The Future of Music Copyright Collectives in the Digital Streaming Age

Copyright collectives are critical to the economic health of the music industry, but they are at a curious crossroads. Collective copyright management is used more extensively in the music business than ever before. Expanded collective copyright management for digital streaming is the centerpiece of the Music Modernization Act (MMA)—the most extensive revision to the Copyright Act in two decades. At the same time, major music publishers, who rely heavily on collective licensing revenue, are on a years-long mission to end collective licensing for certain digital streaming rights. These trends reflect changes that streaming technology has caused in music consumption, distribution, and revenue generation. Digital streaming has emerged as the dominant music consumption model, accounting for eighty-three percent of music revenues in the United States in 2020.1 This rapid rise to dominance naturally has profound implications for the future of music licensing. The licensing needs of streaming service providers are unprecedented in scale. Spotify, for example, currently hosts over 70 million recordings, with more than 60,000 new recordings uploaded every day.2 Most of these recordings encompass two copyrighted works that must be licensed separately: a copyrighted sound recording and a copyrighted underlying musical composition. Streaming services’ need for such a massive number of licenses highlights the value of collectives that enable streaming services to interface with a manageable number of licensors. It also highlights the importance of blanket licenses that permit spontaneous use of millions of works relatively free from infringement liability. At the same time, the importance of collective licensing to copyright owners has decreased in the streaming age. Streaming is a highly concentrated market: Spotify, Apple Music, and Amazon Music together control two-thirds of the global streaming


INTRODUCTION
Copyright collectives are critical to the economic health of the music industry, but they are at a curious crossroads. Collective copyright management is used more extensively in the music business than ever before. Expanded collective copyright management for digital streaming is the centerpiece of the Music Modernization Act (MMA)-the most extensive revision to the Copyright Act in two decades. At the same time, major music publishers, who rely heavily on collective licensing revenue, are on a years-long mission to end collective licensing for certain digital streaming rights. These trends reflect changes that streaming technology has caused in music consumption, distribution, and revenue generation.
Digital streaming has emerged as the dominant music consumption model, accounting for eighty-three percent of music revenues in the United States in 2020. 1 This rapid rise to dominance naturally has profound implications for the future of music licensing. The licensing needs of streaming service providers are unprecedented in scale. Spotify, for example, currently hosts over 70 million recordings, with more than 60,000 new recordings uploaded every day. 2 Most of these recordings encompass two copyrighted works that must be licensed separately: a copyrighted sound recording and a copyrighted underlying musical composition. Streaming services' need for such a massive number of licenses highlights the value of collectives that enable streaming services to interface with a manageable number of licensors. It also highlights the importance of blanket licenses that permit spontaneous use of millions of works relatively free from infringement liability.
At the same time, the importance of collective licensing to copyright owners has decreased in the streaming age. Streaming is a highly concentrated market: Spotify, Apple Music, and Amazon Music together control two-thirds of the global streaming together rights to be administered and licensed by a single intermediary. The intermediary, a licensing collective, is empowered to negotiate and collect licensing fees from users and is obligated to redistribute those fees to its members after deducting administrative costs. Because collective licensing is voluntary, users require the copyright owner's permission to make the desired use. Rather than licensing directly from the copyright owner, however, the user obtains the license from the collective to which the owner has delegated the right to negotiate and administer licenses.
Direct licensing. In direct licensing arrangements, individual copyright owners directly license rights to users without licensing through a collective intermediary.
Compulsory licensing. A compulsory license enables one to use another's intellectual property without having to obtain the owner's permission, though availability of the license may be conditioned on the user's compliance with statutory formalities such as provision of notice and payment of a set fee. 8 An example of a compulsory license is the traditional "mechanical" license established by Section 115 of the Copyright Act, under which anyone who satisfies the statutory requirements may make and distribute recordings of any song that has been previously recorded and lawfully distributed in the United States. Per § 115(a), if you wish to record and distribute your own version of the song "Rainbow in the Dark," a hit song in the United States in 1983 by the band Dio, you can do so lawfully without permission from the musical composition's copyright owners as long as you satisfy the formalities specified in § 115 (a)-(b). In short, compulsory licenses are involuntary licenses. 9 Blanket licensing. A blanket license authorizes the licensee to use any of the works in the licensor's catalog an unlimited number of times within the term and scope of the license without having to obtain permission for any individual use. Blanket licenses may be compulsory or voluntary. 10 Some licensing collectives issue blanket licenses, but not all do. The Harry Fox Agency (HFA), for example, is a licensing collective that issues per-song rather than blanket licenses. 11 Collective management. Collective copyright management is an umbrella term referring to rights administration on behalf of many copyright owners. Entities that engage in collective licensing are a subset of collective management organizations (CMOs). Some CMOs have the authority to license rights as well as administer them, while other CMOs do not grant licenses but merely provide administrative services- 9. Some compulsory licenses are administered by collectives. For example, the compulsory mechanical license for streaming created by the MMA is administered by the MLC, a collective copyright management organization. In such cases, the collective administers the statutory license, including royalty collection and distribution, but does not grant licenses since permission to use the works is granted by statute, not by the collective or copyright owner.
10. The blanket compulsory mechanical license for the digital distribution and streaming of musical works created by the MMA is an example of a blanket compulsory license. 17 U.S.C. § 115(a)(1)(A)(ii).
11. In recent years, however, HFA has developed procedures for issuing "bulk licenses" for largescale digital uses. See BRABEC & BRABEC, supra note 5, at 85. COLUMBIA JOURNAL OF LAW & THE ARTS [45:1 that is, they track and report uses of copyrighted works and collect and distribute payment for those uses to copyright owners. For example, SoundExchange administers collection and distribution of royalties paid on certain digital public performances of sound recordings. However, SoundExchange has no authority to grant or withhold licenses for the digital performances it administers as such authority is granted via a statutory (compulsory) license.

II. DIGITAL MUSIC RIGHTS SUBJECT TO COLLECTIVE COPYRIGHT MANAGEMENT IN THE UNITED STATES
The consumption of a digital music file online, whether by streaming or download, potentially involves two separately copyrightable works: the sound recording and the underlying musical composition. Not all recordings involve musical compositions that are subject to copyright. In the case of many classical recordings, for example, the recording is of recent vintage and is subject to copyright protection, but the underlying work-say, a Bach violin concerto-has long been in the public domain. Still, most sound recordings accessed via online streaming or download services encompass a copyrighted recording and a copyrighted underlying composition. Collective copyright management plays a role in managing the rights related to both works.

