In re Pandora Media, Inc. v. American Society of Composers, Authors, & Publishers

6 F. Supp. 3d 317

• Court: District Court, S.D. New York
• Decided: March 18, 2014
• Docket: Nos. 12 Civ. 8035(DLC), 41 Civ. 1395(DLC)
• Status: Published

Opinion

OPINION & ORDER

DENISE COTE, District Judge.

TABLE OF CONTENTS

INTRODUCTION. co CO o

FINDINGS OF FACT. CO CO CO

I. The American Society of Composers, Authors and Publishers . Co co CO

A. ASCAP Background. CO CO CO

B. The ASCAP Consent Decree. CO CO CO

II.The Evolution of the Radio Industry. CO CO CO

III. The RMLC-ASCAP License Agreement for the Period 2010-2016 ... lO <M CO

IV. Pandora. CO CO -3

A. Pandora’s Music Genome Project. CO CO —3

B. Pandora Premieres. CO CO OO

C. Pandora’s Comedy Programming. CO CO OO

D. Pandora’s Revenue. CO CO OO

E. Pandora’s Competitive Environment. CO CO OO

V.Pandora’s Licensing History with ASCAP. o CO CO

VI.The April 2011 ASCAP Compendium Modification. A. Overview and Context. T — t T — H CO CO CO CO

B. Public Performance Rights for Compositions versus Sound Recordings. CO CO CO

C. ASCAP-Publisher Negotiations Modification. CO CO CO

D. The Compendium Modification Allowing New Media Withdrawals is Enacted. CO CO CO

E. ASCAP Provides Administrative Services for Withdrawing Publishers. Cr-CO CO

VII.A Second Compendium Modification in December 2012 the “Standard Services” Agreement. 00 CO CO

VIII.Pandora Negotiates Direct Licenses with EMI, Sony, and UMPG and Fails to Negotiate an Agreement with ASCAP . CO CO

A. The Pandora-EMI License Negotiations. CO CO

B. The Pandora-ASCAP License Negotiations. CO ^

C. The Pandora-Sony License Negotiations. CO

D. The Pandora-UMPG License Negotiations. CO ^

IX.September 17 Partial Summary Judgment Opinion. CO cn o

X.Other Licensing Agreements Put Forth as Benchmarks. CO cn 1 — «•

A. The Pandora-SESAC License .

B. Apple’s iTunes Radio Licenses with Publishers and PROs. CO cn Í — l

CONCLUSIONS OF LAW. CO lO CO

I. ASCAP’s Rate Proposal of 1.85% for 2011 and 2012. CO cn cn

II. ASCAP’s Rate Proposal of 2.50% for 2013 and 3.00% for 2014 and 2015. CO cn 01

A. Presumption of a Single Rate. CO cn 03

B. Pandora’s Direct Licenses with Sony and UMPG. CO cn

1. ASCAP and Publisher Coordination. CO cn

2. The Pandora-Sony License. CO cn

3. The Pandora-UMPG License . CO 0

C. ASCAP s Secondary Benchmarks: the SESAC and Apple LÍC6HS6S

1. The Pandora-SESAC license.

2. The Apple Licenses.

D. ASCAP’s Theoretical Arguments and Motivations.

1. ' ' An Increase in Competition W

2. Demand for Variety.

3. Disparity Between Sound Recording and Composition Fees

4. Cannibalization of Music Sales.

5. Music Intensity.

6. Pandora’s Success.,.

III. Whether Pandora is Entitled to the RMLC 1.70% Rate. 00 Oí to

IV. Publisher Concerns Regarding the Consent Decree and the Rate Court. CO

CONCLUSION 00 >0

INTRODUCTION

Pandora Media Inc. (“Pandora”) has applied for a through-to-the-audience blanket license to perform the musical compositions in the repertoire of the American Society of Composers, Authors and Publishers (“ASCAP”) for the period of January 1, 2011 through December 31, 2015. The parties having been unable to reach agreement on an appropriate licensing fee, pursuant to Article IX of the consent decree under which ASCAP operates— known as the Second Amended Final Judgment (“AFJ2”), see United States v. ASCAP, Civ. No. 41-Civ-1395, 2001 WL 1589999 (S.D.N.Y. June 11, 2001) — Pandora requested on November 5, 2012 that this Court set a rate for that licensing fee.

The parties disagree as to which are the most appropriate benchmarks for the license rate here. Pandora asserts principally that it is similarly situated to radio stations licensed through a 2012 agreement between the Radio Music License Committee (“RMLC”), which represents commercial radio stations, and ASCAP, and is therefore entitled to the rate in that license. Pandora also points to a direct license agreement between Pandora and EMI Music Publishing Ltd. (“EMI”) that was entered into after EMI purported to withdraw its new media¹ licensing rights from ASCAP in 2011.

ASCAP proposes a variety of benchmarks, including the direct licensing agreement into which Pandora entered with EMI, as well as Pandora’s direct licenses with Sony/ATV Music Publishing LLC (“Sony”) and Universal Music Publishing Group (“UMPG”) in the wake of those publishers’ putative withdrawals of new media licensing rights from ASCAP. ASCAP also puts forward other agreements between music rights holders and music users as secondary benchmarks.

The parties have proposed the following rates, expressed as a percentage of revenue: ASCAP proposes a rate of 1.85% for the years 2011 and 2012, 2.50% for 2013, and 3.00% for the years 2014 and 2015. Pandora proposes a rate of 1.70% for all five years. This Opinion sets the rate for all five years at 1.85%.

The task at hand is to determine the fair market value of a blanket license for the public performance of music. As this Court explained in a prior rate court proceeding:

The challenges of [determining a fair market rate for a blanket music license] include discerning a rate that will give composers an economic incentive to keep enriching our lives with music, that avoids compensating composers for contributions made by others either to the creative work or to the delivery of that work to the public, and that does not create distorting incentives in the marketplace that will improperly affect the choices made by composers, inventors, investors, consumers and other economic players.

In re Application of MobiTV, Inc., 712 F.Supp.2d 206, 209 (S.D.N.Y.2010), aff'd sub nom. ASCAP v. MobiTV, Inc., 681 F.3d 76 (2d Cir.2012).

A bench trial was held from January 21 through February 10, 2014. Without objection from the parties, the trial was conducted in accordance with the Court’s customary practices for non-jury proceedings, which includes taking direct testimony from witnesses under a party’s control through affidavits submitted with the Joint Pretrial Order. The parties also served with the Joint Pretrial Order copies of all exhibits and deposition testimony that they intended to offer as evidence in chief at trial.

Prior to trial, ASCAP presented affidavits constituting the direct testimony from ten witnesses, including four ASCAP employees, two music publishing executives, one composer, and three experts.² The ASCAP employees are CEO John LoFru-mentó; Vice President for New Media and Technology Matthew DeFilippis; Director of Licensing Vincent Candilora; and Senior Vice President Seth Saltzman. AS-CAP’s music publisher witnesses were Peter Brodsky, the Executive Vice President of Business and Legal Affairs at Sony; and Zach Horowitz, the Chairman and CEO of UMPG. ASCAP’s composer-witness was Brett James. ASCAP’s experts were Dr. Kevin Murphy, Dr. Orley Ashen-felter, and Robin Flynn. ASCAP also .provided designated deposition excerpts from six witnesses.³

Pandora provided affidavits constituting the direct testimony of eight witnesses, four of whom are current or former Pandora employees and four of whom are experts. The current or former Pandora employees were founder and Chief Strategy Officer Timothy Westergren; former CEO Joseph Kennedy; Chief Technology Officer Thomas Conrad; and Chief Marketing Officer Simon Fleming-Wood. Pandora’s experts were Dr. Leslie Marx, Dr. Roger Noll, William Rosenblatt, and Fred McIntyre. Pandora also provided the designated deposition testimony of a number of witnesses.⁴

The parties also offered deposition designations of certain witnesses as joint exhibits.⁵ At the trial, the parties waived their right to cross-examine several of the witnesses. The witnesses who testified at trial were Brodsky, LoFrumento, DeFilip-pis, Murphy, Flynn, 'Marx, Rosenblatt, Horowitz, Saltzman, Noll, Conrad, Kennedy, Fleming-Wood, and McIntyre. In addition, Pandora called its outside counsel, Robert Rosenbloum, as a witness.

This Opinion constitutes the Court’s findings of fact and conclusions of law following that trial. The factual findings are principally set forth in the first section of this Opinion, but appear as well in the second section.

FINDINGS OF FACT

I. The American Society of Composers, Authors and Publishers

A. ASCAP Background

ASCAP is an unincorporated membership organization of music copyright holders created and controlled by music writers and publishers.⁶ Its function is to coordinate the licensing of copyrighted musical works, and the distribution of royalties, on behalf of its nearly 500,000 members. ASCAP members grant ASCAP the non-exclusive right to license non-dramatic public performances of their music. AS-CAP licenses these works on behalf of the copyright holders to a broad array of music users, including television networks, radio stations, digital music services, colleges, restaurants, and many other venues in which music is performed.

Employing ASCAP to perform these functions is efficient for both music users and copyright holders. A music user can license an enormous portfolio of copyrighted music through the execution of a single license without having to contact each copyright holder. Copyright holders benefit from ASCAP’s expertise and resources in policing the market, negotiating licenses, and distributing the revenue from a vast array of licenses promptly and reliably among the multiple owners of the public performance copyrights in each work. The ability of ASCAP and other performing rights organizations (“PROs”) to grant licenses covering a large number of compositions creates significant economies of scale in the market for music licensing.

ASCAP offers the option of blanket licenses to users. A blanket license is a license that gives the music user the right to perform all of the works in ASCAP’s repertoire, the fee for which does not vary depending on how much of the music the user actually uses. These blanket licenses

reduce the costs of licensing copyrighted musical compositions. They eliminate costly, multiple negotiations of the various rights and provide an efficient means of monitoring the use of musical compositions. They also allow users of copyrighted music to avoid exposure to liability for copyright infringement.

Buffalo Broadcasting Co. v. ASCAP, 744 F.2d 917, 934 (2d Cir.1984) (Winter, J., concurring).

