regulation 236 (2014).

        [11].     The term, intended to connote both evasion and avoidance, was first coined academically by Leo Katz. Leo Katz, Ill-Gotten Gains: Evasion, Blackmail, Fraud, and Kindred Puzzles of the Law 4 (1996). In the same year, an episode of the popular TV series The Simpsons featured the phrase in reference to Krusty the Klown’s “massive tax avoision.” The Simpsons: Bart the Fink (Fox television broadcast Feb. 11, 1996). It is beyond the scope of this Article to determine who copied whom.

        [12].     Ronald Turner, Reactions of the Regulated: A Federal Labor Law Example, 17 Lab. Law. 479, 479 (2002) (discussing avoision in the labor law context).

        [13].     Subsection 1(e) of the Copyright Act of 1909 was the predecessor to the current Act’s Section 106(4). For more on the story of the jukebox, see H.R. Rep. No. 2222, 60th Cong., 2d Sess. 7–9 (1909), and Jessica Litman, War Stories, 20 Cardozo Arts & Ent. L.J. 337, 350–54 (2002).

        [14].     See, e.g., Dan L. Burk, Perverse Innovation, 58 Wm. & Mary L. Rev. 1 (2016) (describing how technological regulation can lead to innovation designed to avoid the regulation that prompted it); Tim Wu, When Code Isn’t Law, 89 Va. L. Rev. 679 (2003) (discussing code as a mechanism of technological avoidance).

        [15].     Wu, supra note 14, at 726–45.

        [16].     Burk, supra note 14, at 3–5.

        [17].     See, e.g., Mark A. Lemley, Is the Sky Falling on the Content Industries?, 9 J. Telecomm. & High Tech. L. 125 (2011) (reviewing a series of technological innovations and arguing that concerns regarding negative impact on the copyright industry are often overblown).

        [18].     See id.

        [19].     Ian Ayres & Amy Kapczynski, Innovation Sticks: The Limited Case for Penalizing Failures to Innovate, 82 U. Chi. L. Rev. 1781, 1812–27 (2015).

        [20].     I.e., the “incentive-access” paradigm, or the inherent tension between incentivizing creators to produce new works on the one hand, and ensuring consumer access to those works on the other. See generally, e.g., Glynn S. Lunney, Jr., Reexamining Copyright’s Incentive-Access Paradigm, 49 Vand. L. Rev. 483 (1996) (discussing the paradox created by the paradigm’s attempt at balance).

        [21].     The incentive theory of copyright says that society encourages the production of creative works by offering protections designed to maximize financial rewards for creators. Harper & Row Publishers, Inc. v. Nation Enter., 471 U.S. 539, 558 (1985) (“By establishing a marketable right to the use of one’s expression, copyright supplies the economic incentive to create and disseminate ideas.”).

        [22].     See infra Part II.A.

        [23].     U.S. Const. art. I, § 8, cl. 8.

        [24].     See, e.g., Stephen Breyer, The Uneasy Case for Copyright: A Study of Copyright in Books, Photocopies, and Computer Programs, 84 Harv. L. Rev. 281, 350 (1970) (“The period of copyright protection is at present too long and should not be extended beyond fifty-six years.”); Brief of George A. Akerlof et al., Eldred v. Ashcroft, 537 U.S. 186 (2003) (No. 01-618), 2002 WL 1041846, at *3 (“Taken as a whole, it is highly unlikely that the economic benefits from copyright extension under the CTEA outweigh the additional costs.”). But see, e.g., Brief of Amicus Curiae American Intellectual Property Law Association in Support of Respondents, Eldred v. Ashcroft, 537 U.S. 186 (2003) (No. 01-618), 2002 WL 1822117, at *16–17 (“[E]xtending the term of existing copyrights makes it easier for copyright holders, and other creators, to pursue opportunities to further develop, disseminate, and exploit existing works.”).

        [25].     See Kristelia A. García & Justin McCrary, Reconceptualizing Copyright’s Term (unpublished manuscript) (on file with author).

        [26].     For a classic account of economic arbitrage, see, for example, Andrei Shleifer & Robert W. Vishny, The Limits of Arbitrage, 52 J. Fin. 35, 35 (1997) (citing William F. Sharpe & Gordon J. Alexander, Investments (4th ed. 1990)) (defining arbitrage as “the simultaneous purchase and sale of the same, or essentially similar, security in two different markets for advantageously different prices”).

        [27].     See, e.g., Richard L. Hasen, Lobbying, Rent-Seeking, and the Constitution, 64 Stan. L. Rev. 191 (2012) (describing lobbyists as threatening economic welfare by facilitating rent-seeking, capturing legislators, and wasting government resources).

        [28].     See Emily A. Parker, Acting Chief Counsel, Internal Revenue Serv., Address at the TEI/LMSB Financial Services Industry Conference (Sept. 22, 2003), [].

        [29].     Sloan G. Speck, Tax Planning and Policy Drift, 69 Tax L. Rev. 549 (2016) (introducing the term “planning drift,” and offering examples of the phenomenon in the federal tax context).

        [30].     Elizabeth Pollman & Jordan M. Barry, Regulatory Entrepreneurship, 90 S. Cal. L. Rev. 383, 385 (2017).

        [31].     See, e.g., Del Code Ann. tit. 8, § 154 (1953).

        [32].     Take, for example, Yahoo’s most recent tender offer, with a par value of $0.001 per share. Innisfree, Yahoo!: Tender Offer, [] (last visited Dec. 21, 2018).

        [33].     See, e.g., Zachary Brennan, ‘FAST’ Generics Act Seeks to Thwart Abuse of FDA Safety Programs, Reg. Aff. Profs. Soc’y (Apr. 10, 2017), [].

        [34].     Fair Access for Safe and Timely Generics Act of 2017, 115th Cong. (2017), [] (proposed bill text as of April 6, 2017).

        [35].     See Burk, supra note 14, at 15–18.

        [36].     Wu, supra note 14, at 731.

        [37].     Id. at 732.

        [38].     Burk, supra note 14, at 7.

        [39].     Nadeem Muaddi, Chrysler Discontinues PT Cruiser, U.S. News & World Rep. (June 30, 2010), [].

        [40].     See Claire A. Hill, Is Secured Debt Efficient?, 80 Tex. L. Rev. 1117, 1128 n.46 (2002) (discussing double dipping in the lease context).

        [41].     See generally R.H. Coase, The Nature of the Firm, 4 Economica 386, 386–405 (1937).

        [42].     Fleischer, supra note 8, at 231; see also Ronald J. Gilson, Value Creation by Business Lawyers: Legal Skills and Asset Pricing, 94 Yale L.J. 239, 298 (1984) (describing the trade-off between transaction costs and regulatory costs as demonstrative of the value added to a transaction by a business lawyer).

        [43].     Fleischer, supra note 8, at 244–52.

        [44].     Id. at 244–47.

        [45].     Id. at 247–48.

        [46].     Id.

        [47].     Id. at 248–49.

        [48].     Id.

        [49].     Maria A. Pallante, The Next Great Copyright Act, 36 Colum. J.L. & Arts 315 (2013).

        [50].     See, e.g., SoundExchange Statement: Copyright Royalty Board Issues Decision in Royalty Rate Determination Case, SoundExchange (Dec. 15, 2017), [].

        [51].     Fleischer, supra note 8, at 250–51.

        [52].     Id. at 252–74.