A. RIGHTS INVOLVED IN MUSIC STREAMING
Section 106 of the Copyright Act lists the rights that vest in the author of a copyrighted work. 12 Digital music business models generally involve three of these rights: the reproduction right, the distribution right, and the public performance right. 13 Practically speaking, only the reproduction and public performance rights are relevant for licensed digital music delivery models, since the "distribution" of a digital file via streaming or download occurs simultaneously with the performance or reproduction, and therefore distribution rights are generally not compensated separately from performance or reproduction rights. However, the type of digital delivery model will dictate whether performance or reproduction rights are implicated and how those rights are procured and administered.
Two general modes of digital music delivery predominate in the United States: digital downloads and streams. 14 As I detail below, streaming music services come 12. 17 U.S.C. § 106. 13. The reproduction and performance rights for sound recordings are narrower than those of other types of copyrighted works-including musical compositions-in two ways. First, the sound recording reproduction right is limited to the mechanical reproduction of the recording. 17 U.S.C. § 114(b). Thus, independently made sound-alike recordings that mimic the original but do not mechanically reproduce the original do not infringe the sound recording reproduction right. Second, the sound recording performance right is limited to "digital audio transmission [s]." 17 U.S.C. § § 106(6), 114(a). Thus, the sound recording performance right extends to online digital streaming models but does not extend to terrestrial broadcasts and other nondigital performances of the sound recording.
14. This is of course a vast oversimplification of the number and variety of business models in the ever-evolving digital music delivery market. The point here is not to catalog all business models and 2021] THE FUTURE OF MUSIC COPYRIGHT COLLECTIVES 7 in two varieties (interactive and noninteractive), and the process of procuring and paying for the necessary rights involves a different set of procedures and players for each model. Permanent digital downloads involve digital delivery of files that are not consumed by the user in real time but instead are permanently saved to the user's device for unlimited future access in exchange for a one-time, up-front fee. (This is distinguished from the temporary or "tethered" downloads that are permitted by some streaming services for offline listening.) Apple's iTunes Store still offers permanent digital downloads for sale at the time of this writing, as do some digital music stores that cater to audiophiles by selling higher bitrate music files, such as HDtracks and 7digital. Digital streaming, by contrast, involves the delivery of music files to be consumed in real time and not saved for future use. Digital streaming businesses generate revenue through subscription and advertising models. U.S. copyright law treats streams and downloads differently. Digital download business models involve creating a permanent copy of the sound recording on the user's device, thereby implicating the reproduction right of both the sound recording and the underlying musical composition. 15 Because a copy is made for future use but the work is not "performed" for the user at the time of purchase, downloads do not implicate the public performance right of either the sound recording or the musical composition. 16 By contrast, although any stream technically involves a download, streaming services are designed for the immediate, real-time performance of the work for the user, with no data permanently saved to the user's device. Streamed content is therefore deemed "performed" for the user, thereby implicating the public performance right of both the sound recording-technically, the right of public performance by digital audio transmission 17 -and the underlying musical composition.
Some streaming business models function as substitutes for sales of physical or digital copies. 18 The law treats them accordingly and requires operators of streaming services that give the users substantial control over what they hear-and are therefore gradations thereof but rather to map out at a high level the legal landscape in which CMOs operate and collective licensing practices occur.
15. The reproduction right for the underlying musical composition in a sound recording has a unique history and nomenclature: It is called a "mechanical" right because it developed as the right to be paid for reproductions of the composition in media that enable mechanical performance of the composition as sound, thus extending the musical composition's reproduction right beyond reproduction as sheet music. the user has sufficient control over the interactive service such that she can predict the songs she will hear, much as she would if she owned the music herself and could play each song at will, she would have no need to purchase the music she wishes to hear. Therefore, part and parcel of the concern about a diminution in record sales is the concern that an interactive service provides a degree of predictability-based on choices made by the user-that approximates the predictability the music listener seeks when purchasing music."). COLUMBIA JOURNAL OF LAW & THE ARTS [45:1 more likely to replace sales-to license the sound recording performance right and the musical composition reproduction right (i.e., the mechanical right). The law therefore distinguishes between two types of streaming services: interactive and noninteractive. 19 Interactive streaming services are those that enable users to "receive transmission of a program specially created for the recipient." 20 In other words, interactive streaming services provide users with on-demand functionality: The user can search for and immediately play the desired recording. Currently popular interactive streaming services include Spotify, Apple Music, and Amazon Music. Noninteractive streaming services, by contrast, substantially limit user control. Once the user selects a general stylistic orientation ("smooth jazz" or "artists that sound like Pink Floyd"), the service dictates which songs are played. The original version of Pandora is an example of a popular noninteractive streaming service. In the case of noninteractive streaming, the user has minimal control over what is played, and such services are not considered to be a direct substitute for sales of copies. Noninteractive streaming services therefore do not implicate the reproduction right of the underlying musical composition. YouTube is another example of a major music streaming service. Indeed, it is the biggest streaming music provider in the United States. 21 Because YouTube streams audiovisual works rather than audio alone, YouTube cannot use any of the compulsory licenses available to certain music streaming services. This means that YouTube must (and does) negotiate with publishers and PROs to license the music it uses, 22 subject to a very large wrinkle: Because the content uploaded to YouTube is hosted not at YouTube's initiative but rather at the direction of users, YouTube is eligible for immunity from secondary infringement liability even if it fails to secure licenses. As a passive platform for user-uploaded content, YouTube operates within the scheme of § 512(c) of the Copyright Act, which limits the potential liability of service providers that host content at the direction of users so long as the service provider meets certain statutory requirements. These requirements include implementation of a system by which copyright owners may notify the service provider of infringements on its site and the service provider "expeditiously" removes the identified content upon receipt of such notice. 23 This scheme-often called the Digital Millennium Copyright Act (DMCA) "notice and takedown" regime-involves many nuances beyond the scope of this Article. But the key point is that so long as YouTube and similar services host content at the direction of users (and not at their own initiative) and comply with the requirements of § 512, then such services need no licenses at all. Nevertheless, YouTube does obtain some licenses 19 necessary to effectively operate as an interactive streaming service. That said, some copyright owners complain that YouTube executives use the § 512(c) "safe harbor" as negotiating leverage to keep licensing fees below market rates. 24 In sum, the law requires digital download services to procure licenses for the reproduction of the sound recording and the underlying musical composition. Noninteractive streaming services are required to procure licenses for the public performance of the sound recording and the underlying musical composition. Interactive streaming services are required to procure licenses for the public performance of the sound recording and the underlying musical composition, as well as for the reproduction of the underlying musical composition. 25 To facilitate new business models and reduce transaction costs, most of these licenses may be obtained via compulsory licenses or blanket licenses. The following paragraphs outline whether compulsory or blanket licenses are available for the various rights related to sound recordings and musical works, and if so, which CMOs administer the licenses.
Sound recording rights. Sound recording rights are required for both types of streaming services as well as digital download services. Noninteractive streaming services may obtain the sound recording digital public performance right via compulsory license. 26 Revenues generated by the sound recording digital public performance right compulsory license are collected and distributed to copyright owners by SoundExchange, a CMO designated by Congress for that purpose. Interactive streaming services cannot use this compulsory license and must directly license the required sound recording performance rights from the sound recording copyright owners (typically record companies). 27 Digital download services also cannot use this compulsory license because it covers only performances and not reproductions of the sound recording. Instead, digital download services must directly license the right to reproduce sound recordings from sound recording copyright owners.
Musical composition rights. The rights relevant to streaming the underlying musical compositions are the public performance right and the reproduction (mechanical) right. Musical composition public performance rights are required for both interactive and noninteractive streaming services (but not for download services). These rights are not subject to compulsory licenses but may be obtained 24 , which went into effect in January 2021, created a compulsory blanket license that enables any interactive streaming provider to obtain mechanical licenses for copyrighted musical compositions underlying any sound recording that the service is lawfully authorized to distribute. 28 For digital downloads, record labels typically bundle the mechanical rights for musical works encompassed in their recordings with the sound recording reproduction right licensed to digital download services. The labels then pay a percentage of their royalties from downloads to publishers. However, the MWMA permits download services to use the compulsory blanket license for mechanical rights should they so choose. 29 The blanket mechanical compulsory licensing process is managed by the MLC, a nonprofit CMO authorized under the MWMA to issue and administer the blanket compulsory licenses, collect and distribute royalties, maintain a comprehensive database of musical work copyright ownership, identify and locate copyright owners of musical works, and "enforce rights and obligations," among other powers and duties. 30

B. DEVELOPMENT OF THE MMA'S BLANKET COMPULSORY MECHANICAL LICENSE FOR STREAMING
The blanket mechanical compulsory license and the MLC were designed to remedy some long-festering pain points between music publishers and digital streaming service providers. In the early days of digital streaming, it was uncertain whether digital streams constitute a performance of the musical work, a reproduction of the musical work, or both. If streams are deemed to be only performances of the musical work and not reproductions, then streaming services could fulfill their obligations to publishers by simply acquiring blanket performance licenses from the PROs. In that case, streaming services would not be obligated to pay publishers a licensing fee for the reproduction of the musical work (i.e., a "mechanical" license). Publishers predictably argued that a stream is simultaneously a performance and a reproduction and that streaming services should compensate publishers for the use of both rights. 31 Streaming services, however, argued that such a requirement would constitute double-dipping by publishers-i.e., that a stream is either a performance or a reproduction, and payment for one right or the other makes publishers whole.
The problem with the streaming services' argument is that, in the digital streaming era, streaming services increasingly replace revenue streams from broadcasts (implicating performance rights) and physical sales (implicating the reproduction/mechanical right). This dispute was partially resolved in 2008 when interactive streaming companies entered into a settlement agreement concerning streaming and mechanical rights, the terms of which the Copyright Royalty Judges adopted as final regulations. 32  to publishers as a percentage of total revenues, less amounts owed for performance royalties. 33 However, there were immense logistical challenges associated with licensing a new music delivery paradigm that exploded out of the gate before the new licensing framework was settled. At the time, the mechanical reproduction right in musical works was not subject to blanket licensing, so licenses had to be obtained individually. Prior to the streaming era, the mechanical right was subject to compulsory licensing upon compliance with statutory requirements, but relatively few people used it. Most publishers authorized the Harry Fox Agency (HFA)-a private party-to issue direct licenses to artists and labels that released new cover versions or reissues of previously released songs in the publishers' catalog. When streaming models came online, publishers and streaming services naturally employed HFA to handle the administration of mechanical licenses, given the agency's experience and capabilities in the field of mechanical licensing. But this new endeavor for HFA involved the complex-and entirely novel-tasks of administering bulk mechanical licenses to interactive streaming services and collecting and distributing the resulting mechanical royalties.
This was a flawed, stop-gap solution. By 2015, it became evident that a large percentage of mechanical royalties generated by streaming services-allegedly as much as twenty-five percent-went unpaid. 34 Streaming services claimed to be unable to locate the owners of many songs (even songs written by superstars), so the services withheld mechanical royalty payments and persisted in the use of millions of musical works by filing bulk "notices of intent" with the Copyright Office. 35 Although streaming services' "inability" to locate copyright owners was partly selfserving, it is also true that ownership data problems plague the music publishing industry. 36 Neither HFA nor any other entity in the music industry had a comprehensive, centralized, authoritative database of ownership information that could be used to match all sound recordings with the proper right holders of the underlying compositions. 37 Copyright owners alleged HFA failed to pay-or significantly delayed the distribution of-mechanical royalties it had collected from streaming services, leading to pointed criticism of the agency. 38 Compounding matters, the copyright owners of twenty percent of the songs on services such as Spotify did not even use HFA, making it even more challenging to match royalties 33 For copyright owners and collectives, the conflict over mechanical licenses was about revenue stream stability. For a century, the modern music publishing industry was built on three pillars of income: performance royalties, synchronization royalties (the use of music in audiovisual works such as films, television programs, and commercials), and mechanical royalties. 42 Two of those-performance and mechanical royalties-are involved in audio streaming models, and losing either performance or mechanical royalties as a revenue stream would be devastating for publishers and songwriters. 43 So publishers were adamant that streaming revenues adequately replace any existing income sources diminished by streaming. Of course, publishers are not the only intermediaries from the pre-streaming era reliant on established revenue streams. Whether the new revenue streams are called "performance royalties" or "mechanical royalties" might not matter much to publishers if the aggregate revenues are adequate, but the nomenclature was of existential importance to PROs and HFA, which, respectively, specialized in collections for performance and mechanical royalties in particular. In at least one commentator's view, therefore, the fight over whether streaming services owed mechanical royalties in addition to performance royalties boiled down to a turf war between the PROs and HFA. 44 Regardless of various stakeholders' motives, by the late 2010s the market for interactive streaming had become sufficiently mature and lucrative that all sides needed clarity and resolution. The solutions developed during the 2010s were patchwork, ad hoc responses to a quickly evolving market. The impromptu solutions supplied sufficient grease to keep the streaming ecosystem from grinding to a halt, but too much uncertainty and potential liability remained for those ad hoc agreements and procedures to provide a sustainable framework. Copyright owners wanted certainty regarding the extension of mechanical rights to interactive streaming, and they wanted mechanisms for ensuring accountability and fair royalty rates. Interactive streaming services wanted certainty regarding permissions and payment distribution: If a streaming provider received authorization from the copyright owner of a sound recording to distribute the recording, the streaming provider wanted assurance it would not be liable to unknown or unidentified owners of composition copyrights. 39 Congress's answer via the MWMA was the blanket compulsory license for streaming mechanical rights and the establishment of the MLC to administer it. 45 The compulsory license provides peace of mind to publishers and songwriters by codifying the right to remuneration from mechanicals in the interactive streaming context. The law also directs the Copyright Royalty Board (CRB), the administrative tribunal that sets compulsory license rates in the absence of voluntary agreement between the parties, to employ a "willing buyer, willing seller" standard to reduce the potential for compulsory licenses to undervalue the mechanical right. 46 In return, the streaming services receive the option of a blanket compulsory license that provides a shield against future mechanical right infringement claims. 47 Further, establishment of the MLC is intended, once and for all, to remedy the composition copyright ownership data deficiencies by establishing an authoritative, centralized, publicly accessible ownership database. While the database will never be entirely complete or accurate, copyright owners are incentivized to make it as comprehensive and accurate as possible. The immunity that the blanket compulsory licensing regime confers on service providers puts the onus on copyright owners to register their works with the MLC if they wish to be remunerated. This doubtless comparatively disadvantages independent artists who lack the resources of large publishing companies. 48 The flip side, however, is that registering and collecting through the MLC is entirely free to copyright owners, as the MLC is bankrolled by the digital service providers (DSPs). 49