ASCAP’s board of directors is comprised of an equal number of composers and music publishers. The head of the ASCAP board is typically a songwriter. The present head of ASCAP is composer Paul Williams.

ASCAP competes with two other United States PROs: Broadcast Music, Inc. (“BMI”) and SESAC, LLC. (“SESAC”), each of whom also offers blanket licenses. BMI, which is slightly smaller than AS-CAP, operates under a consent decree that is similar to the one that governs ASCAP’s licenses. SESAC is a PRO that is not currently bound by any consent decree.⁷

B. The ASCAP Consent Decree

Since 1941, ASCAP has operated under a consent decree stemming from a Department of Justice antitrust lawsuit. This consent decree has been modified from time to time. The most recent version of the consent decree was issued in 2001 and is known as “AFJ2.” AFJ2 governs here.⁸

In an attempt to ameliorate the anti-competitive concerns raised by ASCAP’s consolidation of music licenses, AFJ2 restricts how ASCAP may issue licenses in a variety of ways. First, AFJ2 provides a mechanism whereby a court, known as the rate court, will determine a reasonable fee for ASCAP licenses when ASCAP and an applicant for a license cannot reach an agreement. AFJ2 § IX. This Court is presently the ASCAP rate court. Second, AFJ2 requires ASCAP to grant a license to perform all of the musical compositions in ASCAP’s repertoire to any entity that requests such a license. AFJ2 §§ VI, IX(E). And third, AFJ2 prevents ASCAP from discriminating in pricing or with respect to other terms or conditions between “similarly situated” licensees. AFJ2 § IV(C). ASCAP members agree to be bound in the exercise of their copyright rights by the terms of AFJ2. For example, the 1996 Agreement Between Sony and ASCAP provides that “[t]he grant [of rights to ASCAP] ... is modified by and subject to the provisions of [AFJ2].”

In addition to operating under a consent decree, ASCAP is governed by a series of internal rules and contracts. The most important internal rule set for purposes of this litigation is the ASCAP Compendium. The ASCAP Compendium can be modified by the ASCAP Board and reflects many of the important rules that govern ASCAP’s obligations to its copyright holder members and vice versa.

II. The Evolution of the Radio Industry

Much of the focus at trial was on the question of whether Pandora can be properly classified as “radio.” A description of the evolution of the radio industry will provide context in understanding Pandora’s features.and its place within the music business.

Radio is a form of media in which a provider transmits audio programming to a listener, where the programming is not directly selected by the listener but is programmed by the provider. As a result, in the context of a music station, the listener does not choose the songs and does not know what composition will be played next. This radio experience has remained constant through the years, regardless of whether radio programming is transmitted by broadcasting, through a cable, from a satellite, or over the internet.

Radio made its debut approximately a century ago and has been a dominant force in the music industry évér since. The first commercial radio station in the United States was located in Pittsburgh, Pennsylvania and was licensed in 1920. The original technological means for delivering radio programming was by broadcasting an “amplitude modulation,” or “AM” signal. By the 1930s, “frequency modulation” or “FM” signal technology was developed. FM broadcasting offered better audio quality but over a smaller range.

Another important moment in the history of radio occurred with the passage of the Telecommunications Act of 1996. Pub.L. No. 104-104, 110 Stat. 56 (1996). Empowered by that legislation, the FCC eliminated most caps on the number of stations that a single company could own. Following that change in the law, there was a large-scale expansion of group ownership of stations. Many terrestrial radio stations are now owned by large conglomerates, such as Clear Channel Communications, Inc. (“Clear Channel”), which owns over 800 stations.

In the 1990s, the first successful national cable radio network was launched, using cable TV transmission lines. Over time, what came to be known as digital TV radio was transmitted through means of cable, satellite,⁹ and telephone-company lines. Some of the major competitors in this market are Music Choice, SiriusXM Satellite Radio, Muzak, and DMX.¹⁰

Also in the 1990s, the nascent internet provided a new means of radio transmission. The introduction in the early 1990s of MP3 digital audio encoding and compression format permitted music to be compressed in way that facilitated distribution over the internet.¹¹ In 1994, the first simulcast of an AM/FM broadcast occurred over the internet. As of today, over 10,000 AM and FM stations stream online. Internet radio includes not just the simulcasting of signals broadcast by AM and FM stations, but also the creation of internet-only radio stations. Over time, some independent companies built directories of internet radio stations. These directories can contain tens of thousands of radio stations. Thus, the internet has enabled providers to present listeners with a vast library of radio programming, the likes of which has never been available before.

Clear Channel and CBS Radio, two major commercial radio companies, launched their own internet radio services in 2008 and 2010, respectively. Clear Channel’s internet radio service is called iHeartRa-dio, and began as a vehicle to simulcast Clear Channel’s own stations.

The arrival of the internet as a radio delivery platform has also permitted radio providers to introduce a level of instantaneous user interactivity for the first time. With the internet, each listener’s device gets its own data stream, in contrast to the broadcasting of a common signal across a geographic area. As will be explained in greater detail below, this permits internet radio services to offer customized music programming based on user feedback. Thus, while a listener to a customized radio service cannot select and does not know what song will be played next, that listener can often give feedback to the customized station to shape the nature of the music that will be played.

As of today, Pandora is the most successful customized radio service. But it was not the first. Prior to Pandora’s launch in 2005, LAUNCHcast and Lastfm, two customized radio services, began in 1999 and 2002, respectively.¹² Recently, three major competitors have emerged as challengers to Pandora’s dominance. In 2011, Clear Channel launched a customized radio offering within its iHeartRadio service, called “Create Station.” Spotify launched a customized radio feature called Spotify Radio in late 2011. And in September 2013, Apple launched its customized radio service called “iTunes Radio.”¹³

In addition to programmed and customized radio, the presence of digital technology and the internet have allowed for the emergence of a third means of delivering music: “on-demand” streaming services. These services provide users with access to large libraries of songs, from which they can select exactly which song to play at any time. A leading on-demand service is Spotify, which had 24 million active users globally as of March 2013. Launched in 2008 in Europe and in the United States in 2011, Spotify has a library of over 20 million songs.¹⁴ Other popular services with on-demand offerings include Rhapsody and Grooveshark. The most popular on-demand services offer both advertising supported and subscription options, but seek to persuade- consumers to elect the subscription model.

Through its century of existence, radio’s popularity has remained robust. The radio industry is a $15 billion industry. It is understood by those in the music business to account for roughly 80% of the music listening experience in the United States. This percentage has remained roughly constant despite the rapid evolution in technology. Thus, while the 80% figure was once confined to listening to music over AM/FM radios, that figure now includes music delivered over radio stations playing through'-TV cable systems and over the internet. Almost half of radio listening occurs while the listener is in an automobile. The other 20% or so of music listening in the United States is experienced by listeners who seek more control over the music that they hear, whether through the purchase and playing of a record album or. a CD, or the subscription to an on-demand digital music service such as Spotify.

III. The RMLC-ASCAP License Agreement for the Period 2010-2016

Much of the radio industry obtains its license for the public performance of AS-CAP music through the RMLC, which is a trade association that represents the commercial radio industry. Between 2003 and 2009, the RMLC paid ASCAP for public performance licensing rights in the form of a “fixed fee” agreement.¹⁵ As a result of the deep recession that hit the country in 2008, the RMLC’s members’ revenues contracted and the fixed fee license began to constitute an increasingly' high percentage of RMLC member revenue. Consequently, in 2009, the RMLC began to negotiate new licensing terms to apply to commercial radio stations effective January 1, 2010 through December 31, 2016. The RMLC wanted a return to the 1.615% rate at which it had paid ASCAP before the fixed-fee arrangement was adopted. It also wanted a license not only for new media transmissions by RMLC member stations, but also new media transmissions by Clear Channel which aggregates many stations for delivery over the internet and mobile devices.

At one point in those negotiations, the RMLC offered ASCAP a dual rate structure for new media and terrestrial broadcasts, under which the RMLC would pay a proposed rate of [REDACTED] of revenue for terrestrial broadcasting and simulcasting of terrestrial radio over the internet, and a separate rate of [REDACTED] of revenue for other new media uses of the ASCAP repertoire. The revenue for the latter category of uses was miniscule in comparison with the revenue to be covered at the [REDACTED] rate. RMLC licensees still derive almost all of their revenue from traditional broadcasting services.

ASCAP and the RMLC ultimately reached an agreement on terms for a license that established separate rates for radio stations depending on their intensity of music use. The rate for Music Format Stations¹⁶ was set at 1.70% of all revenue, including revenue derived from new media uses.¹⁷ The RMLC and ASCAP memorialized their agreement in a binding letter of December 21, 2011. And on January 27, 2012, this Court approved the agreement and its terms became public.

It is of significance to the issues litigated in this trial, that the 1.70% rate applies both to revenue derived from terrestrial broadcasting and from internet transmissions by RMLC members. In addition, the 1.70% rate applies not only to simulcast radio stations that are streamed over the internet by terrestrial broadcasting RMLC members but also to programmed and customized internet radio stations owned by RMLC members. One RMLC member, Clear Channel, is licensed at this rate under a “Group License” form, which covers new media royalty payments for revenues not associated with an individual station. Thus, iHeartRadio’s customized radio Create Station feature, which competes head on with Pandora, is licensed at the 1.70% rate. Until July of 2012, however, there was no revenue generated by the Create Station service.

In 2011, the year of Create Station’s launch, as well as the year that ASCAP and the RMLC agreed to license terms, Create Station constituted just [REDACTED] of the listener hours on iHeartRadio. For the month of March of 2013, which is the last month for which there was data at trial, it had grown to comprise approximately [REDACTED] of its total listener hours on iHeartRadio for that month. Thus, a rapidly increasing share of the iHeartRadio listeners are choosing the customized Create Station feature when listening to iHeartRadio. But from its inauguration in September of 2011 until June of 2012, Clear Channel ran Create Station as an advertising-free service. It began to contribute revenue for the first time in July of 2012. Despite its-growth in audience, the contribution to digital revenue from the Create Station feature [REDACTED]. For the first three months of 2013, it contributed substantially less than [REDACTED] to Clear Channel’s digital revenue, which itself is only a small component of Clear Channel’s entire revenue stream.