        [53].     Tax Reduction and Reform Act of 2007, H.R. 3970, 110th Cong. (2007) (introducing the bill); see also Howard Abrams, A Technical Analysis of Proposed Section 710 (Emory Pub. Law Research Paper No. 07-25, 2007), [].

        [54].     Fleischer, supra note 8, at 253.

        [55].     Id. at 255–56.

        [56].     See generally Boris I. Bittker & Lawrence Lokken, Federal Taxation of Income, Estates and Gifts ¶¶ 4.3.3–.5 (3d ed. Cum. Supp. No. 3 2017) [hereinafter Bittker]. The substance-over-form doctrine allows the IRS to reclassify a transaction according to its substance, regardless of the form that it takes. The business purpose doctrine allows the IRS to challenge and disallow claimed business expenses for which it finds no legitimate business purpose (such as a nonexistent home office). A transaction satisfies the economic substance test if it has both substantive economic effects and a purpose other than reducing or avoiding tax. Finally, the step transaction test enables the IRS to collapse a series of separate transactions into one single transaction for the purposes of determining the proper tax consequences. Id.

        [57].     See Fleischer, supra note 8, at 257–59.

        [58].     Id.

        [59].     See Kristelia A. García, Penalty Default Licenses: A Case for Uncertainty, 89 N.Y.U. L. Rev. 1117 (2014). This is not merely a fairness argument. For example, a company like Spotify may find it worthwhile to engage in private deals only with respect to its high-value content and artists, but not others, leaving smaller players to the regulatory regime. In the case of musical content, this means that a performance rights organization (PRO) like the American Society of Composers, Authors, and Publishers (ASCAP) may find itself both diminished in funding (since it is collecting less money in both membership fees and royalties), and less efficient (with a blanket license offering that is less valuable with high-value content extracted.).

        [60].     Id. at 1137.

        [61].     Id.

        [62].     See Kristelia A. García, Facilitating Competition by Remedial Regulation, 31 Berkeley Tech. L.J. 183, 189 (2016).

        [63].     Fleischer, supra note 8, at 265.

        [64].     See supra Introduction.

        [65].     See Columbia Pictures Indus., Inc. v. Aveco, Inc., 800 F.2d 59 (3d Cir. 1986); Columbia Pictures Indus., Inc. v. Redd Horne, Inc., 749 F.2d 154 (3d Cir. 1984).

        [66].     Columbia Picture Indus., Inc. v. Prof’l Real Estate Inv’rs, Inc., 866 F.2d 278, 281 (9th Cir. 1989).

        [67].     Sony Corp. of Am. v. Universal City Studios, Inc., 465 U.S. 1112 (1984).

        [68].     Cartoon Network LP v. CSC Holdings, Inc., 536 F.3d 121 (2d Cir. 2008).

        [69].     A&M Records, Inc. v. Napster, Inc., 239 F.3d 1004 (9th Cir. 2001).

        [70].     For a detailed analysis of this evolution, see Peter S. Menell, Envisioning Copyright Law’s Digital Future, 46 N.Y.L. Sch. L. Rev. 63 (2002–2003).

        [71].     For more on the challenges that digital technology poses to copyright doctrine, see Aaron Perzanowski & Jason Schultz, Digital Exhaustion, 58 UCLA L. Rev. 889 (2011).

        [72].     For more on the challenges posed by the digital evolution to fundamental copyright concepts such as reproduction, distribution, and public performance, see Robert Brauneis & Roger E. Schechter, Copyright: A Contemporary Approach 991–93 (2d ed. 2017).

        [73].     See, e.g., Oliver Herzfeld & Marc Aaron Melzer, Fair Use in the Age of Social Media, Forbes (May 26, 2016), [] (discussing how most social media content-sharing does not qualify as fair use, despite obvious architectural encouragement).

        [74].     Jessica Litman, Real Copyright Reform, 96 Iowa L. Rev. 1, 3 (2010).

        [75].     See, e.g., infra Part II.B.1 (discussing the DSPRA’s disparate treatment of terrestrial and Internet radio).

        [76].     See Kristelia A. García, Private Copyright Reform, 20 Mich. Telecomm. & Tech. L. Rev. 1 (2013).

        [77].     A “music publisher” is a company that owns the copyrights on various musical compositions, and licenses the use of those compositions to such entities as radio stations, sports stadiums, filmmakers, restaurants, and record labels—traditionally through a PRO like ASCAP. The three major music publishers in the US today are Sony Music Publishing, LLC, Universal Music Publishing Group, and Warner Music Publishing. See, e.g., Ed Christman, Music Publishers’ 4th Quarter Report: Top 3 Companies Have the Same No. 1 Song, Billboard (Feb. 3, 2017), [].

        [78].     As amended in 2001, ASCAP’s consent decree (Second Amended Final Judgment or AFJ2) names the SDNY as rate court. United States v. Am. Soc’y of Composers, Authors and Publishers, No. 41-1395, 2001 WL 1589999, at *6–8 (S.D.N.Y. 2001).

        [79].     Nathan McAlone, Pandora Bought a Radio Station in South Dakota in a Clever Plot to Pay Lower Music Royalties, Bus. Insider (June 16, 2015), [].

        [80].     Formerly known as Clear Channel.

        [81].     McAlone, supra note 79; see also David Oxenford, Details of the ASCAP Settlement with the Radio Industry—What Will Your Station Pay?, Broadcast L. Blog (Jan. 30, 2012), [].

        [82].     See McAlone, supra note 79.

        [83].     “Total royalties” here includes not only mechanical royalties, but also public performance royalties on sound recordings. See id.

        [84].     Christopher Harrison, Why Pandora Bought an FM Radio Station, Hill (June 11, 2013), []. In the end, Pandora’s purchase of KXMZ-FM did succeed in lowering their effective royalty rate vis-à-vis ASCAP, but not vis-à-vis BMI (another PRO). Brad Hill, BMI Wins Court Ruling vs. Pandora in Royalty Rate Lawsuit, Radio & Internet News (May 15, 2015), []. In June 2017, Pandora announced that it would be selling the station to Haugo Broadcasting for $300,000 (they originally purchased the station for $600,000). See Pandora Abandons Broadcast Radio with Sale of South Dakota FM, All Access Music Grp. (June 19, 2017), [].

        [85].     Digital Performance Right in Sound Recordings Act of 1995, Pub. L. No. 104-39, 109 Stat. 336 (1995).

        [86].     17 U.S.C. § 114(d) (2012).

        [87].     I.e., sheet music, or the underlying score to a song. Importantly, the Section 115 compulsory license does not apply to use of a sound recording, which is handled in Section 114.

        [88].     17 U.S.C. § 115(c)(3) (2012) (adding that “[a] compulsory license under this section includes the right of the compulsory licensee to distribute or authorize the distribution of a phonorecord of a nondramatic musical work by means of a digital transmission which constitutes a digital phonorecord delivery, regardless of whether the digital transmission is also a public performance of the sound recording under section 106(6) of this title”).

        [89].     Id.

        [90].     See 17 U.S.C. §§ 801(b)(1), 804(b)(4) (2012). A typical CRB ratemaking for phonorecords involves the filing of Proposed Findings of Fact and Conclusions of Law from both copyright owners and platforms. See, e.g., Determination of Rates and Terms for Making and Distributing Phonorecords (Phonorecords III), Proposed Findings of Fact and Conclusions of Law, U.S. Copyright Royalty Board, [].

        [91].     17 U.S.C. § 115(b) (2012).