C. COMPULSORY LICENSES AS A RESPONSE TO MARKET TRANSFORMATION
To navigate licensing uncertainty in uncharted technological waters, the MWMA's blanket compulsory mechanical license for streaming services and its establishment of the MLC employ a combination of time-honored devices in copyright jurisprudence: compulsory licenses and collective copyright management. Indeed, the original compulsory mechanical license established under the Copyright Act of 1909 arose in a context that bore similarities to the conflicts over streaming. In the early twentieth century, the copyright owners of musical works-generally, music publishers-relied on one revenue stream: royalties from the sale of printed sheet music. (The public performance right in musical works was nascent, having just been added to the Copyright Act in 1897.) 50

2021] THE FUTURE OF MUSIC COPYRIGHT COLLECTIVES 15
developed that threatened to erode sheet music royalties: player pianos. 51 Player pianos were automated devices that played themselves with the aid of paper rolls inserted into a mechanism embedded in the piano, which would operate the piano keys according to the information recorded as hole punches on the roll. With no human player, there was no need to buy sheet music. Publishers faced a major threat if the exclusive right to reproduce a musical work did not include the right to reproduce it in forms beyond sheet music-that is, in forms in which the work could be performed by a mechanism rather than by a human. 52 With the arrival of player pianos and contemporaneous rapid advances in recording technologies, one could already imagine a world in which sales of copies in machine-playable form would outnumber sales of sheet music. 53 But uncertainty clouded the extent and nature of the publishers' rights in these uncharted technological waters. Did the copyright owners' exclusive right to reproduce a musical work encompass only humanreadable formats, or did it extend to the reproduction of the work in machine-readable formats? Lawsuits ensued, ultimately resulting in a Supreme Court decision against the publishers. In White-Smith Music Publishing v. Apollo, the Court held that the text of the Copyright Act did not support extending copyright to machine-readable-only formats. 54 A year later, in 1909, Congress reversed the decision, amending the Copyright Act to extend the reproduction right in new compositions to "any form of record in which the thought of an author may be recorded and from which it may be read or reproduced," including by "controlling the parts of instruments serving to reproduce mechanically the musical work." 55 In the same provision, however, Congress subjected this new right to a compulsory license: Once the copyright owner produced and distributed-or licensed others to produce and distribute-the musical works in machine-readable form, anyone could do the same so long as they met statutory requirements, including paying a predetermined fee per copy sold. Unlike the compulsory license established by the MWMA, early compulsory mechanical licenses were individual and not issued as blanket licenses.
The Interestingly, when Congress created the digital performance right for sound recordings in 1995, it sought this balance with respect to noninteractive streaming by creating a compulsory license in that context but denied interactive streaming services access to the compulsory license. Nevertheless, interactive streaming has come to dominate the streaming market, 59 suggesting there is some limit to copyright owners' leverage even in a nascent market enabled by transformative technology. As I discuss in the next Part, music publishers have been hoping to similarly directlicense musical composition performance rights in the streaming context. If they do, streaming services might lobby Congress to create a compulsory performance license for musical works. In Part IV, I suggest that the successful development of streaming services despite the lack of a compulsory license for sound recording performance rights is evidence that a functioning market for direct licenses is possible and a compulsory license may be unnecessary.

III. COLLECTIVE LICENSING AND STREAMING: PERFORMANCE RIGHTS ORGANIZATIONS (PROS)
As noted in Part I, collective licensing involves voluntary licensing through a collective that sets licensing fees for members' works. In the streaming context, collective licensing is used to license performance rights in streamed musical works (that is, the right to perform the song underlying a recording).

A. PRO COLLECTIVE LICENSING AS A SOLUTION TO TRANSACTION COSTS
The practice of collective licensing would not be so deeply entrenched in the modern music industry were its advantages not remarkable. The primary advantage of collective licensing is that it markedly reduces transaction costs for rights holders as well as music users. 60 Millions of businesses publicly perform music daily in the United States: digital audio and video streaming services, radio stations, stores, malls, bars, restaurants, dance schools, ice rinks, gyms, bowling alleys, airports, airlines, hotels, office buildings, places of worship, and countless other businesses. Musicians "cover" popular songs in live public performances at thousands of coffee shops, bars, restaurants, theaters, and concert arenas across the country nightly. Many of these performances would infringe if unlicensed. Historically, it would have been impracticable for every business and organization to negotiate with myriad copyright owners for the rights to perform each individual work. It would be an even more Herculean task (and prohibitively expensive) for copyright owners to negotiate and execute licenses with every user across the globe and collect and administer the licensing fees-often micropayments-on a per-work basis. (As I discuss in Part IV, however, technology is dramatically reducing these transaction costs. Direct licensing is far more feasible now than in the pre-streaming era-raising some doubts about the value that PROs add.) Collective licensing organizations reduce these transaction costs by creating a one-stop clearinghouse for rights. The major PROs in the United States are the American Society of Composers, Authors and Publishers (ASCAP), Broadcast Music, Inc. (BMI), and SESAC (originally the Society of European Stage Authors and Composers). Members (music publishers and songwriters) grant the PRO the non-exclusive right to license nondramatic public performances of all works to which the member has rights, so long as the licensor remains a member. 61 Each PRO amasses a large catalog of works from thousands or millions of members (combined, ASCAP and BMI represent nearly 25 million songs on behalf of more than 2 million members) and grants music users (venues, broadcasters, streaming services, and so on) a blanket license to use all represented works an unlimited number of times for an annual fee. 62 Many countries have national collection societies that license and collect on behalf of those countries' publishers and songwriters. Those societies, in turn, have reciprocal agreements with sister collection societies across the globe, including the U.S. PROs. 63 The end result is remarkable transactional efficiencies on a global scale. A U.S.-based songwriter member of ASCAP, for example, might open their ASCAP statement to see an accounting of a song's performance royalties from dozens of countries all combined into one royalty check.
In short, members grant the PRO a "portfolio" license and the PRO, in turn, grants its users an "all-you-can-eat" blanket license to perform the entire corpus of works that the PRO has licensed from its members. Publishers need only license their works to one entity, and venues and broadcasters need only deal with one entity to acquire the performance rights. 61. See, e.g., ASCAP Publisher Member Agreement, at 1, https://perma.cc/Z522-YPTN. 62. BRABEC & BRABEC, supra note 5, at 468; BMI, About, https://perma.cc/9W7C-PXPD (last visited Nov. 14, 2021) (reporting that BMI represents more than 1.2 million members); ASCAP, About Us, https://perma.cc/B4GM-CLZP (last visited Nov. 14, 2021) (reporting membership of more than 850,000 music creators In practice, the previous sentence is an oversimplification, at least in the United States: Although PROs markedly reduce transaction costs, they are not quite onestop shops for the venues and broadcasters that publicly perform musical works. No single PRO has the right to license all the potential works that a venue or broadcaster might perform, nor does any single PRO catalog contain all of the popular works most likely to be performed. Nevertheless, some users obtain a license from only one of the PROs. This is a risky strategy for such businesses. It would be difficult for any licensee to ensure that the works it performs are limited exclusively to works in the single PRO's catalog. The challenging task of divining the subset of all copyrighted musical works available through a single PRO, and then limiting oneself thereto, is compounded by the practice of fractional licensing. That is, some works in a PRO's catalog have multiple copyright owners and, for such works, the PRO may represent some, but not all, owners of the work. A single PRO with fractional rights cannot unilaterally license the right to perform the work. 64 The upshot is that streaming services, broadcasters, and businesses seeking maximum flexibility and immunity from liability must obtain blanket licenses from four PROs. 65 In addition to ASCAP, BMI, and SESAC, a fourth PRO-Global Music Rights (GMR)-has recently emerged on the scene. GMR has a comparatively small catalog (about 55,000 songs at the time of this writing), but it focuses on representing mega-hits from contemporary and legacy superstars including John Lennon, Bruce Springsteen, Don Henley, Prince, Drake, and Bruno Mars. Unlike ASACP, BMI, and SESAC, GMR is the exclusive licensor of performance rights for the songs in its catalog, which includes many of the most popular songs in pop and rock music history.
The public also benefits from the efficiencies that collective licensing engenders. Collective licensing facilitates fast, mostly frictionless transactions that give consumers of live, broadcast, and streaming music immediate access to an enormous variety of licensed music. Consumers need not wait out protracted negotiations between individual venues, broadcasters, or webcasters and myriad copyright owners before their favorite music becomes accessible. 66 64. See United States v. Broad. Music, Inc., 720 F. App'x 14, 18 (2d Cir. 2017). 65. There are exceptions to this general rule of thumb. As discussed later in this Article, ASCAP, BMI, and SESAC permit their members to directly license public performances of their works. Broadcasters and venues can, therefore, negotiate direct licenses from copyright owners and can even, under certain circumstances, reduce their PRO blanket licensing costs to account for direct-licensing fees paid for works in the PROs' catalogs. See BRABEC & BRABEC, supra note 5, at 469 (discussing adjustable fee/carve-out blanket licensing arrangements). This practice remains the exception rather than the rule, but I argue in Part IV that direct licensing of performance rights will play an increasingly important role in the future. Additionally, venues may be exempt from licensing public performances of radio and television broadcasts based on the size and type of venue, the sophistication of the equipment used to play the broadcast, and whether the establishment directly charged patrons to see or hear the broadcast. 17 U.S.C. § 110(5)(A)-(B).
66. For a discussion of the downsides of collective licensing, however, see Parts III.C-D, infra.