IV. Pandora

Pandora is the most successful internet radio service operating in the United States today. It is estimated to have approximately 200 million registered users worldwide¹⁸ and an approximately 70% share of the internet radio market in the United States. Pandora launched its internet radio service in 2005. Roughly eight years later, it had achieved great popularity, streaming an average of 17.7 billion songs per month in the fiscal year 2013.

A. Pandora’s Music Genome Project

Pandora’s exponential growth and popularity can be directly attributed to its substantial investment in its proprietary Music Genome Project (“MGP”) database and associated algorithms. Pandora uses the MGP database to create customized internet radio stations for each of its customers. A Pandora customer creates a station by “seeding” it with a song, artist, genre, or composer. That seed serves as a starting point to which Pandora then applies the information in its MGP database to match that seed with other songs that Pandora’s algorithms predict that the listener is likely to enjoy. The listener continues to give feedback by giving a thumbs-up or thumbs-down when a composition is played, or by signaling that a song should be skipped.

The MGP contains a wealth of data for every composition in its database. Trained music analysts, many of whom have music related degrees or are musicians, listen to the compositions selected for inclusion in the database and register the composition in reference to as many as 450 characteristics.¹⁹ For pop and rock songs, for example, Pandora analyzes between 150 to 200 musical traits. Rap has about 350 “genes,” and classical works have between 300 to 500 MGP-defined attributes. As Pandora’s Conrad testified, “[b]ecause Pandora utilizes trained musicologists to analyze songs, the MGP is able to differentiate not only between an alto and tenor saxophone, but also between various styles of playing a tenor saxophone.” When a Pandora listener seeds a station or registers a thumbs-up reaction, Pandora records that feedback and draws upon the MGP to locate other compositions that the listener is likely to enjoy. Conversely, when the feedback is a “thumbs down,” the song will not reoccur in the user’s playlist, and songs sharing its attributes will appear less frequently.

Besides listening to as many as 100 of their own customized stations, Pandora users can opt to listen to programmed “genre” stations. The most popular Pandora genre stations include “Today’s Hits,” “Today’s Country,” and “Today’s Hip Hop and Pop Hits.” These genre stations are populated by songs which are hand selected by Pandora curators.

Pandora has a catalog of between approximately 1,000,000 to 2,000,000 songs, somewhat less than half of which are licensed through ASCAP. This number is considerably lower than the catalog size of an on-demand service like Spotify, which must have the ability to play virtually any composition any customer might select. Successful on-demand services have catalogs in the range of 20 million songs.

B. Pandora Premieres

Pandora has a small on-demand music service, but it is not part of this license application. The Pandora on-demand service is called Pandora Premieres. It features at any one point in time a few dozen songs, each available for listening on-demand for a limited period of time. This is music that artists and music publishers provide to Pandora for promotional purposes, typically before the commercial release of an album. This on-demand component of Pandora’s service is not a significant part of Pandora; it constitutes a “barely measurable portion of Pandora listening.” Pandora secures the rights to play the songs on Pandora Premiers by negotiating direct licenses with the copyright holders.

C. Pandora’s Comedy Programming

While Pandora is overwhelmingly a service that plays music, in 2011 it introduced comedy offerings. The comedy content constitutes a very small percentage of Pandora’s played content.

D. Pandora’s Revenue

Pandora derives revenue from two principal sources: advertising and subscription fees. As of today, Pandora derives approximately 80% of its revenue from the sale of display, audio and visual advertising, and the remaining 20% or so from a paid subscription service without advertising called “Pandora One.”

Pandora’s revenue has grown exponentially since its inception. For fiscal year 2009, Pandora reported revenue of approximately $19 million. By fiscal year 2013, Pandora’s revenue had risen to over $400 million. As of today, however, Pandora has yet to demonstrate sustained profitability.

Pandora’s payment of licensing fees for the use of music consumes a very significant portion of its revenue. In 2013, Pandora’s content acquisition costs were close to $260 million, or over 60% of its revenue for that fiscal year. A very substantial portion of these costs are for the fees paid to record companies for licenses for sound recordings, as described in more detail below.

E.Pandora’s Competitive Environment

Pandora’s competitive environment is dictated by the nature of its service. Pandora is a radio service, albeit a customized radio service. Unlike traditional broadcast AM/FM radio, in which one program is played for many listeners, Pandora’s digital radio service provides the opportunity to have a unique program created for the enjoyment of each listener. This distinction between programmed and customized radio has been referred to as the one to many, versus the one to one distinction. But, despite that differentiation, made possible by digital technology, Pandora is radio. The listener does not control what song will next be played and doesn’t know what that next song will be. As with other forms of radio, the listener may be introduced to new music she has not heard before. There is an industry term for distinguishing among types of listening experiences: lean-back versus lean-forward. Like radio, Pandora is a lean-back service, in contrast to the on-demand lean-forward services like Spotify.

Not surprisingly, therefore, Pandora competes aggressively with other radio stations for listeners. It competes directly with internet radio stations, whether they are programmed music streaming services or customized radio stations. But, because the internet radio market is comparatively small and because Pandora already holds a significant share of that market, Pandora expects its increased audience, listening hours, and advertising revenue to come largely at the expense of terrestrial radio. While Pandora has a 71% share of the internet radio market, it has less than an 8% share of the overall radio market.

Pandora is attempting to make itself ubiquitous, so that its listeners have Pandora available to them throughout their day, whether they are at home, at work, in the car, or somewhere else. As Pandora explained in a 2013 SEC 10-Q filing, “[o]ne key element of our strategy is to make the Pandora service available everywhere that there is internet connectivity.” Pandora has consequently expanded its service to smartphones, tablets, and television streaming devices. And because almost half of radio listening takes place in cars, Pandora has negotiated agreements to integrate its service into new cars built by a number of auto companies.

Besides competing with traditional radio for listeners, Pandora also competes with traditional radio for advertising dollars. Most terrestrial radio advertising revenue comes from local advertising. To compete for these advertising dollars, Pandora has hired a large in-house local advertising sales force. And to improve its ability to compete for advertising dollars with terrestrial radio stations, Pandora contracts with third party Triton Digital, a firm that collects radio audience data on both a local and national level, in order for radio advertising buyers to better understand the reach of advertisements run on Pandora.

Despite this intensive effort to build advertising revenue, Pandora is still unable to play as many minutes of advertising per hour as its broadcasting competitors. Therefore, as of today, Pandora plays on average approximately 15 songs per hour as compared to terrestrial radio’s roughly 11 songs per hour.²⁰ While Pandora’s free radio service now runs less audio advertising per hour than do terrestrial radio stations, this gap may lessen as Pandora’s business matures.

In accordance with the above, in its public filings with the SEC Pandora identifies its principal competitors as broadcast radio providers, including terrestrial radio providers such as Clear Channel and CBS, satellite radio providers such as Sirius XM, and online radio providers such as CBS’s Last.fm and Clear Channel’s iHeartRadio. But, while programmed radio and customized radio are Pandora’s primary competition, Pandora also competes with interactive,²¹ on-demand internet music services.²² Its identified competitors in this market include Apple’s iTunes Store, RDIO, Rhapsody, Spotify, and Amazon.

V. Pandora’s Licensing History with ASCAP

On July 11, 2005, Pandora first entered into an agreement with ASCAP for a blanket license to publicly perform the compositions in the ASCAP repertoire. This license was in effect from 2005 to 2010, when Pandora exercised its option to cancel it. The license which ASCAP issued to Pandora during this span of years was a form license.

Pandora was licensed by ASCAP from 2005 to 2010 under the ASCAP Experimental License Agreement for Internet Sites & Services — Release 5.0 (“5.0 License”). ASCAP first adopted the 5.0 License in 2004.²³ As of 2004, internet radio had been in existence for roughly ten years, customized internet radio had been in existence for approximately five years, and on-demand services had been in existence at least three years.²⁴ With its adoption of this form license, ASCAP made two important distinctions. First, it raised the rate for new media licenses, reflecting a judgment that those services made more intensive use of music than broadcast radio. Second, it made a distinction between interactive and non-interactive new media services. It did not, however, make any distinction between programmed and customized internet music services.

The 5.0 License allowed non-interactive users to choose between three rate schedules.²⁵ Schedule A of the 5.0 License, which Pandora chose, required it to pay the higher of 1.85% of revenue or a per-session rate. The 1.85% rate represented an increase in ASCAP’s form license rate from the previous rate. The predecessor to the 5.0 License had an equivalent rate for this schedule of 1.615%.²⁶ ASCAP’s form license for interactive services provided for a substantially higher license rate of 3.0%.²⁷

The interactive/non-interactive distinction in the ASCAP form license agreements is borrowed from 17 U.S.C. § 114’s (“Section 114”) use of the term interactive in the context of the licensing of sound recording rights (Section 114 and sound recording rights are discussed below). Because ASCAP considers its music to be more valuable to the services it classifies as interactive, it has licensed them at a higher rate than non-interactive services.

As noted, under the 5.0 License Pandora was required to pay the greater of either the percentage of revenue corresponding to its applicable rate, or a fee based on a concept known as “sessions.” A “session” is defined in the license as “an individual visit and/or access to [the] Internet Site or Service by a User.” Any visit that exceeded one hour in length began a new session.²⁸ At some point in 2010, Pandora recognized that it had been calculating sessions incorrectly, and that it had substantially underpaid ASCAP. It paid ASCAP over $1 million to account for that error. If the payments Pandora was required to make to ASCAP, when measured by the per session rate, are converted into a flat percentage of Pandora’s annual revenue, the effective rate for the years 2005 through 2010 ranged from a high of 3.63% in 2006 to a low of 1.91% in 2007. But there is no evidence that any party believed that Pandora was obligated to pay above the 1.85% rate until 2010.