        [92].     Harry Fox Agency, []. The Harry Fox agency was established in 1927 by the National Music Publishers’ Association (NMPA) as the sole agency for licensing, collecting, and administering mechanical royalties. The agency was acquired by SESAC, one of three major music publishers in the US, in 2015. Most of the DIY alternatives to Harry Fox, such as RightsFlow’s Limelight, have been acquired by large content licensing platforms (in RightsFlow’s case, by Google/YouTube). A notable exception is Music Reports, an online platform for rights administration that reportedly administered $500M in 2016. Music Reports, Inc., [] (last visited Dec. 21, 2018); see also Ed Christman, Publishing Briefs: Music Reports Inc. Administered Over $500M in 2016, Billboard (March 2, 2017), [].

        [93].     17 U.S.C. § 115(b)(1) (2012).

        [94].     See id.

        [95].     See, e.g., Chris Castle, Executive Summary—Mass NOIs (2010), [].

        [96].     17 U.S.C. § 115(c)(1) (2012).

        [97].     The Copyright Office publishes all NOIs online. Section 115 NOIs Filed with the Copyright Office, U.S. Copyright Office, [] (last visited Dec. 21, 2018). For the avoidance of doubt, the exploitation of Section 115’s “address unknown” NOI protocol does not amount to perjury (since the addresses are technically not available on the Copyright Office’s public site as required by statute), but are still disingenuous at best (since, for example, Google could reasonably surmise that notice on use of a new track by Katy Perry could be served to the same party they served the last dozen Katy Perry track notices). The twenty-five million notices received likely include many duplicates, and do not necessarily represent twenty-five million separate works. Although Rightscorp—the source of these numbers—only recently began tracking the number of such filings, all signs suggest this is a dramatic increase in filings. A likely explanation for the increase is found with a change to the Copyright Office’s internal policy for accepting such filings. As explained in a Billboard report: Until recently, [filing an “address unknown” NOI] was a time-consuming, work-intensive, and costly process. NOIs had to be filed manually, by paper and under a prohibitive pricing structure. . . . Also, the NOI for each song would need to be filed individually, although they could all be batch delivered to the Copyright Office. But about two months ago, the Copyright Office revamped the way it is willing to accept NOIs and changed its pricing structure. Now, NOIs can be filed on excel spreadsheets, with something like 20 columns of relevant data needed to be completed for each song. This electronic filing still requires an upfront fee of $75 but it now only costs 10 cents a track. So now, filing NOIs for 500,00[0] songs will only cost $50,075, instead of $1.000075 million. Ed Christman, Say You Want a Revolution? U.S. Copyright Office Modernizes Key Part of Digital Licensing, Billboard (June 24, 2016), [].

        [98].     Spotify Songwriters and Publishers Administration System, Spotify (Dec. 7, 2015), [].

        [99].     Spotify USA, Inc., Section 115 NOIs Filed with the Copyright Office, U.S. Copyright Office (Mar. 10, 2018), available at USA Inc.-2018.03.10-NOI.xlsx [].

      [100].     George Jones, Chart History: The Window Up Above, Billboard, [] (last visited Dec. 21, 2018).

      [101].     Mickey Gilley, Chart History: Window Up Above, Billboard, [] (last visited Dec. 21, 2018).

      [102].     Search conducted on March 24, 2018.

      [103].     Chris Castle, Meet the New Boss: Tech Giants Rely on Loopholes to Avoid Paying Statutory Royalties with Mass Filings of NOIs at the Copyright Office, 33 Ent. & Sports L. 65, 65 (2017).

      [104].     17 U.S.C. § 115(b)(1) (2012).

      [105].     Registration Portal, U.S. Copyright Office, [] (last visited Dec. 21, 2018).

      [106].     The SXWorks website describes the tool as intended to make it “easier for publishers and songwriters to search notices sent to the U.S. Copyright Office by indexing each filing into a searchable format so they can find potential uses of their musical works and seek royalties.” NOI Lookup, SXWorks, [] (last visited Dec. 21, 2018); see, e.g., Dan Rys, Searching for Unclaimed Royalties Is About to Get Easier for Music Publishers with New SXWorks Database, Billboard (Dec. 14, 2017), [].

      [107].     An Epsteinian argument might be made that equates the Copyright Office’s delay in posting notices to a government taking; however, the implications of such an argument are outside the scope of this Article.

      [108].     Anthony Manker, Recap: David Israelite, President & CEO of NMPA, Copyright Soc’y of the S. (Nov. 5, 2015), [] (emphasis added).

      [109].     See Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013). The Second Circuit has affirmed all relevant portions of this decision. See Capitol Records, LLC v. ReDigi, Inc., No. 16-2321, 2018 WL 6518076 (2d Cir. Dec. 12, 2018).

      [110].     I say “tenuous” not least of all because commentators disagree on what the “spirit of law” is. Compare Brief of Amici Curiae American Library Ass’n, Ass’n of College and Research Libraries, Ass’n of Research Libraries, and Internet Archive in Support of Reversal, Capitol Records, LLC v. ReDigi, Inc., No. 16-2321 (2d Cir. Feb. 14, 2017) (arguing ReDigi’s resell of “used” MP3s constitutes fair use), with Brief for Amicus Curiae The Copyright Alliance in Support of Plaintiff-Appellees, Capitol Records, LLC v. ReDigi, Inc., No. 16-2321 (2d Cir. May 12, 2017) (arguing that the first sale doctrine does not apply to ReDigi, nor is it fair use).

      [111].     17 U.S.C. § 106 (2012).

      [112].     Plaintiff’s Statement of Undisputed Facts Pursuant to Local Rule 56.1, ¶ 6, Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013) (No. 1:12-cv-00095) [hereinafter Capitol’s Rule 56.1 Statement].

      [113].     Id. ¶¶ 6, 9.

      [114].     17 U.S.C. § 109 (2012).

      [115].     17 U.S.C. § 109(a) (2012).

      [116].     Defendant’s Statement of Undisputed Facts Pursuant to Local Rule 56.1, ¶ 8, Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013) (No. 1:12-cv-00095) [hereinafter ReDigi’s Rule 56.1 Statement].

      [117].     Id. Per ReDigi, a file is considered “legally acquired” if its metadata show that it was purchased (from iTunes or another ReDigi user). Files ripped from a CD, or downloaded from a P2P or bit torrent site are deemed ineligible for sale.

      [118].     ReDigi’s Rule 56.1 Statement, supra note 116, ¶¶ 9, 11; Capitol’s Rule 56.1 Statement, supra note 112, ¶ 22.

      [119].     ReDigi’s Rule 56.1 Statement, supra note 116, ¶¶ 14, 36.

      [120].     Capitol’s Rule 56.1 Statement, supra note 112, ¶ 14.

      [121].     17 U.S.C. § 106(1) (2012).

      [122].     Capitol’s Rule 56.1 Statement, supra note 112, ¶¶ 59–61, 63.

      [123].     See Complaint, Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013) (No. 12 Civ. 0095), 2012 WL 32056 (alleging multiple counts of direct copyright infringement, inducement, contributory and vicarious copyright infringement, and common law copyright infringement; and seeking preliminary and permanent injunctions, as well as damages, attorneys’ fees and costs, and interest).

      [124].     Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013).

      [125].     Capitol Records, LLC v. ReDigi, Inc., No. 16-2321, 2018 WL 6518076 (2d Cir. Dec. 12, 2018).

      [126].     For a balanced summary of the various issues presented, see, for example, Clark D. Asay, Kirtsaeng and the First-Sale Doctrine’s Digital Problem, Stan. L. Rev. Online (May 2013), [].