B. OTHER ADVANTAGES OF COLLECTIVE LICENSING FOR COPYRIGHT OWNERS: REDUCED ADMINISTRATIVE AND ENFORCEMENT COSTS
Collective licensing is about leveraging scale. A group of copyright owners pooling resources can accomplish far more, far more efficiently, than thousands of copyright owners could do individually. 67 In addition to spreading transaction costs, collective licensing spreads administrative costs, reducing the individual costs of copyright administration for all members. The tasks of monitoring usage of copyrighted works by thousands of live venues, broadcasters, and streaming services, billing for that usage, collecting and distributing those revenues, and performing the related accounting functions are daunting for any publisher, not to mention an unrepresented songwriter. Licensing collectives enable copyright owners to outsource these tasks to the collective and share the costs industry wide.
Collective licensing similarly reduces copyright owners' enforcement costs. Indeed, the necessity of sharing enforcement costs was a major reason ASCAP was established. 68 Few venue owners or broadcasters would bother to purchase licenses for performance rights-an arcane concept to many venue owners-without a credible threat of litigation. But venues and broadcasters are too many and too widely dispersed for copyright owners to effectively monitor unlicensed performances everywhere or bring a credible threat of enforcement to every wouldbe infringer. Thus, an important role of PROs is to initiate actions on behalf of their members against unlicensed venues and broadcasters across the country. 69 Individual songwriters and smaller music publishers, for whom enforcement costs would be prohibitive, especially benefit from consolidated enforcement.
The result is that collectives provide these highly specialized, cost-and laborintensive services for a fraction of what it would cost any given copyright owner to establish comparable services in house. Still, administrative fees are increasingly a point of contention between PROs and publishers. 70 ASCAP and BMI reportedly charge members between ten and twelve percent of gross revenue, but publishersincreasingly capitalized by Wall Street investment firms-are becoming more proactive about scrutinizing PRO inefficiencies, expenditures, and payment practices. 71

C. COLLECTIVE BLANKET LICENSING, MARKET POWER, AND COMPETITION REGULATION
Collective licensing is about leveraging scale, but scale can engender excessive market power. 73 If one PRO held the right to license ninety percent of the hit songs on the market, and that entity's entire catalog was licensed as a bundle, few broadcasters or venues could survive without acquiring that bundle even if they are knowingly paying for the right to use songs they will never perform. An entity that represents the majority of desirable (and theoretically competing) songs can engage in monopoly pricing. This is especially true regarding the collective blanket licensing model, in which members empower the collective to bundle works and set the bundle's price. 74 Given the pricing power that blanket collective licensing affords PROs, it is small wonder that antitrust concerns have persistently dogged PROs since their inception.

ASCAP, BMI, and DOJ Consent Decrees
The market power-conferring characteristic of collective blanket licensing has profoundly shaped performance rights licensing in the United States for the past century. ASCAP, the first PRO, was established in 1914. 75 By the 1920s, ASCAP licensed eighty percent of the music played on the radio, and by the 1930s, its market power was so extensive that the Department of Justice (DOJ) sought to break it up as an unlawful combination. 76 A DOJ suit was ultimately dropped, and in 1939 ASCAP attempted to use its market leverage to double radio broadcasters' licensing fees in one stroke. 77 Broadcasters revolted en masse, founding BMI as a broadcasterfriendly alternative to ASCAP. 78 Although the formation of BMI in 1939 (and SESAC in 1930) created competition for ASCAP in the collective licensing marketplace, the anticompetitive threat remained sufficiently concerning that, in 1941, the DOJ sued both ASCAP and BMI for anticompetitive conduct under § 1 of the Sherman Act. 79 These actions precipitated a settlement with the DOJ that resulted in consent decrees that regulate ASCAP and BMI to this day in order to limit the collectives' market power.
The consent decrees, which are periodically reviewed and amended, regulate ASCAP and BMI's licensing practices as well as the organizations' relationships 73 with their members. To limit the PROs' pricing power, the decrees require ASCAP and BMI to offer blanket licenses for "reasonable fees." 80 The consent decrees establish federal rate courts, which set fees when ASCAP or BMI and a licensee are unable to agree on a fee. Importantly, given the market power that exclusive licenses confer, especially when bundled together, the consent decrees limit ASCAP and BMI to acquiring and granting licenses on a non-exclusive basis. This means that members of ASCAP or BMI retain the right to directly license public performance rights themselves. The consent decrees also restrain ASCAP and BMI by limiting the type of licenses they may issue to public performance rights for musical works, thus barring the organizations from granting mechanical licenses, sound recording digital performance rights, and synchronization licenses.
Antitrust law therefore acts as a limitation on the scope of licensing collectives' pricing and negotiating power. The two smaller PROs-SESAC and GMR-have never been subject to consent decrees, though both have been accused by broadcasters of antitrust violations. SESAC, to settle antitrust litigation brought by radio and television broadcasters, agreed to enter mandatory arbitration when licensing negotiations fail. 81 Like ASCAP and BMI, SESAC only obtains nonexclusive licenses from its members, presumably to minimize antitrust concerns. 82 At the time of this writing, GMR has no such constraints, and obtains exclusive performance rights licenses from its members. 83 Nevertheless, in theory the background threat of antitrust litigation helps keep both smaller PROs' negotiating leverage in check (as evidenced by SESAC's voluntary agreement to submit to arbitration on pricing disputes). 84 Ivan Reidel argues that, in theory, the competitive harms from PROs' "one-sizefits-all" blanket licenses extend not just to music users but also to most songwriters by restricting competition in the market for songs. 85 Because blanket licensing, not direct licensing, is the norm in the music performance licensing space, radio stations only have the option of purchasing expensive blanket licenses, which include works by superstars as well as lesser-known songwriters. 86 Under such circumstances, licensing costs for radio stations are high and writers of lesser-known songs have few opportunities because stations are bound to play only the most popular songs to cover 80 87 To cover the licensing fees, stations also must devote more time to advertising, further reducing the airtime available for lesser-known songs. 88 If lesser-known songwriters were free to negotiate prices with radio stations, competition in the market for songs would increase because lesserknown songs could be offered at a lower rate. 89 The competition, Reidel argues, would lower radio station licensing costs, result in fewer advertisements, create more opportunities for a larger number of songwriters and artists, and enhance the diversity of music on the radio. 90 Glynn Lunney argues a collection society PRO model would provide the benefits of reduced transaction costs while reserving pricing power to the copyright owners, thereby preserving pricing competition that ultimately benefits users and copyright owners. 91 As Reidel's and Lunney's arguments-and the history of PRO regulationillustrate, there can be serious downsides to collective licensing, particularly in the case of blanket collective licensing whereby the collective takes pricing power and licensing decisions out of the hands of individual owners. It is reasonable, therefore, to ask why copyright owners would delegate such significant authority to PROs. One answer, of course, is the reduction of transaction, enforcement, and administrative costs discussed above. That is not a complete answer, however, because as Lunney points out, one can imagine the PROs structured not as licensing collectives but as collection societies that provide the benefit of one-stop shopping and other efficiencies without offering the blanket licenses that wrest away publishers' pricing power. Rather, it seems that the PRO blanket licensing model persists primarily because the most powerful music licensees prefer it. 92

Music Licensees' Demand for Blanket Licenses
Freedom of use and immunity from infringement are the key attributes licensees seek. In the Supreme Court's words, "Most [music performance rights] users want unplanned, rapid and indemnified access to any and all of the repertory of compositions, and [copyright] owners want a reliable method of collecting for the use of their copyrights." 93 As noted in Part III.A, payment of a fixed fee to each of the four PROs unlocks virtually the entire catalog of the world's copyrighted compositions for unlimited public performances largely free from potential liability.

2021] THE FUTURE OF MUSIC COPYRIGHT COLLECTIVES 23
Blanket licenses also have the benefit of predictability: Broadcasters and venues can budget in advance for annual PRO fees, which do not change dramatically each year. Of course, the degree to which users prefer blanket licenses to direct or perprogram licenses depends on how much music they use. The cost and inflexibility of blanket licenses frustrates occasional users of music. 94 Blanket licenses are designed to be as homogenous as possible, since tailoring increases transaction costs. As a result, the PROs primarily offer "all-you-can-eat" licenses with little room for customization and a limited set of pricing tiers designed to treat all relatively similar users alike. 95 The lack of granularity naturally leads to inefficiencies. The limited menu of options chafes licensees who feel that the available licenses do not fit their usage patterns. 96 Inevitably, some licensees will overpay for music they will not or cannot use. Those who expect to underutilize a license might opt to forgo performing music altogether or take their chances performing unlicensed music. Using unlicensed music "on the down low" might not be an option for large, conspicuous broadcasters. However, it is an option for smaller entities that can more easily fly under the radar such as bars, restaurants, and coffee shops-although there are risks even for smaller users. 97 The use of unlicensed music, or the decision by potential users to forgo music altogether, represents a market failure that decreases the availability of music and results in lost revenue for songwriters and publishers. PROs do offer tiered pricing based on business type, audience size, and expected usage, but many establishments must pay a minimum fee, and licenses are typically sold on an annual basis. 98 Even for small establishments, the combined price for blanket licenses from ASCAP, BMI, SESAC, and GMR could easily exceed $1,500 annually. 99 This is a hefty price for 94. See, e.g., National Religious Broadcasters Music License Committee, Comments in the Matter of ASCAP and BMI Consent Decree Review, at 5 (Aug. 6, 2014).
95. To reduce market power, ASCAP's consent decree also requires ASCAP to offer "per-program licenses" as an alternative to blanket licenses. ASCAP Consent Decree, supra note 79, at 5 ( § VII). A per-program license "authorizes a broadcaster to perform ASCAP music in all of the broadcaster's programs, the fee for which varies depending upon which programs contain ASCAP music not otherwise licensed for public performance." Id. at 2 ( § II(J)