Pandora’s systems do not track its customers’ use of its services with any measure that corresponds to the 5.0 License definition of a session, and it was a complex undertaking for Pandora to calculate the amount it owed to ASCAP using that measure.²⁹ As a result of its dissatisfaction with this sessions component of its license, on October 28, 2010, Pandora sent a letter to ASCAP terminating its license and applying for a new license, pursuant to the terms of AFJ2, to run from January 1, 2011 through December 31, 2015.

Upon making a written request to AS-CAP, Pandora obtained, pursuant to AFJ2 § V’s requirement that “ASCAP is hereby ordered and directed to issue, upon request, a through-to-the-audience license to ... [inter alia ] an on-line user,” the right to perform all of the compositions in AS-CAP’s repertoire for that period, with only the proper payment rates to be determined, either through negotiation or by the rate court. Having been unable to agree with ASCAP on the proper price for the license after roughly two years of negotiation, and spurred by Sony’s impending withdrawal from ASCAP (as discussed below), on November 5, 2012, Pandora filed with this Court a petition for determination of reasonable licensing fees pursuant to AFJ2. See AFJ2 § IX.

VI. The April 2011 ASCAP Compendium Modification

A. Overview and Context

In 2011, ASCAP modified its Compendium to permit its members to selectively withdraw from ASCAP the right to license works to new media entities. This was an unprecedented event. Never before had ASCAP granted partial withdrawal rights to its members. As this Court would hold in 2013, the modification violated the terms of AFJ2. AFJ2 requires that ASCAP license to any applicant all of the works in its repertoire, and consequently if a publisher leaves a composition in ASCAP’s repertoire for some licensing purposes AS-CAP is required to license that work to any applicant. See In re Pandora Media, Inc., 12 Civ. 8035(DLC), 2013 WL 5211927, at *1 (S.D.N.Y. Sept. 17, 2013). In the year and a half that followed the adoption of the modification of the Compendium, three of the four largest music publishers withdrew their new media rights from AS-CAP. EMI’s withdrawal was followed by the withdrawal of Sony and then UMPG. Each of these publishers thereafter negotiated direct licenses with Pandora. Those negotiations and licenses have been a central feature of this litigation, and are discussed in detail below.

To place the Compendium modification in broader context, it was simply one of the many ripple effects that have followed the onset of the digital age in the music business, and the industry’s attempt to recover from the concomitant decline in some types of music sales. The modification of the Compendium came in response to pressure from ASCAP’s largest music publishers. These publishers were focused principally on the disparity between the enormous fees paid by Pandora to record companies for sound recording rights and the significantly lower amount it paid to the PROs for public performance rights to compositions. The modification was enacted despite significant concern about the impact of this change on AS-CAP, its writers and its independent publishers.

B. Public Performance Rights for Compositions versus Sound Recordings

A brief overview of the distinction between public performance rights in sound recordings and public performance rights in compositions is necessary to provide context for the discussion of the motivations of the largest publishers in effectuating the Compendium modification. A right to the public performance of a sound recording is the right to control the performance of one recording of a performance of a song. By contrast, a right of public performance in a composition is the right to control the use of the underlying musical composition itself. The latter right has been long recognized; but the right of public performance of a sound recording is a relatively new phenomenon and is restricted to digital services. The licensing fees for sound recordings are paid to an entity called SoundExchange, which collects and distributes these fees to the holders of sound recording copyrights.

In 1995, Congress passed the Digital Performance in Sound Recordings Act (“DPSRA”), which provided for the first time a public performance copyright in sound recordings. See 17 U.S.C. §§ 106, 114 (“Section 114”). Section 114 did not require all music users to obtain a license to perform a sound recording, but only services that “perform the [sound recording] publicly by means of a digital audio transmission.” 17 U.S.C. § 106(6) (emphasis added). Section 114 differentiates among services that are, in the meaning of the statute, “interactive” and “non-interactive.” An interactive service is defined as a service “that enables a member of the public to receive a transmission of a program specially created for the recipient, or on request, a transmission of a particular sound recording ... which is selected by or on behalf of the recipient.” 17 U.S.C. § 114(j)(7). If a digital service does not provide users with this level of control it is non-interactive. The distinction between interactive and non-interactive services is meaningful because “non-interactive” digital music services are eligible for “a compulsory or statutory licensing fee set by the Copyright Royalty Board [“CRB”] made up of Copyright Royalty Judges appointed by the Library of Congress,” see Arista Records, LLC v. Launch Media, Inc., 578 F.3d 148, 151 (2d Cir.2009), whereas interactive services must independently negotiate rates for sound recording licenses. Pandora is a non-interactive service within the meaning of Section 114.

Importantly for purposes of this proceeding, Congress also provided that this rate court (and the BMI rate court) may not take into account sound recording licensing fees in setting a rate for the licensing of the compositions themselves. The DPSRA provides that “[l]ieense fees payable for the public performance of sound recordings ... shall not be taken into account in any ... proceeding to set or adjust the royalties payable to copyright owners of musical works for the public performance of their works.”³⁰ 17 U.S.C. § 114®.

Ultimately, the CRB decided that the market for sound recording rights was materially different from the market for the public performance rights to musical compositions, and set rates for compulsory license fees for sound recordings at rates many times higher than the prevailing rates for the licensing of the public performance of the compositions. Consequently, Pandora pays over half of its revenue to record companies for their sound recording rights, and only approximately four percent to the PROs for the public performance rights to their songs.

The disparity between rates for the public performance of compositions versus sound recordings does not exist for most of ASCAP’s revenue streams since, as just explained, the need to acquire sound recording licenses only applies to services who conduct digital audio transmissions. Thus, there is no disparity at all when it comes to most of ASCAP’s business, including its general licensing program and its licensing of cable TV, broadcast TV, and terrestrial radio. Because only new media music services must acquire sound recording licenses, the PROs end up receiving far more money from public performance rights license fees for compositions than do the record companies from public performance license fees for sound recordings.

C. ASCAP-Publisher Negotiations Pñor to the Compendium Modification

Against this backdrop, music publishers assessed their options. In September 2010, music publisher EMI advised AS-CAP that it was contemplating withdrawing entirely from ASCAP. EMI Chief Executive Roger Faxon has explained that EMI wanted to withdraw because it believed that it was inefficient to license each right in the musical works and recordings it administered through different institutions. Faxon wanted EMI to be able to “unify the rights in the compositions that we represented so that a single negotiation with ... a customer who wanted the rights could encompass all rights ... necessary to empower their business.” Faxon also said that EMI was dissatisfied with the “delays” in ASCAP’s procedures and AS-CAP’s high operational costs.

Spurred by the potential loss of one of the four largest music publishers, ASCAP began in 2010 to explore a proposal to amend its Compendium to allow members to withdraw from ASCAP only the right to license works to new media users. It was, after all, only new media users who needed to acquire both a public performance and sound recording license. Not all of the ASCAP board was in agreement on this proposal. Large publishers were in general enthusiastic about such a change, but the songwriters and independent publishers were less so.

As noted, the largest publishers were fixated on the higher rates that record companies — often their corporate affiliates — were receiving from internet music providers, of which Pandora was the most prominent, for sound recording rights under Section 114’s compulsory license regime when compared to the rates that Pandora and others were paying for public performance rights under AFJ2. The major publishers viewed AFJ2 as preventing them from closing the gap between the composition rates and the sound recording rates. In the words of Sony’s Brodsky:

We were struck by the vast disparity between what record companies received from digital music services for the sound recording rights that they conveyed and what was paid for the performance right.

This concern over the discrepancy between the revenues generated for record labels and those generated for music publishers is repeated in many of the communications related to the adoption of the Compendium modification and the subsequent withdrawals by the publishers of new media rights from ASCAP. In many of those exchanges they focus their attention directly on Pandora. For example, an email from ASCAP General Counsel Joan McGivern to LoFrumento of July 30, 2010, brought up the possibility of permitting partial withdrawals from ASCAP as a means to attempt to close the gap in Pandora’s payments:

I spoke to Peter Brodsky at Sony late yesterday. He would like to meet with us ... to discuss — why publishers are not receiving as much as their record labels from Pandora and what options Sony might have, such as trying to license Pandora directly, withdrawing its rights, etc.

Similarly, an email of July 31, 2010 from McGivern to other top ASCAP officials confirms Sony’s focus on the gap in the payments made by Pandora:

Peter Brodsky at Sony, asked that we meet with him and two [outside lawyers] to discuss our Pandora license, and in particular, why publishers and writers are not receiving as much from Pandora as it is paying to SoundExchange.

ASCAP’s DeFilippis responded to this email by explaining that he had already had conversations with Brodsky about the same issue “about 2 months ago” and that “[h]e seemed to understand the basic reasons, i.e., CRB set rates, ASCAP/BMI market share.”³¹

The publishers believed that AFJ2 stood in the way of their closing this gap. They believed that because the two PROs were required under their consent decrees to issue a license to any music user who requested one, they could not adequately leverage their market power to negotiate a significantly higher rate for a license to publically perform a composition.

On occasion the publishers offered a second rationale for executing the new media withdrawals: the high cost of litigating a case in rate court. Martin Bandier, the CEO of Sony, cited this rationale in an email to Sony employees, in which he contended that “litigating a rate is an expensive and inefficient way to license our repertoire to new digital services.”

Against the backdrop of the urgency felt by the largest music publishers to close the gap between payments for composition rights and sound recording rights, other ASCAP members had their own separate concerns. Songwriters, and at least some independent music publishers, were concerned about the damage that might be wrought from the Compendium modification and the partial withdrawal of rights from ASCAP. Songwriters trusted AS-CAP to account reliably and fairly for the revenues ASCAP collected and to distribute the portion of revenues owed to writers promptly and fully. Songwriters were concerned about the loss of transparency in these functions if publishers took over the tasks of collection and distribution of licensing fees. They were concerned as well that the publishers would not manage with as much care the difficult task of properly accounting for the distribution of fees to multiple rights holders, and might even retain for themselves certain monies, such as advances, in which writers believed they were entitled to share. Overall, they were concerned about the increasing concentration of the publishing industry and the willingness by some, particularly Sony, to engage in direct licensing outside the framework of the PROs. These concerns ripened as the writers learned that Sony intended to follow EMI’s lead and take advantage of the Compendium modification to partially withdraw from ASCAP.