      [127].     See Pollman & Barry, supra note 30.

      [128].     Aaron Perzanowski & Jason Schultz, The End of Ownership (2016) (noting, for example, that “buy now” often refers to a rental/lease in the digital context, rather than a traditional purchase with full ownership rights).

      [129].     See 17 U.S.C. § 109(a) (2012).

      [130].     See generally Perzanowski & Schultz, supra note 128.

      [131].     Kindle Store Terms of Use, Amazon (Oct. 5, 2016), [].

      [132].     Aaron Perzanowski & Chris Jay Hoofnagle, What We Buy When We Buy Now, 165 U. Pa. L. Rev. 315, 322 (2017).

      [133].     For a detailed description of VidAngel’s system, see Disney Enters., Inc. v. VidAngel, Inc., 869 F.3d 848, 853–55 (9th Cir. 2017).

      [134].     Id. at 854–55.

      [135].     Id. at 854 n.3.

      [136].     Specifically, 17 U.S.C. § 106(1), (4) (2012).

      [137].     Specifically, 17 U.S.C. § 1201(a)(1)(A) (2012).

      [138].     17 U.S.C. § 107 (2012).

      [139].     17 U.S.C. § 110(11) (2012).

      [140].     Id.

      [141].     Disney Enters., Inc. v. VidAngel, Inc., 224 F. Supp. 3d 957 (C.D. Cal. 2016).

      [142].     See Disney Enters., Inc. v. VidAngel, Inc., 869 F.3d 848 (9th Cir. 2017).

      [143].     Id. at 859.

      [144].     For a detailed description of Aereo’s system, see Am. Broad. Cos., Inc. v. Aereo, Inc., 134 S. Ct. 2498, 2503 (2014).

      [145].     17 U.S.C. § 106(4) (2012).

      [146].     17 U.S.C. § 101 (2012).

      [147].     The majority recognizes—and the dissent relies upon—a difference between Aereo’s system and a cable company: the latter transmits content constantly, while Aereo’s system waits to begin transmitting until a subscriber requests a program. Ultimately, the Court finds the distinction insignificant. See Aereo, 134 S. Ct. at 2507.

      [148].     Id.

      [149].     17 U.S.C. § 111 (2012).

      [150].     See Joe Mullin, Aereo: Hey, We’re a Cable Company After All!, Ars Technica (July 10, 2014), [].

      [151].     See, e.g., Brian Fung, Aereo to the FCC: Let Us Join the Cable Companies We Tried to Replace, Wash. Post (Oct. 13, 2014), [] (“By accepting the label of MVPD, Aereo would also need to start negotiating with broadcasters over content, or ‘retransmission’ fees—which is precisely what Aereo’s original business model was designed to avoid.”)

      [152].     Letter from Jacqueline C. Charlesworth, Gen. Counsel & Assoc. Register of Copyrights, United States Copyright Office, to Matthew Calabro (July 16, 2014), []; see also Brian Fung, No, Aereo Isn’t a Cable Company, Says the Copyright Office, Wash. Post (July 17, 2014), []. In the meantime, a district court has sided with the Copyright Office in denying Aereo access to the compulsory license in Section 111. See Bill Donahue, Aereo Can’t Use Compulsory License, Judge Says, Law 360 (Oct. 23, 2014), [].

      [153].     See In the Matter of Promoting Innovation & Competition in the Provision of Multichannel Video Programming Distribution Services, 29 FCC Rcd. 1547 (2014). The FCC has still not moved on this issue; the definition proposed in the rulemaking would significantly expand the FCC’s media regulation authority (a move considered somewhat unlikely to succeed in any case).

      [154].     See Fung, supra note 151.

      [155].     David Lumb, Aereo-like NYC Nonprofit Locast Streams Local TV for Free, Engadget (Jan. 19, 2018), [].

      [156].     FAQs, Locast (Feb. 9, 2018), [] (“As in other urban areas, tall buildings often make it difficult to receive an over-the-air signal in an apartment or condo. Technical changes to broadcasting have made it harder to get a signal—the transition to digital broadcasting changed the signal propagation characteristics of over-the-air signals and the recent ‘incentive auction’ at the FCC in which some broadcasters sold their licenses while others will have to change their frequency contributed to this trend. . . . helps to keep the promise of public access to local broadcasting . . . .”).

      [157].     See Memorandum of Law in Support of Applicants’ Motion for Partial Summary Judgment, United States v. Am. Soc’y of Composers, Authors & Publishers, 607 F. Supp. 2d 562 (S.D.N.Y. 2009) (No. 41-1395), 2007 WL 7012850.

      [158].     17 U.S.C. § 106(4) (2012).

      [159].     17 U.S.C. § 101 (2012).

      [160].     Greg Sandoval, Music Publishers: iTunes Not Paying Fair Share, CNET (Sept. 17, 2009), [].

      [161].     Id.

      [162].     Unites States v. Am. Soc’y of Composers, Authors & Publishers, 485 F. Supp. 2d 438, 443–44 (S.D.N.Y. 2007) (citations omitted).

      [163].     See, e.g., A&M Records, Inc. v. Napster, Inc., 239 F.3d 1004, 1014 (9th Cir. 2001) (“Napster users who download files containing copyrighted music violate plaintiffs’ reproduction rights.”).

      [164].     U.S. Copyright Office, DCMA Section 104 Report, at xxvii-iii (2001) (“It is our view that no liability should result from a technical ‘performance’ that takes place in the course of a download.”).

      [165].     Am. Soc’y of Composers, Authors & Publishers, 485 F. Supp. 2d at 447.

      [166].     Id.

      [167].     United States v. Am. Soc’y of Composers, Authors & Publishers, 627 F.3d 64 (2d Cir. 2010).

      [168].     Digital Millennium Copyright Act, Pub. L No. 105-304, 112 Stat. 2860 (1998).

      [169].     17 U.S.C. § 512 (2012).

      [170].     OSPs wishing to take advantage of the DMCA safe harbor must designate an agent to receive takedown notices from content owners and establish a policy for effecting takedown of the infringing material. Notably, there is no requirement on the part of an OSP to make an affirmative determination as to the legitimacy of user-uploaded content, nor is the takedown procedure required to prevent re-upload of the same, or similar, infringing material. 17 U.S.C. § 512(c) (2012).

      [171].     See Digital Millennium Copyright Act, supra note 168, at 8 (“[W]ithout clarification of their liability, service providers may hesitate to make the necessary investment in the expansion of the speed and capacity of the Internet. In the ordinary course of their operations service providers must engage in all kinds of acts that expose them to potential copyright infringement liability. For example, service providers must make innumerable electronic copies by simply transmitting information over the Internet. Certain electronic copies are made to speed up the delivery of information to users. Other electronic copies are made in order to host World Wide Web sites. Many service providers engage in directing users to sites in response to inquiries by users or they volunteer sites that users may find attractive. Some of these sites might contain infringing material. In short, by limiting the liability of service providers, the DMCA ensures that the efficiency of the Internet will continue to improve and that the variety and quality of services on the Internet will continue to expand.”).

      [172].     17 U.S.C. § 512(c)(1) (2012).

      [173].     17 U.S.C. § 512(k)(1)(B) (2012).

      [174].     See Viacom Int’l, Inc. v. YouTube, Inc., 676 F.3d 19, 39 (2d Cir. 2012).

      [175].     Another part of the problem is that platforms like Spotify largely compete with YouTube’s Red service, but are not direct competitors with the free video service.