Major Publishers' Push to Direct-License Streaming Services
The homogeneity of PRO blanket licenses-especially under DOJ control-also exasperates publishers. The rapid rise of streaming and the attendant shift in consumption and monetization models has left publishers bridling at the constraints of collective licensing and seeking more direct control. Dissatisfied with the rates that ASCAP negotiated with Pandora, which were subject to rate court review and adjustment, EMI, Sony Music Publishing (Sony), and Universal Music Publishing Group (Universal) threatened in 2010 to withdraw their digital rights from ASCAP. This would force Pandora to negotiate directly with the publishers for performance rights. 100 While ASCAP's internal governing documents did not allow for members' partial withdrawal of rights, fear that the major publishers might fully withdraw their rights prompted ASCAP to amend its governing documents to permit partial withdrawal. With the PROs' consent, then, EMI, Sony, and Universal withdrew from ASCAP and BMI the right to license "new media transmissions by new services" and struck direct licensing deals with Pandora for rates considerably higher than the rates the publishers received through the PROs. 101 Pandora challenged the withdrawals, arguing the ASCAP and BMI consent decrees prohibit selective grants of rights.
The district court agreed with Pandora: The decrees require publishers to be allin or all-out; selective withdrawal is prohibited. 102 The Second Circuit affirmed, holding that ASCAP's consent decree "unambiguously precludes ASCAP from accepting such partial withdrawals." 103 The decree requires ASCAP to license its entire repertory to all eligible users; therefore, "publishers may not license works to ASCAP for licensing to some eligible users but not others." 104 This ruling demonstrates how the consent decrees have shifted from a competition-promoting regime to a form of price control on performance licensing. Allowing publishers to selectively withdraw rights would increase competition in the licensing space. Indeed, creating space for others to license works-and thereby weaken the PROs' market power-is precisely why the consent decrees stipulate that ASCAP and BMI must permit members to directly license works. Because publishers are denied the ability to partially withdraw rights from ASCAP and BMI, however, the PRO fees set by rate courts act as a de facto cap on the direct-licensing rate publishers can negotiate. If the fee exceeds the PRO rate, then the music user can opt for the PRO license instead.
Having lost on the question of partial rights withdrawal, ASCAP and the publishers sought to use the direct licensing rates negotiated between the publishers and Pandora as evidence of the market rate for purposes of setting Pandora's ASCAP and BMI royalty rates going forward. The ASCAP rate court held that the negotiated rates were not fair benchmarks because the publishers had been coercive in their negotiations, largely because, in the court's view, Sony and Universal leveraged their market power as major publishers to extract an unfair price. 105 Interestingly, the BMI rate court found the rates to be reasonable benchmarks for fair market prices. 106 A significant complication concerning the withdrawal of publishers' rights that has received little attention in the scholarly literature or press, but is well known in the music publishing industry, is that U.S. publishers do not have the right to license foreign writers' performance rights. A writer domiciled in the United Kingdom, for example, will typically be a member of Performing Rights Society Ltd. In a competitive, atomistic market, if one of many right holders refuses to share critical information, then the music user can see if a competitor will be more cooperative. Instead, Pandora and Sony operated in a highly concentrated market. . . . Even if Sony had provided the list of its works to Pandora, Sony would have retained enormous bargaining power; by withholding the list, Sony deprived ASCAP of a chance to argue in any persuasive way that the Sony-Pandora license reflects a fair market price. Id. at 359. The rate court judge likewise held that negotiations between Universal and Pandora were not evidence of fair market rates because "there were virtually no meaningful negotiations between Pandora and [Universal]." Central to the judge's reasoning was that Universal "control[s] roughly 20% of the music market," and could therefore demand a rate "that bore no relation to the then-existing market price." Id. at 360. Given [Universal's] bargaining stance, including its unwillingness in Pandora's eyes to proceed in a businesslike manner, Pandora agreed to a contingent, short-term license, and placed its fate in the hands of the ongoing rate court proceedings. In such circumstances, this license rate cannot be said to represent a bargain arrived at by a willing buyer and seller. unaffiliated with Sony (and there are many such songs), Sony arguably has the right as a co-owner to non-exclusively license the work (at least for performance in the United States) but does not have any way to distribute collected royalties to the foreign writer and publisher. Sony does not have an agreement with PRS to act as PRS's agent and may even lack sufficient information about the foreign writers to match the funds. And, if a streaming platform only direct-licenses performance rights from U.S. publishers and skips U.S. PRO licenses, it would not be licensed to play any works written entirely by foreign writers and represented entirely by foreign rights societies. Many legacy and contemporary hits fall into this category. As withdrawal of rights from ASCAP and BMI may be an inevitability, the various stakeholders would surely work out the problem through private ordering, though a period of turbulence and litigation can be expected. Still, the problem was thorny enough that it gave some U.S. publishers pause, I have been told by a music publishing executive, when it came to the question of rights withdrawal. After the major publishers failed in their bid to withdraw digital rights, ASCAP and BMI had reason to fear for their own existence. Although the transaction costs of licensing thousands of bars, businesses, and broadcasters might be prohibitive, licensing a handful of major streaming services is not. 107 By 2014, it was clear that streaming was the biggest growth area in the music performance rights sector. Since 2015, ASCAP has had a compound annual growth rate of six percent, driven by streaming. 108 Publishers were plainly eager to directly partake in that swelling new revenue stream, and there was a danger that they might withdraw all their rights from ASCAP and BMI if the economics of doing so made sense.
If the publishers were to quit BMI and ASACP, it would be to gain direct control over streaming licensing terms and fees, not to remove the PROs as rent-seeking intermediaries. BMI and ASACP are nonprofits that, as noted above, allow publishers to cost-effectively outsource resource-intensive royalty collection and distribution work. While publishers have pressured the PROs in recent years to reduce costs, 109 if the publishers do direct-license streaming platforms, they will probably continue to employ the PROs to handle the related administrative tasks.
Doubtless concerned for their future, in 2014 ASCAP and BMI petitioned the DOJ to review the consent decrees in light of market changes caused by streaming. 110 ASCAP and BMI argued that the eighty-year-old consent decrees shackle the organizations' ability to modernize and innovate at a time when digital technologies 107. U.S. Dep't of Just., Remarks of Assistant Attorney General Makan Delrahim, supra note 92, at 4 ("Rather than a geographically distributed scheme of terrestrial broadcasters, or an assortment of record and phonograph stores, the principal means through which many Americans now get their music is a collection of digital streaming services.").

2021] THE FUTURE OF MUSIC COPYRIGHT COLLECTIVES 27
and streaming have spectacularly altered how people consume and pay for music. 111 Two concerns underlie this desire-if not desperation-for change. First, so long as the consent decrees prohibit partial grants of publisher rights, the threat always remains that publishers will opt to withdraw their entire catalogs to negotiate higher streaming license fees. 112 Second, the consent decrees limit the revenue streams available to ASCAP and BMI almost exclusively to musical work performance rights. 113 As long as that restriction remains, ASCAP and BMI cannot develop new revenue streams, such as synchronization and mechanical rights licensing, to help offset the losses if publishers withdraw their performance rights. 114 Accordingly, ASCAP and BMI sought to have the decrees modified in three ways. First, they wanted the flexibility to allow publishers to grant partial rights. Second, they sought to streamline the fee review process (which, the PROs argued, could drag on for years, during which time ASCAP and BMI would need to grant provisional licenses to petitioning users). Third, they sought to remove restrictions on engaging in licensing practices beyond performance rights licensing. At first blush, advocating for partial withdrawal seems contrary to the interests of ASCAP and BMI. If the publishers take back large swaths of the performance licensing business from the PROs, that will assuredly weaken the PROs as organizations by reducing their revenues and negotiating leverage. Nevertheless, the PROs decided that losing the major publishers' streaming business was better than losing them as members altogether.
Interestingly, when the U.K. rights society PRS weighed in on the 2014 DOJ consent decree review, it advocated for the modifications ASCAP and BMI sought. That position seems counterintuitive since the sought modifications would assuredly have resulted in mass direct-licensing of streaming platforms by U.S. publishers with whom PRS has no reciprocal relationship. This would potentially have thrust the U.S. performance rights licensing environment-and the ability to collect therefrom-into chaos from the foreign CMO's perspective. Nevertheless, PRS seemed to take the long view: If direct licensing by the majors ( In 2016, after reviewing comments from more than 200 stakeholders ranging from broadcast and restaurant industry advocacy groups to music publishers and composers, the DOJ decided not to modify the decrees. 116 The DOJ's biggest concerns were stability and predictability in the music performance rights market. It was reluctant to disturb the established licensing infrastructure and expectations that had developed over the nearly eight decades since the decrees were imposed. 117 But rather than simply declining to modify the consent decrees to be less restrictive, the DOJ unexpectedly and emphatically recommended a more restrictive reading of the consent decrees by interpreting them to require ASCAP and BMI to grant users "full-work" blanket licenses. 118 Full-work licensing contrasts fractional licensing: ASCAP and BMI maintain they issue fractional licenses, meaning that they only license the portions of works owned by their respective members. Thus, for example, if a song has two writers, one represented by ASCAP and one represented by BMI, a licensee may not lawfully perform the song if it acquires only one blanket license from either ASCAP or BMI. Under a fractional licensing approach, the licensee must have a blanket license from both PROs. The effect of fractional licensing is that, even if a work is in the catalog of ASCAP or BMI, the licensee is not guaranteed immunity from liability unless all of the work's writers are members of the licensing PRO. In practice, therefore, a music user that seeks maximum immunity must obtain a license from every PRO. "Full-work" licensing, by contrast, would enable a PRO to fully license a work even if the PRO represents only a fractional interest. Full-work licensing reduces the cost and burden on music users but also reduces songwriter income and control over uses. In a declaratory judgment action on the question of full-work licensing, the judge overseeing BMI's decree agreed with BMI that the decree permits fractional licensing. 119 The Second Circuit affirmed, reasoning that, because the consent decree "is silent on fractional licensing, BMI may (and perhaps must) offer [fractional licenses] unless a clear and unambiguous command of the decree would thereby be violated." 120 Less than three years after the DOJ closed its consent decree review, it reopened it under Trump administration leadership that was especially skeptical of antitrust consent decrees. 121 The review was part of a massive investigation targeting more BMl's repertoire becoming less valuable: this could lead to a reduction of rates achieved in the market on PRS's behalf by ASCAP and BMI and the application of higher commission rates in order to meet their costs. Clearly, this would be to the detriment of members who may continue to be represented by the US CMOs and by CMOs internationally, particularly the members of PRS.").
116 117. Id. at 3 ("Although stakeholders on all sides have raised some concerns with the status quo, the Division's investigation confirmed that the current system has well served music creators and music users for decades and should remain intact.").
118 than 1,300 legacy antitrust judgments, most of which had no termination date or sunset provision. The DOJ aimed to "pursue the termination of outdated judgments," the "vast majority" of which "no longer protect competition because of changes in industry conditions, changes in economics, changes in law, or for other reasons." 122 The CEOs of ASCAP and BMI issued a joint open letter welcoming the review as an opportunity to modernize the music rights marketplace. 123 They acknowledged that ending the decrees in one stroke would be highly disruptive to licensees. They proposed, therefore, that the DOJ craft new decrees that would keep the present licensing structure largely in place but only for a sunset period designed to facilitate "a thoughtful transition to a free market." 124 The DOJ, doubtless unaware of the hornets' nest it had kicked by reopening the investigation, optimistically predicted in 2019 that it would wrap up its review of the consent decrees before the year's end. 125 This time, the DOJ received over 800 submissions from stakeholders. 126 In January 2021, the DOJ announced yet again that it would not alter the decrees. 127 Again, the DOJ recognized that the music market had changed materially since the decrees were imposed in the 1940s but also recognized that a complex licensing ecosystem had developed around the decrees with reliance interests too deeply entrenched to alter them without great cost. 128 Given that the DOJ declined to amend the consent decrees under two presidential administrations in a row, the likelihood that the decrees will be significantly altered or dissolved in the foreseeable future is low.