Some of this tension is captured in an email sent by ASCAP-member and composer [REDACTED] to LoFrumento on September 6, 2012. In that email, [REDACTED] explained the conflicts that he perceived between the major publishers and writers of ASCAP:

[W]riters and (the major) publishers differ. Writers, I believe are concerned with the health and well being- of AS-CAP. As small business owners we are dependent upon ASCAP for our success. ... Today’s publishers (the majors) are executives not owners. Their focus is on the well being of their company, their investors and their own perceived performance all of which is reflected in the quarterly bottom line. In their vision of the future, ASCAP plays an inconsequential role.

[REDACTED] was not alone among writers in his concern about the publishers’ plan for new media withdrawals. Writer [REDACTED] wrote in an email of August 28, 2012 to LoFrumento, that there was a “disintegration of trust between writers and publishers,” and that “the new breed of publishers was understood by writers to be motivated primarily by profits, and that writers would not look positively on ASCAP becoming a clearinghouse for processing direct licensing royalties.” [REDACTED] concluded by expressing his opinion about “the vital role ASCAP plays in protecting writers from the shark-infested waters of the music business.”³²

Tension between the major- publishers and the writers of ASCAP is not surprising given that the two groups’ interests are not perfectly aligned. To balance their competing interests, ASCAP’s internal rules are premised on equality in decision-making between writers and publishers. As LoFrumento testified:

[ASCAP’s] rules are geared towards equality between writers and publishers. [For example] [o]ur rules say that if you get a stream of revenue that was an adjustment -for the past, we normally go back to the past and make that' adjustment. There is not necessarily the same credibility that a publisher who got extra money would say, oh, well, this is for two years ago, well, let’s go back and make that distribution. It’s not a certainty what they would actually do with it.

As significantly, ASCAP provides writers with transparency. Again, in the words of LoFrumento:

Major, major driving issue is [that] with ASCAP [the writers] get transparency .... [They] know our rules and we take the money that we collect, take off our overhead and split it fifty-fifty. Our writers get that part of the fifty percent, the publishers get the other parts. It’s an equal division .... The writer’s greatest fear is that in the world of publishers collecting the money the splits will not be reflective of how AS-CAP splits the money.

Finally, the writers were concerned that to the extent that the major publishers pulled their significant resources out of ASCAP, the writers would have to shoulder a larger burden in paying for activities like licensing, advocacy, and litigation. In that vein, [REDACTED] urged LoFru-mento to “not let ASCAP become the ‘accounting firm’ for the publishers who want to withdraw rights.”

The large publishers were well aware of the. discomfort that at least some writers felt with the new media withdrawals and made the following argument to convince them to come on board: if the major publishers could get higher license rates by direct negotiations with new media companies outside, of ASCAP then those rates could be used in rate court litigation to raise the ASCAP license fees. The publishers found an ally on this issue in writer and ASCAP chairman Williams, who agreed with the new media rights withdrawal strategy. His email illustrates the strategy he pursued to get writers to support the publishers’ partial withdrawal of rights from ASCAP:

My job is to make this transition as smoothly as possible in the board room ... to assuage the fears of the writers who may see this as an ASCAP death knoll .... [W]e are in fact giving [the major publishers] the right to negotiate. The end result being that they will set a higher market price which will give us bargaining power in rate court.

As an internal debate swirled, the AS-CAP Board authorized management on September 16, 2010 to “examine alternative means of licensing digital media and to engage antitrust counsel.” In March 2011, ASCAP notified the Department of Justice of its consideration of a proposal to allow the withdrawals of new media licensing rights from ASCAP.³³

D. The Compendium Modification Allowing. New Media Withdrawals is Enacted.

On April 27, 2011, the ASCAP Board adopted a resolution to amend its Compendium to allow a member to withdraw from ASCAP its rights to license music to new media outlets, while allowing ASCAP to retain the right to license those works to other outlets. Six songwriter members of the Board abstained from the vote, but there was no vote in opposition.

The Compendium modification was executed by creation of Compendium Rule 1.12. It allowed any ASCAP member, on six months notice,³⁴ to “modify the grant of rights made to ASCAP ... by withdrawing from ASCAP the right to license the right of public performance of certain New [M]edia Transmissions.” The modified Compendium defines “New Media Services” — i.e., entities which make “New Media Transmissions” and which would be purportedly subject to a decrease in their ASCAP rights as the result of publisher withdrawals — as

any standalone offering by a ‘Music User’ ... by which a New Media Transmission of musical compositions is made available or accessible (i) exclusively by means of the Internet, a wireless mobile telecommunications network, and/or a computer network and (ii) to the public, whether or not, in exchange for a subscription fee, other fee or charge.³⁵

In the Compendium modification, ASCAP provided that ASCAP would

continue to have the right to license such works only to those New Media Services which are licensed under Li-eenses-in-Effect on the Effective Date of the Membership Modification, and only for the duration of such Licenses-in-Effect.

One effect of the Compendium modification was that major publishers could pull a writer’s works out of the PRO that the writer had decided to join. Although publishers had in the past considered a work to belong to the repertoire of the PRO to which the writer of the work belonged, in fact, it was a publisher that generally had contractual control over the licensing decisions for the work. With the withdrawal of rights from the PRO, the withdrawing publisher unilaterally removed the work from the PRO insofar as new media licensing rights were concerned.

The Compendium modification also allowed the withdrawing publishers to rejoin ASCAP at any point, eliminating any risk to the publisher if a withdrawal proved to be a bad idea. Section 1.12.6 of the Compendium provided that: “[a]ny Member may terminate its Membership Modification at any time upon written notice to ASCAP, and thereby grant back to ASCAP the rights previously withdrawn.” (Emphasis added.)

To manage the withdrawal process, the Compendium modification mandated, in Section 1.12.4, ASCAP’s creation of a list of works subject to any publisher’s withdrawal by “[n]o later than ninety days before the Effective Date” of the withdrawal. The publisher was required to notify ASCAP of any errors or omissions “within ten days of receipt of the List of Works.” Thus, ASCAP and the publisher would both have a list of works that would be affected by the withdrawal well in advance of the effective date of the withdrawal.

E. ASCAP Provides Administrative Services for Withdrawing Publishers.

EMI publicly announced in early May of 2011, within days of the adoption of the Compendium modification, that it would be withdrawing new media rights from AS-CAP. The turmoil caused by EMI’s decision was widespread. Confronted with the reality of losing this major publisher, on May 5, ASCAP’s LoFrumento made a proposal to EMI. He offered ASCAP’s services in distributing the EMI revenues to ASCAP members and to other songwriters and publishers who would be entitled to share in the revenues. LoFrumento argued at the time that

ASCAP is uniquely positioned to handle the distribution of these rights because it already distributes royalties from the online licensees; its operating ratio remains one of the lowest in the world and certainly the lowest in the US; its technology is leading edge and its databases are authoritative; and finally, its staff is truly professional.

LoFrumento also advised EMI’s Faxon that ASCAP had been flooded with inquiries since EMI’s announcement from both foreign and domestic rights holders and organizations. As he explained, many writers were concerned that EMI would not distribute royalties as carefully, accurately, and promptly as they had relied upon ASCAP to do.

Ultimately, EMI and other withdrawing publishers agreed to let ASCAP handle the distribution of royalties collected for the new media direct licenses that they negotiated. They executed “Administration Agreements”' fór this purpose. AS-CAP charged a fee of [REDACTED] for this service, which represented a very substantial discount from its ordinary charge to members. Essentially, ASCAP set a rate based on the direct costs associated with these functions. ASCAP was concerned that without a low rate, the withdrawing publishers would be tempted to use competing PROs to perform the administration services.

As a result of the publishers’ partial withdrawals from ASCAP, the burden on remaining ASCAP members to pay for all of the other functions that ASCAP performs for its members, including in Lo-Frumento’s words at trial, “membership, legislative, legal, senior management, [and] international costs,” increased. On the other hand, because ASCAP continued to administer the distribution of licensing revenues, the writers could continue to have confidence that they would actually receive the monies owed them by the withdrawing publishers. Finally, the Administration Agreements meant that the withdrawing publishers faced little downside in withdrawing new media rights. They could continue to enjoy the benefits of having ASCAP perform burdensome back-office tasks while licensing internet music entities directly.

VII. A Second Compendium Modification in December 2012: the “Standard Services” Agreement

At the urging of Sony, another change to the Compendium, executed in December 2012, further reduced the burdens on withdrawing publishers. The modification allowed the publishers to target large new media entities for direct licensing negotiations and to effect withdrawals of rights from ASCAP solely with respect to those large licensees.³⁶

The December 2012 amendment permitted a member that was withdrawing under Section 1.12 of the Compendium to indicate that it wished to leave to ASCAP the right to license certain new media services that paid to ASCAP license fees of less than $5,000 per year. Where the withdrawing member indicated that it was only withdrawing new media rights “in part,” ASCAP continued to license new media services for the member that were defined in the Compendium as “Standard Services.” As a consequence, smaller new media entities could avail themselves of an ASCAP license so long as they accepted ASCAP’s 5.0 License (or its successor licenses) without negotiation.

ASCAP’s DeFilippis offered the following explanation for the adoption of the Standard Services exception for the withdrawal of new media licensing authority:

Given the rapidly changing marketplace and the low barriers to entry, new digital music services launch quite frequently. Many will never gain traction with listeners or generate substantial revenue. From the perspective of the withdrawing music publishers, they lacked the necessary staff and infrastructure to track the thousands of small music users that wished to license their music.

Sony’s Brodsky stated that Sony wanted this revision to the Compendium so that Sony’s withdrawal could be limited “to just the music services that we wanted to enter into direct deals with.”³⁷

VIII. Pandora Negotiates Direct Licenses with EMI, Sony, and UMPG and Fails to Negotiate an Agreement with ASCAP.

A. The Pandorar-EMI License Negotiations

Upon learning in May 2011 of EMI’s withdrawal of its new media licensing rights from ASCAP, Pandora immediately began to negotiate with EMI for a license to its catalog. The negotiations were not contentious and the contours of the license were quickly settled. Indeed, in their very first substantive discussion, which occurred on June 6, EMI confirmed that it would be using 1.85% as the headline rate, and hoped to have the agreement effective as of January 1, 2012. The collegial tone is reflected in handwritten notes by Pandora’s Rosenbloum. Rosenbloum colorfully recorded that EMI was “not looking to screw anyone.”