      [176].     Todd C. Frankel, Why Musicians Are So Angry at the World’s Most Popular Music Streaming Service, Wash. Post (July 14, 2017), []; see also Int’l Fed’n of the Phonographic Indus., Global Music Report 2016: State of the Industry 23 (2016), [] [hereinafter IFPI Music Report] (providing a full breakdown of this figure) (“Public data suggests a striking disparity between the amount of revenues proportionately being returned to rights holders by two leading companies in the different sectors, namely Spotify and YouTube. From publicly-available data, IFPI estimates that Spotify paid record companies US$18 per user in 2014, the last year of available data; by contrast it is estimated that YouTube delivered less than US$1 per user to rights holders in 2015. These estimates provide a realistic illustration of the ‘value gap’ in practice.”).

      [177].     Frankel, supra note 176.

      [178].     This ability to re-upload infringing content was previously limited to some extent by the voluntary graduated response system (commonly referred to as the “copyright alert system”) in which participating ISPs will send notices to repeat infringers, eventually leading to, for example, a throttling of the user’s Internet speed. That program was terminated at the beginning of 2017. See, e.g., Ted Johnson, Internet Service Providers, Studios and Record Labels Call It Quits on Copyright Alert System, Variety (Jan. 27, 2017), []. YouTube’s proprietary fingerprinting technology, Content ID, also helps to some extent by attempting to identify (and claim for copyright owners) infringing content upon upload. See infra Part III.B.

      [179].     Mathew Ingram, YouTube Pays Billions, But the Music Industry Says It’s Not Enough, Fortune (July 13, 2016), [].

      [180].     Alphabet Inc., Form 10-K: Annual Report Pursuant to Section 13 or 15(d) of the Securities and Exchange Act of 1934, at 9 (2016), [].

      [181].     Of course, the same might be said of any ad-supported platform, such as a broadcast TV station. In any event, YouTube might contend that Content ID affords content owners the opportunity to monetize (however poorly) where there is no statutory requirement to do so.

      [182].     Determining what YouTube “should” be paying, however, is difficult, not least of all because there is no “market rate” for OSP streaming, since it operates under the statute. By definition, a market rate is determined in a market, and markets require both willing buyers and willing sellers. Because of the existence of the safe harbor—which obviates the need for OSPs to bother with licensing—there is no market (in the economic sense) for the licensing of these rights. There never has been, and there never will be, unless and until Section 512 is revoked (a highly unlikely and in any event ill-advised solution) or amended. This means that we don’t know what the market rate is, or should be, for OSP streaming. What we do know is that a “negotiation” held in the shadow of the safe harbor lacks a willing buyer (as content owners claim to be competing with free content—meaning pirated or otherwise unlicensed content). As Warner Music Group’s CEO Steve Cooper put it: “There’s no getting around the fact that, even if YouTube doesn’t have licenses, our music will still be available but not monetized at all. Under those circumstances, there can be no free-market ‘willing buyer, willing seller’ negotiation.” Daniel Sanchez, Indie Labels Say Spotify Is Paying 3X Better Than YouTube, Digital Music News (June 12, 2017), []. Nonetheless, Warner and YouTube reached an agreement in May 2017. While the terms of the deal are not publicly available, Cooper described the circumstances surrounding the negotiation as “very difficult.” Tim Ingham, Warner and YouTube Sign New Deal . . . ‘Under Very Difficult Circumstances,Music Bus. Worldwide (May 5, 2017), []. To be clear, since we do not have a market, we also don’t know if we have, as content owners claim, a supra-competitive rate. The only way to eventually determine a market rate for public performance royalties is to do away with the Section 512 safe harbor, a decision which itself presents a number of concerns—including the potential for content owners to shut down certain services in favor of others. A recent dispute in music publishing is illustrative of this point. When two of the three major U.S. music publishers decided to withdraw their digital rights from ASCAP in order to negotiate private rates in the market, Sony—which at the time controlled roughly 30 percent of the market—began its negotiation with prospective licensee Pandora by stating: “[I]t’s not our intention to shut down Pandora.” In re Petition of Pandora Media, Inc., 6 F. Supp. 3d 317, 343 (S.D.N.Y. 2014). The not-so-subtle implication was that Pandora could agree to the rate set by Sony, or face crippling copyright litigation and withdrawal of content that could handily shut the service down. Pandora acknowledged as much in its comments to the Federal Trade Commission when Sony first sought to acquire competing music publisher EMI, noting that the combination would afford Pandora “no choice” but to accept whatever terms Sony offered, as Pandora “could not survive without access to the combined Sony and EMI catalogues.” Id. For more on this dynamic, see García, supra note 62. In other words, the threat to shut down Pandora was intended precisely to increase the rate above what a willing buyer (i.e. Pandora) was willing to pay, and can be seen as a type of arbitrage on the part of the music publishers. The impact of this arbitrage is undoubtedly exacerbated by the oligopolistic nature of the music publishing industry, in which a mere three companies control over 60 percent of the market. See Ed Christman, Publishers Quarterly: Sony/ATV Reclaims No. 1 Spot in Fourth Quarter of 2017, Billboard (Feb. 1, 2018), [].

      [183].     See H.R. Rep. No. 105-551, pt. 2, at 21 (1998) (discussing the purpose of Section 512 as “balanc[ing] the interests of content owners, on-line and other service providers, and information users in a way that will foster the continued development of electronic commerce and the growth of the Internet”).

      [184].     To understand how YouTube qualifies as an OSP for the purposes of Section 512, while Spotify does not, even though the companies may be “similarly situated,” it is helpful to recognize YouTube as a participant in two different markets: UGC (which earns it the protection of the safe harbor, and in which Spotify does not compete) and official, paid-for content (for which it negotiates payment, and in which market it competes with Spotify). Since both types of content reside on YouTube’s platform, however, unlicensed content “competes” with licensed content on the service.

      [185].     See, e.g., Hugh McIntyre, Report: YouTube Is the Most Popular Site for On-Demand Music Streaming, Forbes (Sept. 27, 2017), [].

      [186].     Updated! Streaming Price Bible w/ 2016 Rates: Spotify, Apple Music, YouTube, Tidal, Amazon, Pandora, Etc., Trichordist (Jan. 16, 2017), [].

      [187].     See, e.g., Takedown Hall of Shame Honorees, Elec. Frontier Found., [] (last visited Dec. 21, 2018).

      [188].     17 U.S.C. § 512(f) requires a level of good faith from filers of takedown notices, prohibiting “material misreprent[ations]” and establishing damages for so doing. 17 U.S.C. § 512(f) (2012).

      [189].     See, e.g., Mike Masnick, DMCA Misuse: Trying to Take Down a Negative Movie Review, Techdirt (Jan. 14, 2008), [] (describing misuse of a takedown notice to remove a negative film review).

      [190].     See, e.g., Mike Masnick, DMCA Abused to Get Sucks Site Taken Down, Techdirt (Oct. 11, 2007), [] (describing use of a takedown notice by a home building company to take down a suck site commenting on their work).

      [191].     Jennifer M. Urban, Joe Karaganis & Brianna L. Schofield, Notice and Takedown in Everyday Practice 2 (2017), [].

      [192].     Id.

      [193].     Id; see also Lenz v. Universal Music Corp., 815 F.3d 1145 (9th Cir. 2016) (requiring fair use to be considered prior to issuance of a takedown notice under Section 512).

      [194].     Ayres & Kapczynski, supra note 19, at 1812–21.