D. FURTHER DOWNSIDES TO PRO LICENSING: INEQUITIES, INACCURACIES, AND INEFFICIENCIES IN USAGE SAMPLING AND DISTRIBUTION
Blanket licensing fees are up-front, lump payments that permit unlimited use of a portfolio of works. Collected royalties are pooled and distributed to copyright owners based on complex weighted formulas that factor in frequency, type, and duration of uses, audience size, and many other factors. 129 The payment distribution formulas of ASCAP and BMI have been criticized as byzantine and opaque, and the methods used to capture the usage data fed into the formulas critiqued as antiquated.  131 That may be so, but for independent artists, the system of performance royalty collection and distribution can still be a bewildering black box. 132 One area in which the reporting and distribution is potentially rife with inequities and inefficiencies is live concert performances. Live concerts fall into two categories for PROs: (1) songs performed in the 300 largest concert tours and festivals in the United States, as well as songs performed at approximately a dozen selected major venues such as Madison Square Garden, Radio City Music Hall, and the Hollywood Bowl; and (2) songs performed live at all other concerts and venues. 133 The PROs track all songs performed on the 300 largest concert tours, and the associated licensing revenue is generally divided according to whether the song was performed by a headlining or supporting act (typically ninety percent of revenues from a show are allotted to the former and ten percent to the latter). 134 As regards songs performed by musicians and DJs at smaller venues, the PROs do not track these performances at all. All live musical performances that fall outside the 300 largest concert tours and a few select venues in the United States are simply unreported (except for self-reporting programs that allow musicians to receive royalties for their own live performances of original music). 135 The thousands of cover songs performed live daily at nightclubs, bars, coffee shops, restaurants, and similar venues across the country are not counted for revenue distribution purposes, even though revenues collected from such venues presumably amount to tens of millions of dollars annually. Because no data is collected for performances at these venues, the venues' fees are distributed to publishers and songwriters according to performance data from the year's top 300 concert tours and a few major venues, or by using radio and television performances as a proxy for live performance market share. 136 It is reasonable to assume that the songs performed in bars and coffee shops diverge considerably from the set lists of songs on current major concert tours and map imperfectly at best onto current radio and television performances. 137  137. This is admittedly an assumption, since we lack data concerning live performances at smaller venues. Of course, there will be some overlap, since many legacy artists with popular (and oft-covered) PROs' approach to distributing concert performance revenue could result in a windfall for writers of current hits at the expense of writers and publishers of many legacy and jazz songs covered in the thousands of smaller performances outside the major concert tours.
It can be jarring for independent musicians to discover that venues may deduct a portion of the venue's annual PRO fee from the door receipts, and that that fee will be distributed to the world's biggest music stars. Cellist and composer Zoë Keating blogged about her frustration that her concert receipts were being diverted in this way: [A]t one concert I played last month, the gross ticket sales for the night were $9336. Of the many expenses deducted, one of the items was $86 to ASCAP. What is this? This is the nightly portion of a license fee that the hall pays to ASCAP for the permission to perform music by ASCAP artists in their venue. . . . I have composer and publisher accounts with ASCAP, so I should get this money eventually, right? Wrong. . . . Every day, thousands of venues are required to pay a percentage of their gross ticket sales to ASCAP who then gives that money to . . . let's look here on Pollstar and find the highest-grossing concerts for 2011 . . . U2, Taylor Swift, Kenny Chesney, Lady Gaga, Bon Jovi, etc. 138 The amount that PROs collect from smaller live music venues is significant. ASCAP and BMI are not transparent about small venue licensing revenue, which they include within the category of "general licensing" along with revenue from other businesses such as retail stores, fitness centers, and so on. General licensing revenue for ASCAP and BMI combined is at least $300 million annually. 139 If one assumes that twenty percent of general licensing revenue derives from restaurants, bars, taverns, music halls, coffee shops, theaters, and similar venues, that amounts to $60 million annually being distributed irrespective of which songs are actually performed in those venues. 140 PRO members have tolerated this system presumably because tracking live performances in smaller venues is notoriously difficult. The PROs, for their part, probably have little incentive to change the rules if their prize members-major publishers and star writers-benefit most from the system. However, technological solutions on the horizon could make live performance tracking and reporting more hits are still touring and cover bands and DJs do play songs by stars currently on tour. But any overlap is coincidental and does not reflect the proportional usage of songs played in most venues.
138. Keating, supra note 132. 140. Because the PROs publish almost no data about small venue licensing, a back-of-the-envelope estimate will have to suffice. The estimate here assumes that small live performance venues account for six percent-or 48,000-of the 800,000 businesses ASCAP reports constitute its general licensing customer base. See ASCAP, 2018 ANNUAL REPORT 5 (2019). It further assumes (probably conservatively) that those live venues pay on average $1,250 in total annual PRO licensing fees to ASCAP and BMI (48,000*1,250=60,000,000). For simplicity's sake, this estimate does not take into account any small venue licensing revenue generated by SESAC or GMR. COLUMBIA JOURNAL OF LAW & THE ARTS  [45:1 feasible, opening the door for more equitable revenue distribution and potentially disruptive live performance licensing business models. 141 The monetary value of small-venue live performance licensing may be comparatively small but this function remains foundational to the PROs' identities. Many members of the public-and many musicians-know the PROs primarily as the entities that license bars and restaurants. Licensing venues was ASCAP's main function before the advent of radio, and general licensing remains an area where the efficiencies of collective licensing are still readily apparent even in the digital age. While major publishers believe they have the capacity to direct-license the relatively few digital streaming services, no publisher has shown an interest in taking on the burden of directly licensing small venues across the country and engaging in endless enforcement. Live licensing is perhaps the area in which the PROs are best able to demonstrate their value to their members going forward. It is therefore troubling that it is also the area in which their sampling and distribution is most deficient.

IV. THE FUTURE OF COLLECTIVE MUSIC COPYRIGHT MANAGEMENT IN THE UNITED STATES
Technological development has collective copyright management in the U.S. music industry in a state of flux. Compulsory licenses for streamed sound recordings and musical compositions have brought some clarity and stability to the streaming ecosystem. New collective management organizations have been created in recent years to meet the challenge of administering a new breed of licenses and royalties from digital streaming. 142 If the MLC database lives up to its potential, it would be a boon to the music industry compared to the data disarray that it replaces. But serious questions remain. How will the MWMA's willing buyer, willing seller standard-intended to avoid the rate stagnation that has plagued earlier compulsory licenses-work in practice? Will the MLC database provide an effective solution to the problem of matching sound recordings to underlying musical compositions to ensure proper distribution of mechanical royalties?
The PROs' future is far more uncertain. Whether-or in what form-ASCAP and BMI survive has been the elephant in the music licensing room since 2011, when major publishers attempted to withdraw digital rights. 143 The failure of two attempts to modify or abolish the consent decrees in a matter of a few years does not bode well for the long-term viability of ASCAP and BMI in their present form. Although the DOJ's decision not to amend the decrees has hindered publishers' selective rights withdrawal plans, it likely only increased publishers' desire to free themselves entirely from the yoke of ASCAP and BMI.

A. COMPULSORY LICENSES: SOUNDEXCHANGE AND THE MLC
SoundExchange, an inveterate CMO in the streaming music space, has been operating since the early 2000s to administer compulsory licenses for the digital performance right in sound recordings. That said, as a CMO that primarily collects from noninteractive streaming platforms, SoundExchange is presiding over a gradually shrinking domain. 144 SoundExchange's royalty distributions as of 2021 account for just fourteen percent of U.S. recorded music revenue, down from sixteen percent in 2015. 145 As interactive streaming subscriptions rise, the percentage of users listening to noninteractive online radio decreases. 146 SoundExchange's services will remain relevant for the foreseeable future as the compulsory license for noninteractive streaming seems here to stay, but SoundExchange is not operating in a growth market.
Nevertheless, as the earlier of the digital streaming compulsory license administrators, SoundExchange serves as the canary in the coal mine for the MLC. SoundExchange's first two decades have been marred by prolonged and contentious rate-setting disputes, persistent data problems, and allegations of underpayment. 147 Still, this process has certainly helped pave the way for the MLC. SoundExchange has operated under a willing buyer, willing seller rate-setting model since 2002, 148 and has therefore spent decades hashing out with the CRB and streaming services how the standard works in practice. Commentators and music services have long pointed out that the standard itself is paradoxical, since there has never been a free market for digital sound recording performance rights or mechanical rights. 149 In practice, the standard might not lead to materially different results than any other similar standards, 150 with CRB rate hearings predictably boiling down to, as one observer puts it, "SoundExchange asking for a substantial increase in the royalties and the webcasting community . . . asking for decreases." 151 Nevertheless, the standard now applies to compulsory blanket mechanical license rate-setting, and the CRB has had time to contemplate thorny and time-consuming questions such as who is a hypothetical "willing seller" for purposes of the standard and what kinds of negotiations are admissible evidence of market rates. The rate-setting process for the blanket mechanical compulsory license is likely to hew closely to the approach established in the SoundExchange context, albeit with some new considerations to be worked out, such as how evidence of direct licensing rates for sound recordings may be considered for purposes of market rate approximation. 152 Publishers and songwriters believe the ability to introduce rates negotiated by streaming services for sound recordings is a major advantage of the MWMA's willing buyer, willing seller standard for the blanket mechanical compulsory license. At the time of this writing, about fifteen percent of total streaming revenues goes to songwriters and publishers, while record labels and artists receive more than three times that amount. 153 Songwriters hope that because CRB judges are now empowered to consider all market evidence, including sound recording royalties, they will be persuaded to reduce the substantial disparity between publisher/songwriter royalties and record label royalties. 154 SoundExchange's experience with data deficiency and royalty distribution problems are also instructive for the MLC. Indeed, SoundExchange, invoking its own "extensive experience developing and providing public access to music repertoire information," has publicly criticized the MLC's use of "non-authoritative" sources of sound recording identifiers in the MLC database. 155 SoundExchange's comments about the MLC might evince a percolating turf war between the CMOs,  . 16, 2021), https://perma.cc/V3G5-H2XP (reporting that labels receive more than three times the streaming revenue that publishers and songwriters receive).
154. See Gorgoni, supra note 35, at 11. Streaming services believe that how the royalties they pay are divided between copyright owners should be resolved between copyright owners themselves and should not affect the overall amount that streaming services pay. See infra notes 205-206 and accompanying text.
155. Comments of SoundExchange, Inc. In the Matter of the Public Musical Works Database and Transparency of the Mechanical Licensing Collective 1, 5-6 (Oct. 19, 2020).