EMI’s Faxon testified that the rate in the EMI-Pandora license was “freely agreed to.” EMI was less concerned with the precise rate than the flow of revenue into EMI. Because the ASCAP deductions from gross receipts would be smaller, EMI viewed the license terms with Pandora as a “substantial improvement.”

The 1.85% rate in the Pandora-EMI agreement was the same rate that was available to Pandora under ASCAP’s 5.0 License for a non-interactive service. A July term sheet with EMI reflected this rate and an expectation that the agreement would have a two year term. It also reflected calculations premised on EMI’s estimate that it had approximately a 20% market share at the time.

During the ensuing months, the parties discussed the size of an advance that Pandora would pay to EMI, among other things. Meanwhile, Pandora continued to pay its licensing fees to ASCAP.

The licensing agreement, although not executed until March 16, 2012, covered the two-year period January 1, 2012 to December 31, 2013. Pandora agreed to a license that provided EMI with a pro-rata share of 1.85% of Pandora’s revenues.³⁸ The EMI agreement did not contain any “per-session” component like that included in the ASCAP 5.0 License. It required Pandora to pay a non-refundable advance of [REDACTED], which Pandora was confident would be exceeded by its payments to EMI over the course of the license. In addition, the EMI agreement permitted Pandora to take up to a [REDACTED] adjustment to revenue for advertising expenses.³⁹ This adjustment included commissions paid to internal advertising sales personnel, but only if Pandora were able to obtain an adjustment for internal advertising expenses in connection with an agreement from another major music publisher or PRO.

Finally, the agreement included a most-favored-nation clause, or “MFN,” for the benefit of Pandora. The agreement contemplated a prospective decrease in the headline rate from 1.85% to as low as 1.70% if Pandora succeeded in obtaining a lower rate for licensing a repertoire as large or larger than EMI’s catalog. It similarly allowed for an increase in the advertising expense adjustment up to [REDACTED],

The reference to the 1.70% rate was prompted by the recently announced settlement of rate court litigation between ASCAP and the RMLC in January 2012. As described above, that license provided for a blanket license rate of 1.70% of revenue and a 25% deduction for advertising expenses in connection with new media.

B. The Pandorctr-ASCAP License Negotiations

As noted above, Pandora had terminated its license with ASCAP on October 28, 2010 because of its concern over the calculation of the per-session rate in the 5.0 License, and had applied at that time for a new license for the calendar years 2011 through 2015. It remained an applicant for such a license throughout 2011 and 2012, as ASCAP adopted its modification to the Compendium and as EMI withdrew new media rights from ASCAP.

On September 16, 2011, Pandora executed an interim license agreement with AS-CAP effective as of January 1, 2011. It adopted the 5.0 License rate of 1.85% without any per session fee. The agreement noted the parties’ competing positions on several issues, including Pandora’s position that the adjustment for advertising expenses should apply to its internal advertising expenses.

Roughly a year later, on September 28, 2012, Pandora learned that Sony was also withdrawing its new media rights from ASCAP. With its discussions with ASCAP “languishfing]”, and with Sony’s withdrawal from ASCAP due to take effect at year end, which was just weeks away, Pandora filed this rate court petition on November 5.

Pandora’s filing in rate court angered some in the ASCAP community, particularly the major publishers. They expressed their outrage not only to Pandora, but also to its outside counsel, the law firm Greenberg Traurig, LLP. The day after the rate court filing, UMPG’s Horowitz called one of Pandora’s attorneys at Greenberg Traurig. As Horowitz promptly memorialized in an email to ASCAP’s LoFrumento, Horowitz

told [Pandora’s outside counsel], as a “friend” of the firm, that I thought both the firm and Pandora are completely tone deaf. That whether his firm has the legal right to rep Pandora in litigation, the firm has lost huge goodwill with writers and artists by doing so. And that filing now for a rate court proceeding against ASCAP ... had the effect of unifying artists, writers, and PROs against Pandora.

Horowitz also gave some advice to Lo-Frumento regarding ASCAP’s negotiating stance with Pandora. His advice boiled down to two words: be strong. Horowitz wrote:

My take: [Pandora’s outside counsel] and Pandora are scared. They just want to settle with ASCAP and settle fast. Be strong. Time is on your side. Pandora is now under intense pressure to settle with ASCAP. They have to put this behind them. You can really push Pandora and get a much better settlement as a result. They are reeling. They will pay more, a lot more than they originally intended, to do that.

Horowitz forwarded this same email to other ASCAP board members, including Sony’s Martin Bandier, and BMG Music Publishing’s Laurent Hubert. Besides these ASCAP Board members, Horowitz sent the email to David Israelite of the National Music Publishers Association (“NMPA”), which is a music industry trade group based in Washington, D.C. LoFru-mento assured Horowitz that he was approaching Pandora with the mindset Horowitz advocated.

Horowitz continued to apply pressure on Pandora. He called Pandora’s outside counsel a second time, about a week after his first call, and reported once more to LoFrumento. As Horowitz explained to LoFrumento on November 14, Pandora’s outside counsel “has been spending hours on fallout from their repping Pandora. They are embarrassed. [Pandora’s counsel] said they will withdraw from repping Pandora in the next few weeks if the [rate court litigation with ASCAP] doesn’t settle.”

Not surprisingly, given the fallout from Pandora’s filing of the rate court petition, and with the deadline for Sony’s withdrawal from ASCAP approaching, the negotiations between Pandora and ASCAP intensified. Had those negotiations succeeded, of course, this rate court action would have become moot.

By the end of November, Pandora believed that it had reached an agreement on terms with ASCAP, although it understood that the agreement needed final approval from ASCAP. Pandora emailed a term sheet to Pandora on November 29. AS-CAP had assured Pandora that if they finalized their agreement before the end of 2012, the license would cover the Sony repertoire since the Sony withdrawal from ASCAP was only effective as of January 1, 2013.

LoFrumento decided to reject the license that his team had negotiated with Pandora. He knew that either way he faced litigation. He knew that if he executed the license, Sony would sue ASCAP. Sony had threatened to sue ASCAP in the event any license agreement with Pandora that encompassed the Sony repertoire was executed before the end of 2012.⁴⁰ Sony had also notified ASCAP that it might not use ASCAP for administration services if ASCAP issued a license to Pandora. Lo-Frumento was already facing rate court litigation with Pandora. Given the pressure being exerted on him by both Sony and UMPG, LoFrumento was only willing to execute a license with Pandora that included a headline rate of at least 2.5%, and he knew Pandora was not willing to pay that much.

LoFrumento advised the Law and Licensing Committee of ASCAP’s Board of Directors on December 12 that he intended to reject the terms Pandora and AS-CAP had negotiated. Everyone understood that that meant that the rate court proceeding would go forward. None of the Committee members asked for a description of terms Pandora and ASCAP had negotiated or to discuss LoFrumento’s decision.⁴¹

Thus, in mid-December 2012, ASCAP set itself on a course to have its rate for licensing Pandora set in this rate court proceeding, despite the cost associated with that litigation. The decision was made in the midst of great turmoil, uncertainty and pressure. The partial withdrawals of new media rights by major publishers, who collectively controlled about 50% of ASCAP’s music, threatened to make ASCAP a weaker organization. Sony and UMPG had also made clear to LoFrumento that they wanted to negotiate direct licenses with Pandora and opposed ASCAP entering into a final license with Pandora. There was, of course, a chance that by placating the major publishers, they might later exercise their option to rejoin ASCAP for all purposes. LoFru-mento also had to consider the writers who had become restive and were doubtful about the supposed benefits of the publisher withdrawals. In the midst of all of this, LoFrumento cast the lot of ASCAP with the withdrawing major publishers and chose to let the rate court decide the dispute between Pandora and ASCAP. On December 14, ASCAP surprised Pandora and rejected the terms they had negotiated.

C. The Pandora-Sony License Negotiations

Since the Fall of 2010, Sony had been discussing with ASCAP the possibility of a withdrawal of rights so that it could directly negotiate with Pandora. In July 2012, Sony notified ASCAP that it would exercise its right under the modified Compendium to withdraw new media rights. In late September, Pandora (along with the rest of the world) learned that Sony would be withdrawing new media rights from ASCAP effective January 1, 2013. As already described, Sony worked with AS-CAP during late 2012 to effect a second change to the Compendium that would permit a partial withdrawal of new media rights from ASCAP. Under the Standard Services exception, Sony allowed ASCAP to retain licensing authority for smaller new media services while assuming responsibility for the direct licensing of larger entities such as Pandora.

As of the Fall of 2012, Sony was the world’s largest music publisher. It owned or controlled between 25% and 30% of the market. It had taken this frontrunner position in the summer of 2012, when it became responsible for licensing EMI’s catalog.⁴² Combined, the Sony and EMI catalogs contain roughly 3 million songs.

While the effective date of the withdrawal came as a surprise to Pandora, Pandora had been aware that the withdrawal was a possibility ever since ASCAP adopted the Compendium modification. Indeed, in the Spring of 2012, Pandora wrote to the Federal Trade Commission in opposition to Sony’s acquisition of EMI and referred to this very possibility. Noting that a Sony withdrawal from the PROs would require Pandora to negotiate directly with Sony and that Pandora would be faced with a choice of either paying higher rates “or continuing to operate without Sony’s songs,” Pandora’s Kennedy expressed concern that the combination of the Sony and EMI catalogs would give Pandora “no choice” but to enter into a direct license for the content. While Pandora “could survive without access to Sony’s musical content,” it “could not survive without access to the combined Sony and EMI cata-logues.”