      [195].     Id. at 1813.

      [196].     Id. at 1815–16.

      [197].     See, e.g., Hasen, supra note 27, at 191 (describing lobbyists as threatening economic welfare by facilitating rent-seeking, capturing legislators, and wasting government resources).

      [198].     Id. at 231.

      [199].     Timothy B. Lee, 15 Years Ago, Congress Kept Mickey Mouse Out of the Public Domain. Will They Do It Again?, Wash Post: The Switch (Oct. 25, 2013), [].

      [200].     See, e.g., Brief of George A. Akerlof et al., Eldred v. Ashcroft, 537 U.S. 186 (2003) (No. 01-618), 2002 WL 1041846, at *3 (“Taken as a whole, it is highly unlikely that the economic benefits from copyright [term] extension . . . outweigh the additional costs.”).

      [201].     Sonny Bono Memorial Collection, [] (last visited Dec. 21, 2018).

      [202].     17 U.S.C. § 108(h) (2012).

      [203].     Brewster Kahle, Books from 1923 to 1941 Now Liberated!, Internet Archive Blogs (Oct. 10, 2017), [].

      [204].     See, e.g., Leading Multimedia Websites in the United States in November 2016, Based on Market Share of Visits, Statista, [] (last visited Dec. 21, 2018) (showing YouTube in first place at 78 percent market share and Netflix in second place at 8 percent).

      [205].     The incentive theory of copyright considers the motivation of creators to be central to copyright’s purpose. In Mazer v. Stein, the Supreme Court described the philosophy as the “encouragement of individual effort by personal gain.” 347 U.S. 201, 219 (1954).

      [206].     See Glynn Lunney, Copyright’s Excess: Money and Music in the U.S. Recording Industry (2018).

      [207].     Harriet Taylor, Pandora CEO: ‘We Intend to be Profitable This Year,CNBC (Mar. 13, 2017), [].

      [208].     Spotify Tech. S.A., Consolidated Financial Statements (December 31, 2016), [].

      [209].     Ben Sisario, Spotify Is Growing, But So Are Its Losses, N.Y. Times (June 15, 2017), []; see also Hugh McIntyre, Spotify’s Revenue Spiked in 2016, But Its Losses Rose by a Greater Margin, Forbes (June 15, 2017), [] (citing a year-over-year increase in losses of 133 percent).

      [210].     Dominic Paschel, Vice President, Pandora, Naveen Chopra, CFO and Interim CEO, Pandora, Karen Walker, Chief Accounting Officer, Pandora, Scripts for Pandora (P) Q217 Financial Results Conference Call (July 31, 2017), [].

      [211].     For tiers and pricing, see Pandora, [] (last visited Dec. 21, 2018).

      [212].     Press Release: Pandora Reports Q2 2017 Financial Results, PANDORA (July 31, 2017), [].

      [213].     See supra text accompanying note 97.

      [214].     Alternatively, it might encourage purchases by consumers interested in resale, although it should be noted there is nothing to stop consumers from reselling “used” MP3s. Therefore, on balance, sales are likely to head in only one direction for iTunes.

      [215].     It is of course true that Section 109’s first sale doctrine is itself not an incentivizing measure; rather, Congress intended it to enlarge the scope of user rights as part of the broader incentive-access balance. Unlike copyright arbitrage, however, the doctrine is presumed to be the product of robust debate and compromise, and applies across the board to all similarly situated content.

      [216].     In contrast with its free, ad-supported service, YouTube’s paid services (currently branded as YouTube Music and YouTube Premium) pay for streams at a privately negotiated rate. While this may matter in the future, it is currently of little consequence: of YouTube’s estimated 1 billion monthly users, a mere 1.5 million bother paying for the service. See Micah Singleton, YouTube Is Still Having Trouble Getting People to Pay for YouTube, Verge (Nov. 2, 2016), [].

      [217].     Frankel, supra note 176.

      [218].     The EU has formally recognized this “value gap,” while the jury is still out in the U.S. See European Commission Press Release, State of the Union 2016: Comm’n Proposes Modern EU Copyright Rules for European Culture to Flourish and Circulate IP/16/310 (Sept. 14, 2016), [].

      [219].     144 Cong. Rec. S4886 (daily ed. May 14, 1998) (statement of Sen. Leahy).

      [220].     Cary Sherman, Medium: Five Stubborn Truths About YouTube and the Value Gap, RIAA: News (Aug. 18, 2017), [].

      [221].     Peter Kafka, Here’s Why the Music Labels Are Furious at YouTube. Again., Recode (Apr. 11, 2016), []. Kafka’s only half right: the safe harbor was intended for passive, neutral intermediaries, but it’s not clear to this author that Verizon doesn’t make money off of distributing content; if there was no content on the Internet, no one would subscribe to Verizon’s Internet service.

      [222].     Frankel, supra note 176.

      [223].     Because monetization is not an option under Section 512(c), Content ID is often referred to as a “DMCA-plus” agreement. How Content ID Works, Youtube, [] (last visited Dec. 21, 2018).

      [224].     See, e.g., Saba Hamedy, YouTube Has Paid $2 Billion to Rights Holders Through Content ID, Mashable (July 13, 2016), [] (citing the percentage of copyright claims handled through Content ID at 98 percent, with only 2 percent resorting to takedown notices).

      [225].     IFPI Music Report, supra note 176. While beyond the scope of this Article, Content ID itself may be a source for future competitive concerns as it potentially sets an industry expectation of “DMCA-plus” accommodation. See, e.g., European Comm’n, Directorate-General for Communications Networks, Content, and Technology, Proposal for a Directive of the European Parliament and of the Council on copyright in the Digital Single Market, Title IV, Ch. 2, Art. 13, [] (“[S]ervice providers that store and provide to the public access to large amounts of works or other subject-matter uploaded by their users shall, in cooperation with rightholders, take measures to ensure the functioning of agreements concluded with rightholders for the use of their works or other subject-matter or to prevent the availability on their services of works or other subject-matter identified by rightholders through the cooperation with the service providers. Those measures, such as the use of effective content recognition technologies, shall be appropriate and proportionate. . . . Member States shall facilitate, where appropriate, the cooperation between the information society service providers and rightholders through stakeholder dialogues to define best practices, such as appropriate and proportionate content recognition technologies . . . .”).

      [226].     Frankel, supra note 176 (quoting Warner Music CEO Steve Cooper).

      [227].     Id.

      [228].     Thereby serving as yet another argument against complete elimination of the safe harbor. For more on the severity of statutory damages in copyright, see, for example Pamela Samuelson & Ben Sheffner, Debate, Unconstitutionally Excessive Statutory Damage Awards in Copyright Cases, 158 U. PA. L. Rev. Online 53 (2009), [].

      [229].     See, e.g., Christman, supra note 182 (showing three music publishing companies controlling over 60 percent of the total market).

      [230].     Also referred to as “network externalities,” “network effects” refer to the increased value to a consumer deriving from an increase in the number of other consumers enjoying the same product or service. Consumers are more likely to join the music-streaming platform that their friends use, for example, so that they can exchange playlists and recommendations.

      [231].     See García, supra note 76.

      [232].     The statutory distribution of royalties as described in 17 U.S.C. § 114(g)(2) of the Copyright Act is as follows: 50 percent of receipts shall be paid to the copyright owner; 2.5 percent of receipts shall be deposited in an escrow account for distribution to nonfeatured musicians; 2.5 percent of receipts shall be deposited in an escrow account for nonfeatured vocalists; and 45 percent of receipts shall be paid to the featured recording artist on the sound recording. 17 U.S.C. § 114(g)(2) (2012).