2021] THE FUTURE OF MUSIC COPYRIGHT COLLECTIVES 35
but they nonetheless echo critiques that others have made concerning the development, functionality, and implementation of the MLC database. Critics have raised concerns about the accuracy and completeness of the MLC database, the transparency of the MLC organization, whether the MLC database is an outdated solution, and whether creating a new CMO risks introducing the same inefficiencies and rent-seeking that foreseeably arise with the introduction of any powerful intermediary. 156 While the MLC's centralized database will never solve all the music industry's information problems, it will help enormously compared with the scattered and patchy data conditions that preceded it. And because ownership information is dynamic as owners of works can change over time through assignment and devise, metadata embedded in files will become likewise outdated unless it is updatable and updated in real time. This makes the existence of a central, authoritative database valuable, especially if a solution is developed for pushing database updates in real time to service providers.
Finally, there is the question of whether the MWMA's structure locks in the winners of yesterday and today to the detriment of future market entrants. Tyler Ochoa observes that the provisions of the MWMA "read like negotiated agreements between existing stakeholders because that is exactly what they are: Congress invites existing stakeholders to negotiate what they want, and it enacts whatever contractual provisions the stakeholders agree upon, so long as they keep the campaign contributions flowing." 157 The benefits of the MWMA "come with the costs of further entrenching existing stakeholders and business models in statutory language, erecting barriers to entry for new entities in the form of an impenetrable tangle of statutory language that deters all but repeat players from engaging in the system." 158 This mode of legislating imposes costs, to be sure, especially given that future stakeholders by definition have no place at the table when the rules are set. 159 However, the fact that the MLC and the MWMA's blanket compulsory mechanical licensing framework arose from agreements between current stakeholders does not ensure that new entrants will be frozen out of the market. First, 156. See id.; Lowery, supra note 38 (alleging that the fact that the database is starting from a poor baseline means that its improvement will only result from a convoluted process of manual data entry in which "songwriters will do all the heavy lifting" if they hope to be paid); David Lowery, MLC Selects as "Digital Services Provider" the Company that Sent Fraudulent License Notices To Songwriters, TRICHORDIST BLOG (Apr. 11, 2021), https://perma.cc/X6LM-EY6M (expressing outrage that once formed, the MLC immediately turned around and outsourced its royalty payment processing to HFA, which was much criticized for its pre-MMA mechanical royalty distribution practices); Chris Castle, Who Owns the MLC Database of Songs?, MUSIC TECH. SOLS. (June 19, 2020), https://perma.cc/M92T-8Z3X (arguing that the MLC database is an "11th Century solution to a 21st Century problem-a list of things that will be very difficult to get right and even more difficult to keep right" since "[s]tatic lists of dynamic things necessarily are out of date the moment they are fixed" regardless of the rules' provenance, replacing informal custom with formal rules increases certainty-and uncertainty is one of the largest barriers to entry for new market participants, especially in a market reliant on copyright licensing. To be sure, the MWMA is not a model of clarity, nor do rules always beget certainty, but the blanket compulsory license alone makes it less risky to enter the market as a streaming startup today than it was before the MWMA.
Second, the history of such a legislative approach suggests that it does not necessarily inhibit new market entrants or the development of competing models. 160 The MWMA, as Professor Ochoa notes, is hardly the first copyright statute to codify negotiated settlements between copyright owners and users. 161 For more than a century, private negotiations have routinely been used as an expedient route to copyright revision in the face of new technologies that upend existing markets and create uncertainty around the scope and contours of existing rights. 162 Like many other new copying and delivery technologies that have emerged throughout copyright's history, streaming arose into a legal framework and set of industry practices that poorly fit the emerging model. The stakeholders had to construct a workable solution on the fly using legacy concepts. True, the legislation-bysettlement framework not only enables and facilitates the new model, but also entrenches it. As Professor Litman observes, however, "[i]ndustries . . . adjust in time to even the most inhospitable law. Where the copyright statute failed to accommodate the realities faced by affected industries, the industries devised expedients, exploited loopholes, and negotiated agreements that superseded statutory provisions." 163 Although Professor Litman was referencing developments in the first half of the twentieth century, the narrative fits the development of streaming and will likely fit future paradigms, as well. New entrants will not necessarily be barred because when truly new paradigms-"technological discontinuities" 164 -arise, industry participants will likely forge new practices in the shadow of but outside the existing formal legal framework and the law will be adjusted and resettled again once industry practices crystalize. 165

B. PERFORMANCE RIGHTS ORGANIZATIONS
The story of technology driving disintermediation is even more pronounced in the PRO space. As noted in Part III, the PROs' primary function is to reduce transaction 160 costs. In the twentieth-century world of paper licensing agreements and manual, analog performance tracking, it made sense to have intermediaries act as clearinghouses between music copyright owners and users. Today, however, most performance revenue derives from digital streaming, in which licensing and payment transactions occur digitally and automatically.
The notion that blanket collective licenses are necessary for the efficient operation of digital music services is demonstrably false. Interactive streaming services such as Spotify and Apple Music have always directly licensed all the copyrighted recordings on their services, which has created no serious logistical challenges. Spotify hosts over 70 million recordings at the time of this writing, and that number is rapidly growing. 166 Spotify licenses recording rights through a variety of intermediaries: major record labels (which function as licensing collectives of sorts for their artists); associations such as the Music and Entertainment Rights Licensing Independent Network (Merlin), which represents independent record labels; and distributors such as CD Baby, TuneCore, DistroKid, and The Orchard, which represent artists and independent labels, licensing their rights to streaming and download services globally and handling royalty collection and distribution. 167 There is no logistical or technological impediment to publishers direct-licensing performance rights to streaming services. It seems that the selective withdrawal prohibition is the primary reason the major publishers continue to sublicense streaming through PROs; if publishers want to collect general licensing and terrestrial broadcast licensing revenues through ASCAP and BMI, then the publishers must allow those PROs to license digital service providers, too. 168 But technology has rapidly reduced transaction costs across the range of businesses that use music rights, enabling new entrants into the performance licensing and royalty collection services space. 169 For example, AMRA, a global music rights collecting society owned by independent music publishing giant Kobalt, has established a successful alternative to traditional performance rights societies by offering global performance royalty collection from streaming services. 170 AMRA touts the high transparency and efficiency of its rights management platform, in 166 contrast to many PROs around the world. 171 Many other companies today in the music rights clearance and administration business have advanced licensing, tracking, reporting, collection, and payment systems that could aid publishers large and small in direct-licensing performance rights. 172 Other startups are bypassing the PROs to license the music of independent artists directly to businesses for use as background music, potentially poaching a significant source of PRO general licensing revenue. 173 These trends are transforming the licensing market and creating legitimate, directlicensing alternatives to the consent decree-bound PROs. It certainly would be possible for major publishers to directly license the handful of major streaming services and digital B2B background music providers. 174 The publishers would also have their choice of the rights administrators noted above to manage direct licensing of broadcasters and other users. Or, after withdrawing their rights from the PROs, the major publishers could conceivably hire the PROs-with their extensive royalty collection and payment infrastructure-as collecting societies to administer direct licensing deals if the consent decrees are interpreted to allow this. This would align with Professor Lunney's prediction that PROs might transition from blanket licensors to collecting societies. 175 Smaller publishers and artists, too, could work with the new breed of rights management companies to administer direct licenses across diverse uses, or they could continue to license through the PROs.
Even the sphere of performance licensing seemingly most resistant to disruption-live music performances at smaller venues-may eventually be vulnerable to disaggregation from the PROs. Blanket licensing is especially valuable, and direct licensing especially challenging, in small venues where live performances are not easily tracked. However, researchers are developing technologies to address the challenge of live music recognition and reporting. 176  177. As noted above, although bars and restaurants seek the peace-of-mind liability coverage that blanket licenses offer, they also complain about the inflexibility and over-inclusiveness of blanket licensing that leads to higher costs. A direct-licensing platform with a comprehensive catalog that would Growing tensions between publishers and PROs over operating efficiencies and payment practices may also intensify publishers' desire to take more direct control over licensing. 178 The chairman of Warner Chappell Publishing remarked about PROs in October 2020: Some are good, some aren't so good; some feel like allies and partners, but some are inefficient, holding on to money and charging high commission rates. . . . When you're objectively looking at the PRO landscape, you would rightly ask some questions: Why has it gotten more, not less, complicated in the digital age? Why are some of these guys holding on to so much money? I won't go into individual cases . . . . 179 Although it is unclear the extent to which these comments target U.S. PROs, pressure is mounting on PROs everywhere to root out profligacy and maximize the size and speed of payments. 180 Driving this trend is a recent influx of Wall Street capital into music publishing. 181 As Music Business Worldwide recently reported, "KKR, Providence Equity Partners, Blackstone, and Morgan Stanley have jointly committed billions of dollars to buying music rights. Each has a reputation for fierce financial discipline." 182 Warner Music Group and Universal Music Group are both public companies. 183 If direct licensing can increase revenue by even a few percentage points over collective licensing, publishers could feel compelled to make direct licensing a reality. 184 None of this is to suggest that it would be easy or painless for publishers to extricate themselves from the PRO ecosystem. But it now is, or soon will be, feasible to replace major functions of the PROs with specialized alternatives. ASCAP and BMI recognize this, of course, which is why over the past decade they have been license on a per-performance basis through automated usage tracking-allowing the venue to pay for only the compositions performed and ensuring royalties are remitted to owners of the songs that are actually played-would address many of the problems that plague live music licensing today. Of course, publishers and their direct licensing administrators would have to figure out how to maintain a credible enforcement threat in the absence of the PROs. In theory, the PROs themselves might be well positioned to pivot to a direct licensing model for live performances as tracking technologies mature. But the consent decrees make licensing individual works cumbersome, requiring express member consent for each license. See ASCAP Consent Decree, supra note 79, at 8-9 ( § VI); BMI Consent Decree, supra note 80, at 5-6 ( § IX.C).
178. See Ingham, Collection Societies, supra note 70 ("[T]he cumulative operating expenses of PROs in 2019 stood at comfortably north of half a billion dollars. . . . Understandably, music rightsholders have long raised prickly questions about PRO expenditure and efficiency.").
180. See Ingham, Collection Societies, supra note 70 (explaining the growing pressure on PROs to ensure "not a penny is spent out of place" s copyright law has increasingly become a liability regime, private agreements in the space have gone in the other direction, opting out of collectives and adopting stronger protections than those afforded by law."). desperate to dismantle the consent decrees. 185 They worry that without the ability to negotiate prices, they will forever be locked into artificially and detrimentally low rates and their members will defect to seek higher rates. Collective enforcement is a PRO function that would be especially costly to replace in a direct-licensing model. However, collective enforcement is less consequential in the digital era. Publishers are increasingly less reliant on royalties from smaller licensees, against whom enforcement is most costly.
Another big question is whether, if major publishers defect from ASCAP and BMI, broadcasters and streaming services will respond by lobbying for a blanket compulsory performance rights license akin to the blanket compulsory mechanical rights license established by the MWMA. 186 Broadcasters and streaming services are largely content with the performance rights status quo. 187 The present regime of collective blanket licensing highly regulated by consent decrees already accomplishes essentially what a compulsory license would do: It regulates pricing by vesting courts with ultimate rate-setting power and significantly reduces the licensing discretion of ASCAP and BMI. 188 If music users are forced to negotiate with publishers for rates higher than those available from the PROs, the push for a compulsory license seems inevitable.
Some allege that a shift to direct licensing would not lead to meaningful competition in music licensing because concentration in the publishing market has created an oligopoly that controls and manipulates the market. 189 Such anticompetition concerns might make a compulsory licensing solution seem more attractive. The interplay between market power and music licensing is complex, however. Major publishers have significant market share, but many small publishers enjoy market power that outstrips their "independent" status, helping to create competition. 190 That said, because no streaming service could survive without Sony's or Universal's catalog, a shift to direct licensing of performance rights might sound like a recipe for undue bargaining power. Yet, interactive streaming services face precisely the same situation with major record companies, from which streaming services must acquire direct licenses for sound recordings. Although negotiations between labels and streaming services lead to predictable wrangling, both sides are always motivated to come to terms. 191 If (or when) the push for a performance rights compulsory license comes, therefore, Congress should be skeptical of arguments that the market power of major publishers renders the development of a functional market for performance rights impossible. The functional market for digital sound recording performance rights in the interactive streaming context provides compelling evidence that copyright owners and music users can negotiate an agreement. It is not in copyright owners' interest to exact such high prices that music users' businesses fail. Both sides are mutually dependent: Publishers need their songs disseminated as widely as possible and music users need songs. That shared interest provides powerful motivation to come to mutually sustainable terms.
Alternatively, the § 115 compulsory license for mechanical rights could ultimately swallow performance rights licensing for musical works, at least in the interactive streaming context. The difference between the two is already more a matter of form than function, since the CRB set an "all-in" rate ceiling that includes and therefore caps both performance and mechanical royalties. 192 Because performance royalties count toward the all-in rate, it is unclear whether directly negotiating higher performance royalties would benefit publishers; an increase in performance royalties would presumably decrease the royalties attributed to mechanicals but might not increase the overall pie. Under this structure, an increased performance royalty rate achieved via direct licensing might disadvantage publishers because mechanicals are cheaper to collect. Whereas copyright owners pay an administrative fee to PROs (or to a third-party administrator under a future directlicensing regime), the MLC is funded by streaming services and therefore charges copyright owners no administrative fees for its services. This might incentivize publishers to phase out interactive streaming performance royalties for musical works. 193 However, doing so would not be costless, as it would drastically weaken and perhaps destroy the PROs, forcing the publishers to develop the licensing and collection alternatives for non-streaming performance royalties discussed above. It See  would also lock all interactive streaming revenue for compositions into a compulsory licensing framework for the foreseeable future. A compulsory performance rights license is unlikely to be a positive development for songwriters and publishers because it would lock them into rates substantially lower than those paid to artists and labels. However, a compulsory license might not be the worst outcome for ASCAP and BMI. Just as the MLC immediately awarded HFA a contract to process MLC transactions, one imagines the PROs' performance rights administration expertise would put them at the head of the line of contractors under consideration to administer a new compulsory license.