The first substantive discussion between Pandora and Sony occurred in a telephone call on October 25 between Sony’s Brodsky and Pandora’s Rosenbloum.⁴³ Sony promptly set the tenor for the negotiations with a not-too-veiled threat. Brodsky stated “[fit’s not our intention to shut down Pandora.” In his many years of negotiating music licenses, Rosenbloum testified that had never before heard such a threat. In some ways, this threat put on the table no more than what was obvious. Sony’s works were already being played on Pandora; they were incorporated in the MGP. Unless Pandora could do without those works and remove them from its repertoire by January 1, Pandora had to obtain a license from Sony or face crippling copyright infringement claims. Sony was in the driver’s seat and the clock was ticking.

The remainder of the conversation was largely devoted to Sony’s statement of the reasons why it needed Pandora to pay for the public performance of music at a substantially higher rate. The principal reason was the “massive unfair disparity” between what Pandora was paying the record labels for sound recording rights and what it was paying the music publishers for composition rights. Brod-sky explained that if the labels were getting 50% of Pandora’s revenue, then it would be “fair” for music publishers to get 12% of the revenue, although Brodsky acknowledged that Pandora could not afford to pay that much. As Brodsky emphasized, it was the “differential” between the rates paid to the labels and the publishers that was the problem, and that Pandora was really just caught in the middle of a tug of war between the labels and publishers. Brodsky admitted that if the labels were getting only 25% of Pandora’s revenue, then Pandora’s current industry-wide rate of 4% for the licensing of rights to publicly perform compositions would probably be alright and there wouldn’t be any need to increase it.

Brodsky identified a second, subsidiary reason for needing Pandora to pay more. Referring to the writers’ skepticism over the motives of the publishers in withdrawing from the PROs, Brodsky added that Sony had to show the writer-members of the PROs that there was some “reasonable justification” for Sony’s withdrawal. At the end of the call, Rosenbloum threw out the possibility that Pandora might pay a “modest” increase to Sony for a year as they all waited to see what happened to the rates Pandora was paying the labels.

Following this conversation, Pandora decided on a two-prong strategy. It would intensify its efforts to get an ASCAP license before the end of the year. To bring ASCAP to the negotiating table it filed its petition in this rate court for an ASCAP license on November 5. Secondly, Pandora attempted to obtain leverage in its negotiations with Sony. It requested a list of the Sony catalog so that it could take the Sony works off, or at least threaten to take them off, of the Pandora service if no deal could be reached. In his years of negotiating licenses, this was the first time that Rosenbloum had ever requested a list of works from a publisher.

Pandora’s first request for the list came on November 1, 2012, in an email from Rosenbloum to Brodsky. Rosenbloum advised Brodsky that:

I wanted to follow up with you about our conversation last week regarding Pandora. As I mentioned, given the uncertainties around Sony/ATV’s and EMI’s position with respect to webcasting rates, Pandora has decided that it needs to be prepared to take down all Sony/ ATV and EMI content in the event we are unable to agree on rates by the end of this year. In that regard, please let me know if you can provide us with an electronic listing of Sony/ATV and EMI repertoire.

On a related note, as the end of the year is rapidly approaching, we look forward to receiving a rate proposal as soon as possible (to the extent that EMI and Sony/ATV are still interested in moving forward with a direct license agreement).

Brodsky received this request for a list of the Sony works, but never responded. In their telephone conversations during the month of November, Rosenbloum reiterated the request for a list of works on several occasions but never got any response. Rosenbloum repeated the request once more at a breakfast meeting that he and Pandora’s Kennedy had with Sony’s Brodsky and Bandier on November 30. Again, Sony did not respond.

The list of Sony works was potentially important for several purposes, and Pandora referred to those several purposes in its discussions with Sony. In addition to wanting to be able to remove the Sony works from its service if Pandora and Sony could not come to terms, Pandora needed the list so that it could understand how to apportion any payments between the EMI and Sony catalogues since the payments would apparently be made at two different rates. Pandora also wanted the list so it could evaluate whether the substantial, non-refundable advance that Sony was demanding would likely be recouped.

Sony had a list readily at hand, since the Compendium required that a publisher and ASCAP work together during the 90 day period before the effective withdrawal date to confirm precisely which works were being withdrawn. Sony understood that it would lose an advantage in its negotiations with Pandora if it provided the list of works and deliberately chose not to do so. Brodsky’s explanation at trial that he did not provide the list because he believed that negotiations were proceeding smoothly and did not want to impose an unnecessary “burden” on Sony’s staff is not credible. The negotiations were not going smoothly; the list had already been prepared and its production imposed no burden. As Brodsky recognized in his testimony, the list was “necessary” to Pandora in the event the parties did not reach a deal. Sony decided quite deliberately to withhold from Pandora the information Pandora needed to strengthen its hand in its negotiations with Sony.

Ultimately, Sony made an offer to Pandora in early December. Still hoping to reach an agreement with ASCAP which would obviate the need for license from Sony, Pandora did not respond to the offer or to a follow-up email of December 6.

On Friday, December 14, with two weeks left in the year, and one week remaining before the music industry took its annual holiday break, ASCAP notified Pandora that it would not execute the agreement they had negotiated. The following Monday, Pandora urgently made two renewed written requests for' the list of Sony’s works, one to Sony and another to ASCAP.

Since the repeated requests from Pandora’s outside counsel Rosenbloum had gone unanswered, Pandora’s general counsel Delida Costin sent her own email to Brodsky on December 17 requesting the list of works.⁴⁴ Not wishing to empower Pandora, Sony never responded.⁴⁵

That same day, Pandora also asked AS-CAP for the list of Sony works in ASCAP’s repertoire. It would have taken ASCAP about a day to respond to Pandora’s request with an accurate list of the Sony works. But, ASCAP, like Sony, stonewalled Pandora and refused to provide the list.

In making the request to ASCAP, Pandora’s counsel wrote that “Pandora must prepare for the possibility of being unlicensed by Sony/ATV or ASCAP for [Sony’s] works effective January 1st, so it is important that we get this information from ASCAP as soon as possible.” This request set off a flurry of emails within ASCAP. ASCAP ultimately decided to contact Sony to see if it would give its permission to share the list of works. On Wednesday, December 19, ASCAP notified Sony of Pandora’s request and that it would be providing Pandora with the list of Sony works that ASCAP had previously given to Sony in connection with its withdrawal of rights. Not surprisingly, given its own refusal to 'Share the list with Pandora, Sony did not give ASCAP permission to provide the list.⁴⁶ As a result, neither Sony nor ASCAP provided the list of works to Pandora.

If either Sony or ASCAP had provided Pandora with a list of the Sony works, Pandora would have been able to remove Sony’s compositions from its service within about a week.⁴⁷ Although ASCAP attempted at trial to show that Pandora could have used public sources of information to identify the Sony catalog, it failed to show that such an effort would have produced a reliable, comprehensive list, even if Pandora had made the extraordinary commitment necessary to try to compile such a list from public data.

The terms of the Pandora license with Sony were negotiated in four business days during the single week that ran between ASCAP’s rejection of the Pandora term sheet and the start of the holiday break. On December 18, Brodsky sent Rosenbioum a term sheet. As proposed in that document, the license term would be one year, starting January 1, 2013. It required Pandora to pay a non-refundable but recoupable advance of [REDACTED] and a non-refundable [REDACTED] advance as an administrative fee. The royalty rate was set at Sony’s pro-rata share of an industry-wide rate of 5%. Sony understood this to be a 25% increase over the then prevailing industry rate of approximately 4%. In his March 2013 report to his Board of Directors, Sony’s Bandier bragged that Sony had leveraged its size to get this 25% increase in rate.

The term sheet also allowed Pandora to take an adjustment for advertising expenses of up to [REDACTED]. This would include a deduction for Pandora’s internal advertising sales personnel “to the extent deducted from revenue in connection with the calculation of performance right license fees under Licensee’s agreements with other major music publishers” or PROs.

On a December 21 draft of the agreement, Rosenbioum wrote to Sony that, to the extent Pandora was willing to conclude the license without receiving “actual data” from Sony, it “at least” needed confirmation of the approximate percentage of the ASCAP repertoire that consisted of Sony and EMI compositions. Sony’s Brodsky responded to this request not by giving a list but with a rough estimate that the Sony/EMI share of the ASCAP repertoire was 30%.

Although the agreement was predominately on Sony’s terms, the December 21 draft agreement did include a change in Pandora’s favor regarding the adjustment for advertising sales from a previous draft of the agreement. Unlike a draft delivered from Sony to Pandora on December 18 which only allowed for a deduction from internal advertising costs if Pandora got a similar deduction from another PRO or publisher, the December 21 agreement allowed for a reduction of up to [REDACTED] that included both outside commissions and direct internal costs of such sales without reference to another agreement with a PRO or publisher. The parties executed a Binding Heads of Agreement on December 21, 2012.

By mid-January 2013, and despite the existence of a confidentiality agreement, Sony leaked the key terms of the Pandora license to the press.⁴⁸ The headlines in three articles said it all: “Sony/ATV ‘Now Has the Power to Shut Pandora Down “Sony/ATV gets 25 percent increase in Pandora royalties”; and “Sony/ AIN’S Martin Bandier on new ‘quite reasonable’ Pandora deal.” A New York Post article featured a photograph of Sony’s Bandier in shirt sleeves with a large cigar in his mouth, as it reported that Sony had “wrangled a 25 percent increase in royalties” for a one year license. Bandier was quoted as saying that “[a]t the end of the day, we got a terrific deal for our songwriters. Our thinking has been vindicated.” In his interview with Billboardbiz, reported on January 18, Bandier explained that the rates “are quite reasonable. When you compare it to the rate record companies are getting, it was really miniscule.” One article reported: “[m]any other publishers were rooting for Sony to deliver a higher rate ... so that if [the PRO’s] deal with Pandora heads to rate court, the judge will consider the Sony rate the market rate and raise performance royalties accordingly.” The press coverage focused on Sony’s leverage in negotiations due to its outsize market power: “Look a little closer, and this is ultimately a very lopsided negotiation .... Pandora absolutely needs Sony’s catalog to run an effective radio service. And if they don’t pay what Sony/ATV wants, they can’t use it, by law.”