      [233].     García, supra note 59, at 1151–56.

      [234].     For example, in 2012, digital music service DMX presented to the CRB the rate it had reached privately with music publisher Sony/ATV in 2007, and the rate was adopted. What DMX failed to disclose, however, was a $2.7 million advance that accompanied the rate it reached with Sony/ATV. See Brief and Special Appendix for Petitioner-Appellant, Broad. Music, Inc. v. DMX, Inc., 683 F.3d 32 (No. 10-3429), 2010 WL 5580502, at 20–21 (“When DMX solicited direct licenses from smaller music publishers, it never told them about the advances it had committed to pay Sony. . . . It nonetheless sought to induce publishers to enter into direct licenses that did not include those substantial payments by assuring them they would be treated the same as a sophisticated major publisher who had accepted the same deal.” (emphasis omitted)); see also Broad. Music, Inc. v. DMX, Inc., 683 F.3d 32, 35 (2d Cir. 2012) (affirming lower court’s adoption of DMX’s proposed rate over ASCAP’s and BMI’s).

      [235].     ReDigi launched in October 2011 and was found guilty of copyright infringement on March 20, 2013. See Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013).

      [236].     Perhaps it is; the case is still on appeal. For a detailed read on the status of intellectual property as property, and therefore alienable, see Brief of Copyright Law Scholars as Amici Curiae Supporting Defendants-Appellants, Capitol Records, LLC v. ReDigi, Inc., No. 16-2321 (2d Cir. Feb. 14, 2017).

      [237].     While the ReDigi site is currently unavailable pending appeal, screenshots of the service are available. Mark Langshaw, ReDigi Digital Music Reseller ‘In Breach of Copyright Law, Digital Spy (April 2, 2013), [].

      [238].     See Laura Fitzpatrick, Brief History: YouTube, Time (May 31, 2010),,9171,1990787,00.html [] (“The Site marked the May anniversary of its 2005 beta launch . . . .”).

      [239].     See Stephanie Clifford, Pandora’s Long Strange Trip, Inc. (Oct. 1, 2017), [] (“Pandora has been around in one form or another since 1999 . . . .”).

      [240].     See, e.g., Alex Suskind, 15 Years After Napster: How the Music Service Changed the Industry, Daily Beast (June 6, 2014), [] (discussing how Napster “revolutioniz[ed] a new frontier in music consumption”).

      [241].     See, e.g., Larry Downes and Paul Nunes, Big-Bang Disruption, Harv. Bus. Rev. (March 2013), [] (citing Napster as an example of disruptive innovation).

      [242].     See, e.g., Dr. David C. Lowery, Songwriter Would Need 288 Million Spins to Equal Average Spotify Employee Salary, Trichordist (May 26, 2016), [].

      [243].     See, e.g., Melville B. Nimmer & David Nimmer, 1 Nimmer on Copyright § 2.03 [A] (2015) (“All that can be said by way of summary is that the concept ‘works of authorship’ is intentionally left vague under the Act.”).

      [244].     See Part I for a discussion of copyright arbitrage’s tendency toward negative impact on balance.

      [245].     Sandra M. Aistars, The Next Great Copyright Act, or a New Great Copyright Agency?, 38 Colum. J.L. & Arts 339, 340 (2015).

      [246].     See, e.g., Guido Calabresi & A. Douglas Melamed, Property Rules, Liability Rules, and Inalienability: One View of the Cathedral, 85 Harv. L. Rev. 1089 (1972).

      [247].     Robert P. Merges, Contracting into Liability Rules: Intellectual Property Rights and Collective Rights Organizations, 84 Calif. L. Rev. 1293, 1378 (1996); see also Ivan Reidel, Note, The Taylor Swift Paradox: Superstardom, Excessive Advertising and Blanket Licenses, 7 N.Y.U. J.L. & Bus. 731, 770 (2011) (“The strongest argument for blanket licenses [comparable to compulsory licenses] was then and remains today the capacity of these licenses to deliver large savings in transaction costs.”).

      [248].     See, e.g., David Lowery, Credit Check: Serial “Bad Actors” Should Lose Access to Compulsory Licenses, Trichordist (June 9, 2014), [] (recounting his receipt of checks for content licensed without notice and of content used without notice or payment).

      [249].     Jessica D. Litman, Copyright Legislation and Technological Change, 68 Or. L. Rev. 275, 278 n.9 (1989).

      [250].     113 Cong. Rec. 8581 (1967).

      [251].     Jessica D. Litman, Copyright, Compromise, and Legislative History, 72 Cornell L. Rev. 857, 871 (1987).

      [252].     Id. at 860–61.

      [253].     Hearings Before the Subcomm. on Courts, Civil Liberties, and the Admin. of Justice of the H. Judiciary Comm. on H.R. 2223 Copyright Law Revision, 94th Cong. Pt. 3 at 1853 (1975).

      [254].     Litman, supra note 74, at 39.

      [255].     Herbert Hovenkamp, The Antitrust Enterprise 250 (2005).

      [256].     William F. Patry, Copyright and the Legislative Process: A Personal Perspective, 14 Cardozo Arts & Ent. L.J. 139, 141 (1996).

      [257].     Alicia Parlapiano, How the ‘Small-Business Tax Cut’ Would Also Be a Tax Cut for the Wealthy, N.Y. Times (Dec. 20, 2017), [].

      [258].     Capitol Records, LLC v. ReDigi, Inc., 934 F. Supp. 2d 640 (S.D.N.Y. 2013).

      [259].     Id. at 655 (holding that “the first sale doctrine does not protect ReDigi’s distribution of Capitol’s copyrighted works. This is because, as an unlawful reproduction, a digital music file sold on ReDigi is not ‘lawfully made under this title’”).

      [260].     See Shyamkrishna Balganesh & Gideon Parchomovsky, Equity’s Unstated Domain: The Role of Equity in Shaping Copyright Law, 163 U. Pa. L. Rev. 1859 (2015) (discussing equity’s influence in judicial construction of copyright doctrine).

      [261].     See supra note 144 and accompanying text.

      [262].     Balganesh & Parchomovsky, supra note 260, at 1888.

      [263].     U.S. Const. art. III.

      [264].     H.R. 3945, 115th Cong. (2017). For a robust discussion of what such a court might look like, including potential pros and cons, see Pamela Samuelson & Kathryn Hashimoto, Scholarly Concerns About a Proposed Small Copyright Claims Tribunal, Berkeley Tech. L.J. Online (Nov. 28, 2017), [].

      [265].     Mark A. Lemley, The Regulatory Turn in IP, 36 Harv. J.L. & Pub. Pol’y 109, 115 (2013).

      [266].     Joseph P. Liu, Regulatory Copyright, 83 N.C. L. Rev. 87, 91–92 (2004) (internal citations omitted); see also William W. Fisher III, Promises to Keep: Technology, Law, and the Future of Entertainment 195 (2004) (considering the treatment of copyright as a regulatory regime); Menell, supra note 70, at 195 (describing increasingly polarized positions on the expansion or contraction of copyright protections as “conducive to an outcome in which Congress delegates resolution of the problem to a regulatory agency”).