CONCLUSION
Several years ago, Glynn Lunney predicted that music performance rights licensing, currently dominated by PRO collective licensing, could transition into a collective copyright management model because the "market can solve the transaction cost problem that particular instances of widespread copyright infringement create without the need for a uniform price, collectively set." 194 This led Professor Lunney to ask, "How long will the copyright collective [licensing] model be permitted to endure?" 195 As the foregoing discussion shows, it is still a fair question. But if there is pressure on the PRO licensing model, it is coming from copyright owners, not music users. Major music licensees such as streaming services clearly prefer highly regulated blanket collective licensing to direct licensing. 196 So, if the music publishing industry forgoes the PRO-centric blanket licensing model for performance rights licensing and embraces direct licensing, it will be over the clamorous protests of the streaming industry. 197 Nevertheless, because technological advances in performance tracking and usage data make direct licensing of performance rights more feasible than ever before, music publishers could transition to a direct-licensing model for performance rights. The DOJ reviews of the ASCAP and BMI consent decrees in 2014-16 and 2019-21, both of which failed to effect any changes to the decrees or eliminate the "all-in" licensing requirement, might have made this an inevitability. 198 Accordingly, many open questions remain about the future of ASCAP and BMI (and perhaps SESAC and GMR, as well). If major publishers move to direct-license streaming services and the "all-in" rule remains intact, how would they license and 194. Lunney, supra note 74, at 322. 195. Id. 196. For example, in response to the DOJ's review of ASCAP and BMI's consent decrees, both NPR and Pandora Music urged the department to support compulsory music licensing in order to protect streaming services and warned against modifying the consent decrees. Pandora Media, Inc., supra note 189, at 6 (arguing PROs would extract supracompetititve rates from licensees if consent decrees were repealed); National Public Radio, Inc., supra note 188, at 3-4 ("Compulsory licensing . . . performs an essential function for public broadcasting by facilitating the licensing and use of musical works in a relatively cost effective manner.").
collect for non-streaming performances? As noted in Part IV, technology and competition in the licensing space makes non-PRO alternatives more feasible all the time. However, switching from the PROs, with their licensing expertise, networks, and enforcement capabilities, would be a major and risky transition. And there is no guarantee that making the switch to direct licensing would yield the material difference in streaming licensing fees the publishers hope for.
There is also the question of whether independent publishers would follow suit by removing their rights from ASCAP and BMI. Doing so would gut the PROs as collective licensing organizations. But smaller publishers would have fewer resources available to manage the transition to PRO-less, direct performance rights licensing. Small publishers benefit greatly from the PROs, not only because of reduced transaction costs but also because of valuable administrative and enforcement cost savings that arise from the efficiencies of the collective licensing model. The likely outcome if the market moves to direct licensing of streaming services is that ASCAP and BMI, if permitted by the DOJ, would reinvent themselves as CMOs that administer direct-licensing transactions without issuing blanket licenses and continue to collect on behalf of publishers. Indeed, when Sony briefly purported to withdraw its digital rights from ASCAP and BMI in 2011, it continued to use them to administer its direct licenses. This would allow publishers to continue to enjoy many of the efficiencies of collective copyright management while employing a direct licensing model. 199 Further, there is the question of whether a direct-licensing model would increase songwriters' vulnerability. 200 Today, the PROs pay songwriters their share of performance royalty income directly, providing songwriters with an accountable intermediary that lacks the conflicts of interest inherent in the songwriter-publisher relationship. The PROs implement and effectively enforce the time-honored fiftyfifty royalty split between songwriters and publishers. 201 If publishers direct-license streaming services, they might aim to collect on behalf of songwriters. How this would be addressed in a direct-licensing model is unclear. 202 It is seemingly in the best interests of the songwriter to be paid directly by whatever intermediaries administer direct licenses on behalf of publishers, whether they be PROs reconfigured as CMOs or some other third-party administrators.
Of course, publishers could also opt to maintain the status quo, relying on PROs for performance rights licensing and administration, and on periodic CRB-mandated rate increases to increase revenue. While this is perhaps the most frictionless path forward, it would mean capitulation on a definitive issue: the ability to set their own prices and maximize revenue. Publishers have long bridled at the excessive regulation of their industry. Between compulsory mechanical licensing and the ASCAP and BMI consent decrees, the only major revenue stream over which music publishers enjoy direct pricing control is synchronization. Publishers believe they need not look further than streaming rates for sound recordings to demonstrate their disadvantaged position and the resultant inequities. 203 Record labels and artists earn more than three times what publishers and songwriters earn for the same interactive stream. 204 Of course, sound recording licenses for interactive streaming are neither compulsory nor subject to regulatory oversight, suggesting to publishers that their own copyrights would be far more valuable in a free market.
Streaming services view the situation differently: From their perspective, they already pay most of their revenue to copyright owners. For example, Spotify pays roughly seventy percent of total revenues to rights holders. 205 How copyright owners choose to divide that lucrative pot among themselves, streaming services believe, is not the streaming services' concern. Thus, streaming services maintain that if there is a problem with the status quo, it is that the labels are paid too much and that any increase in publishers' share should result from a concomitant decrease in labels' royalties, not from extracting an additional pound of flesh from streaming services. 206 And streaming services are quick to point out that the major publishers are owned by the same corporations as the major labels. But executives are judged on their ability grow revenue, so record label bosses are unlikely to support any solution that reduces their revenues. The need for intra-industry accord on issues of fair payment seems inevitable.
If publishers do retract their rights from ASCAP and BMI and pursue directlicensing deals, music users-especially streaming services, broadcasters, and venues-may counter by campaigning for a compulsory blanket license for musical performance rights with rates set by the CRB, similar to the blanket compulsory license for mechanical streaming rights established by the MWMA. To achieve this, a separate compulsory license for performance rights might be established, or the MWMA blanket license might be expanded to cover both mechanical and performance rights. In the streaming context, current practice regarding the mechanical compulsory license already effectively merges the two because the royalties owed for mechanicals are determined by subtracting performance royalties paid from an "all-in" rate including mechanicals and performance rights. In the event that music users lobby for a compulsory performance rights license, Congress should be skeptical of arguments that a compulsory license is necessary because major publishers' market power renders the market anticompetitive and dysfunctional. Direct licensing deals between record labels and interactive streaming services are instructive, as the record industry is more concentrated than the music publishing industry but the symbiotic relationship between labels and streaming services has consistently led to mutual agreement for more than a decade.
It is important to remember how nascent the streaming market still is. To put it into perspective, streaming revenues only overtook physical media and digital download sales for the first time in 2018. 207 It is impossible to predict how licensing models will develop in the coming years and decades, partly because it will still take time for the industry to shake off the vestiges of licensing practices developed for twentieth-century music delivery models. The optimal streaming licensing models and rates are still being worked out. The industry is grappling with transformative questions about the equities of paying songwriters a fraction of what artists receive from streaming, and of artists as well as songwriters and publishers receiving only a fraction of what record labels receive. 208 The next decade could witness a mass exodus from labels and publishers as artists and songwriters license directly through digital distributors to disintermediate and keep the majority of the revenue their music generates. 209 That could cause a massive restructuring of the music economy, undermining the legacy intermediaries and resulting in a technology-and information-driven fragmented licensing environment in which a plethora of digital media distributors and rights administrators license discrete industry sectors-streaming, radio, background music, bars and restaurants, and so on. Of course, people have been predicting disintermediation since the advent of the internet, but it has yet to happen on the scale many predicted, in part because it turns out that it is very difficult for artists and songwriters to handle all the functions that labels, publishers, and PROs do. 210 Nevertheless, with the multidirectional pressures on ASCAP and BMI from new competitors, their major publisher members, and regulators, it is easy to imagine that in the coming years they will be reduced to CMOs providing administrative support for direct or compulsory licenses-a far cry from the blanket-licensing behemoths that drew antitrust scrutiny for nearly a century.