      [267].     Compare Liu, supra note 266 (making the case for copyright as an increasingly regulatory body of law), with Adam Mossoff, Is Copyright Property?, 42 San Diego L. Rev. 29 (2005) (supporting a Lockean justification for copyright as property). Both sides of the debate would agree, at least, that the quintessential property model—that of the “right to exclude” as central to property ownership—is problematic for copyright, which deals with public (i.e. nonrivalrous and nonexhaustive) goods.

      [268].     See, e.g., Mark Schultz, Copyright Reform Through Private Ordering, CATO Unbound (Jan. 14, 2013), [] (“Property rights . . . put decisions in the hands of those with the best information and biggest stake in getting things right—the owner of the right and his customers and trading partners.”).

      [269].     Copyright Act of 1909, Pub. L. No. 60-349, 35 Stat. 1075 (1909); see also Pamela Samuelson, Justifications for Copyright Limitations & Exceptions, in Copyright Law in an Age of Limitations and Exceptions (Ruth Okediji ed., 2017) (“The 1909 Act had three L&E [limitation & exception] provisions. One was a codification of the exhaustion doctrine. A second limited the newly created right of composers to control mechanical reproductions of their music in sound recordings by subjecting their works to a compulsory license. Once a copyrighted song had been recorded once, anyone could re-record the song as long as they paid the license fee set forth in the statute. A third exempted coin-operated music machines unless their owners charged for admission to the premises.”) (citations omitted).

      [270].     See, e.g., Menell, supra note 70 (describing copyright’s shift from property right to regulation).

      [271].     Lemley, supra note 265, at 110–11.

      [272].     Timothy Wu, Copyright’s Communications Policy, 103 Mich. L. Rev. 278, 279 (2004).

      [273].     Melville B. Nimmer, 1 Nimmer on Copyright, at vi (1978).

      [274].     17 U.S.C. § 701 (2012).

      [275].     H.R. Rep. No. 105-796, at 64 (1998) (“It is the intention of the conferees that, as is typical with other rulemaking under title 17, and in recognition of the expertise of the Copyright Office, the Register of Copyrights will conduct the rulemaking, including providing notice of the rulemaking, seeking comments from the public . . . and recommending final regulations in a report to the Librarian.”); see also U.S. Copyright Office, Compendium of U.S. Copyright Office Practices, Chapter 100: U.S. Copyright Office and the Copyright Law 1–2 (3d ed. 2017), [].

      [276].     Intercollegiate Broad. Sys., Inc. v. Copyright Royalty Bd., 684 F.3d 1332 (D.C. Cir. 2012).

      [277].     U.S. Const. art. II, § 2, cl. 2.

      [278].     Aistars, supra note 245, at 345–51.

      [279].     Some commentators lament the current Copyright Office’s perceived bias in favor of copyright owners, which, they argue, the Office views as its “customers.” See Liu, supra note 266, at 158 (“One common critique of the Copyright Office has been the extent to which it has largely backed the interests of copyright holders over the interests of consumers more generally. . . . At the same time, it is easy to see how the Office might feel particularly attuned to the interests of copyright holders, as copyright holders are the main consumers of the Copyright Office’s historic function in registering copyrights and administering compulsory licenses. Given this practical tie, it is perhaps unsurprising that the Office might feel particularly aligned with copyright owners.”)

      [280].     Bittker, supra note 56.

      [281].     Id.

      [282].     Leo Katz, A Theory of Loopholes, 39 J. Legal Stud. 1, 7–10 (2010). Use of the term “mismatch theory” in the loophole context should not be confused with the unfortunate use of the same term in the affirmative action context.

      [283].     Id. at 7; see also Katharina Pistor & Chenggang Xu, Incomplete Law, 35 N.Y.U. J. Int’l L. & Pol. 931, 932–33 (2003) (describing how legislative incompleteness can result not only from imperfect drafting, but also from technological or social changes).

      [284].     Katz, supra note 282, at 7.

      [285].     Id. at 2.

      [286].     This conclusion is supported by the fact that as soon as the purchase failed to secure Pandora the rate reduction sought, they unloaded it at a loss. See supra Part II.B.1.

      [287].     Links to Greenbook reports dating back to FY1990 are available at [] (last visited Dec. 21, 2018).

      [288].     U.S. Dep’t of the Treasury, General Explanations of the Administration’s Fiscal Year 2017 Revenue Proposals 105 (Feb. 2016), [].

      [289].     Id. at 112–13.

      [290].     Id. at 112.

      [291].     Id.

      [292].     Recognized Abusive and Listed Transactions, supra note 4.

      [293].     Rev. Rul. 2003-6.

      [294].     Steven Toscher & Bruce I. Hochman, The Multifront War Against Corporate Tax Shelters, 23 L.A. Law. 19, 21 (Nov. 2000).

      [295].     Id.

      [296].     George K. Yin, Getting Serious About Corporate Tax Shelters: Taking a Lesson from History, 54 SMU L. Rev. 209 (2001) (discussing inefficiencies associated with the Tax Avoidance Game).

      [297].     Stacey L. Dogan & Mark A. Lemley, Antitrust Law and Regulatory Gaming, 87 Tex. L. Rev. 685, 687 (2009).

      [298].     Id. at 729.

      [299].     See García, supra note 62, at 222 (identifying a loss of structural competition in copyright industries that makes it difficult for antitrust to function in them). Indeed, a DOJ-mandated break-up of platform giants like Google, and oligopolies like the major record labels, might go some way toward alleviating the prevalence and impact of copyright arbitrage.

      [300].     See Dogan & Lemley, supra note 297, at 686 (noting that “antitrust laws are impliedly repealed by government regulation of a particular industry”).

      [301].     See, e.g., Herbert Hovenkamp et al., IP and Antitrust: An Analysis of Antitrust Principles Applied to Intellectual Property Law § 1.3 (2002) (explaining that “intellectual property laws can be justified by the public goods argument only to the extent that the laws on balance encourage enough creation and dissemination of new works to offset those costs”).

More in this Issue

Arbitration Nation: Data from Four Providers

Forced arbitration has long been controversial. In the 1980s, the Supreme Court expanded the Federal Arbitration Act (FAA), sparking debate about whether private dispute resolution was an elegant alternative to litigation or a rigged system that favors repeat-playing corporations. Recently, these issues have resurfaced, as the Court has decided a rash of cases mandating that […]

Visiting Judges

Despite the fact that Article III judges hold particular seats on particular courts, the federal system rests on judicial interchangeability. Hundreds of judges “visit” other courts each year and collectively help decide thousands of appeals. Anyone from a retired Supreme Court Justice to a judge from the U.S. Court of International Trade to a district […]

The New World of Agency Adjudication

In 1946, the Administrative Procedure Act (APA) set forth the criteria for “formal” adjudication, requiring an administrative law judge to make the initial determination and the agency head to have the final word. That is the lost world. Today, the vast majority of agency adjudications Congress has created are not paradigmatic “formal” adjudications as set […]

Disenfranchisement in the US Presidential Nomination Process Through Caucuses and the Gatekeeping Role of Iowa and New Hampshire

This Note examines inequities in the presidential nomination process. The nomination process has developed such that African American and women voters, compared to white male voters, wield less influence over which candidates parties nominate. By examining data from recent elections and scholarship from the fields of law, political science, and economics, this Note illuminates the […]

Big Data and the Non-Horizontal Merger Guidelines

Data is increasingly valuable as a product, input, and market tool. Exclusive data may be the most valuable asset many firms possess. Yet, regulators in the United States often overlook the importance of data-related mergers, especially between firms that do not directly compete. This is in part because the Non-Horizontal Merger Guidelines (“NHMG”) are out […]