D. The Pandorar-UMPG License Negotiations

Pandora did not have to wait long for the next publisher to leave ASCAP and demand a yet higher rate for a direct license. In February 2013, Pandora learned that UMPG was scheduled to withdraw its new media licensing rights from ASCAP effective July 1, 2013.

UMPG’s Horowitz had notified ASCAP’s LoFrumento at the end of November 2012 that UMPG intended to withdraw new media rights from ASCAP. Horowitz told LoFrumento that the ASCAP rate for Pandora was “too low.” In making this assertion he referred to Spotify’s 10.5% rate.⁴⁹ Horowitz added that he believed Sony would be getting a much better rate from Pandora than ASCAP would achieve. Horowitz asked ASCAP for a waiver of the Compendium’s notice period; he wanted to withdraw effective January 1, 2013. AS-CAP denied the waiver request.

The negotiations between UMPG and Pandora were even more contentious than the negotiations between Sony and Pandora. After difficult conversations in March in which UMPG asked for an industry-wide headline rate of 8%, Pandora essentially placed the negotiations on hold. While a license agreement was executed in June, it was for a six month term only and was contingent on several events.

The negotiations between these parties were conducted principally by Horowitz for UMPG and Rosenbloum and Kennedy for Pandora. Kennedy and Horowitz knew each other fairly well. They had been dealing with each other for years in connection with sound recording rights. Horowitz had run the “label” side of Universal’s business until April 2012, when he was transferred to the music publishing side of the organization. Horowitz brought into these 2013 negotiations with Pandora, therefore, a thorough grasp of the history behind the requirement that Pandora pay a substantial portion of its revenue to obtain sound recording rights, and perhaps as significantly, a desire to show that he could be similarly effective in achieving an enhanced payment from Pandora for composition rights.

In their first substantive meeting, which occurred on March 22, Horowitz quizzed Kennedy at some length about the state of Pandora’s business. Horowitz then moved to a discussion of Pandora’s need for a license from UMPG, uttering what Kennedy took to be an implicit threat. Horowitz said “we want Pandora to survive.”

Like Sony, Horowitz justified a substantial increase in the rate Pandora needed to pay by stressing the disparity between the rates at which Pandora paid for sound recording rights and public performance rights for compositions. Showing confidence that he knew the material terms of the Sony-Pandora license, Horowitz repeatedly asked Kennedy, (as Kennedy paraphrased) “how did you get Marty [Bandier] at Sony to agree to such a low payment?”

Avoiding invitations to discuss the specific terms of the Sony license, Kennedy explained to Horowitz that Pandora felt it should be treated just like entities covered by the ASCAP-RMLC license. Kennedy argued that Pandora was competing for listeners with, and taking listeners from, radio companies covered by the RMLC deal. Those radio services, including iHeartRadio, would be paying the PROs for many years into the future at a rate lower than the roughly 4% range that Pandora had been paying the PROs.

Kennedy indicated a preference for negotiating with the PROs, but added that, if UMPG wanted to negotiate directly with Pandora, then UMPG should provide Pandora with a list of the withdrawn compositions and UMPG’s proposal for a rate. Horowitz said he was “not sure” he was able to provide Pandora with a list, and indicated that Pandora should just make a deal based on UMPG’s representation of its overall market share.

Rosenbloum had his own detailed conversation with Horowitz about a week later, on March 29. Horowitz and David Kokakis of UMPG asked what Rosenbl-oum thought the next steps were in commencing formal license discussions. Ro-senbloum responded that UMPG needed to provide a proposal and a list of UMPG works so that Pandora could “better understand the scope of rights at issue”. Horowitz responded that UMPG was prepared to provide a list so long as it was covered by non-disclosure agreement (“NDA”).

In their conversation, Horowitz expressed amazement at the Sony rate for Pandora, and indicated that he felt an industry-wide rate of 8% of revenue would be reasonable, particularly in light of what Pandora was paying to the record labels for sound recording rights. Rosenbloum was aghast. He told Horowitz that in his 20 years in the music industry he had never encountered a situation in which a licensor suggested that rates should effectively double overnight, going from 4% to 8%. Rosenbloum observed that Sony “was apparently more willing to adopt a businesslike approach” and that Sony’s Bandier “understood Pandora’s realities.” Skipping over the fact that UMPG wanted its rate to serve as a benchmark for all future PRO licenses, Horowitz responded that the 8% rate would not have such a significant impact on Pandora because UMPG’s market share was only about 17%.

Horowitz bluntly reminded Rosenbloum that Pandora did not have much negotiating leverage. Rosenbloum described Horowitz as asking

what Pandora would do if we could not reach an agreement as to rates (suggesting that they have all of the leverage). I told him if UMPG is unwilling to move from its 8% figure it might be creating a situation where Pandora would have no choice other to take down all UMPG repertoire. [Horowitz] indicated that he definitely was not seeking such a result .... It was at that point in the conversation that the tone began to change a bit and [Horowitz] became somewhat less “positional” in his approach.

In late April 2013, UMPG provided to Pandora a complete list of the UMPG works in the ASCAP repertoire, but in a way that prevented Pandora from using the information to remove UMPG compositions from its service. The list was subject to an NDA. The list itself was the very information that the NDA deemed confidential.⁵⁰ The NDA provided that:

[Pandora] has requested that Universal provide to [Pandora] titles of songs in Universal’s music catalog controlled by ASCAP, corresponding writer names and corresponding shares owned or controlled by Universal and such writers, all of which Universal deems to be confidential (“Confidential Information”). [Pandora] agrees not to use any Confidential Information for any purpose except to evaluate and engage in discussions concerning a potential business relationship between the Parties.

The NDA then restricted Pandora’s use of the list. It provided that

Pandora correctly interpreted'this provision as forbidding it from using the list to remove the UMPG works from its service.⁵¹

On May 21, Pandora’s Rosenbloum and Horowitz met. While Horowitz expressed his admiration for Pandora and assured its representatives that UMPG wanted it to thrive, he did not move much from his initial proposal for a 8% industry rate, only revising it downward to 7.5%. Rosenbloum responded by reminding Horowitz that ASCAP had agreed to a 1.70% ASCAP rate with a generous advertising deduction for Pandora’s competitor, iHeartRadio.

Pandora believed that UMPG’s rate request was unreasonable, and that UMPG would be inflexible in any negotiations. Therefore, instead of engaging further with UMPG, Pandora went on the offensive. First, Pandora purchased KXMZ-FM, a terrestrial radio station in Rapid City, South Dakota. With this purchase, Pandora hoped, to shoehorn itself into the ASCAP-RMLC license. Then, Pandora believed, it would be in a position to argue that it was entitled to the RMLC 1.70% rate.⁵² The agreement of purchase is dated June 5, 2013. Second, on June 11, Pandora moved in this Court for partial summary judgment. Its motion argued that any purported new media withdrawals by publishers following the 2011 ASCAP Compendium modification did not affect the scope of the ASCAP repertoire subject to Pandora’s application for an ASCAP license.

During this interim period, as it worked on these two projects, Pandora did not respond to emails from Horowitz. In his emails to Pandora, Horowitz expressed increasing levels of anxiety and exasperation about Pandora’s “radio silence.”⁵³ Then, on the heels of announcements of its purchase of a radio station and the filing of the summary judgment motion, Pandora reached out to UMPG. In an understatement, Pandora observed in a June 13 email that “[a]s you may have read or heard, this week Pandora made a couple of announcements that are related to our discussions regarding a direct license with UMPG.” Pandora expressed optimism that it would win the summary judgment motion and recognition of its entitlement to the RMLC license, all before July 1. But, it added,

In the unlikely event we don’t have a decision on either of these points by July I, it is our preference to continue to perform works in the UMPG catalog. To help facilitate that, we propose accepting UMPG’s 7.5% of revenue offer on a provisional basis starting July 1, 2013, pending the Court’s rulings, with the understanding that if the ASCAP rate court subsequently rules in Pandora’s favor that Pandora will immediately thereafter — -and on a retroactive basis back to July 1, 2013 — license the right to works in the UMPG repertory through ASCAP at whatever rate the rate court decides.

The parties memorialized a six month license agreement on July 1, 2013. The agreement provided for an industry rate of 7.5%, with no deduction for advertising expenses, which would be contingent on the two contingencies outlined in the June 13 email from Pandora. The agreement provided that “in the event that a final decision not subject to any further appeal is rendered in the pending ASCAP Rate Court ... [that] UMPG’s July 1, 2013 withdrawal from ASCAP of [New Media licensing rights] ... is not effective” or if “Pandora’s acquisition of the KXMZ-FM qualifies Pandora for the RMLC-ASCAP license” then the agreement would “be of no further force or effect.”

UMPG refused Pandora’s request that the agreement reflect that it was non-precedential and could not serve as a benchmark in rate court proceedings. Instead, both parties reserved their rights on the question of whether the agreement could serve as a benchmark in this rate court proceeding.

IX. September 17 Partial Summary Judgment Opinion

On September 17, 2013, Pandora’s motion for partial summary judgment was granted. See In re Pandora Media, Inc., 2013 WL 5211927. The Opinion held, inter alia, that AFJ2 prohibited ASCAP from withdrawing from Pandora the rights to perform any compositions over which ASCAP retained any licensing rights. Consequently, the publishers’ purported withdrawals of only new media rights under the Compendium modification were held inoperative. The Court found that AFJ2 prohibited a regime in which publishers allowed ASCAP to license a composition to some music users but not others. AFJ2 required each work that was in the ASCAP repertoire to be available to any user who requested a blanket license. The publishers, of course, remained free to withhold works from ASCAP entirely.⁵⁴

X. Other Licensing Agreements Put Forth as Benchmarks

There are two other sets of licenses which. ASCAP argues are “confirmatory” benchmarks for the reasonableness of AS-CAP’s proposal for its license with Pandora. One is Pandora’s own license with SESAC. The other are licenses held by Pandora’s competitor Apple iTunes Radio. A description of these licenses will conclude this section of the Opinion.

A. The Pandorar-SESAC License

Since 2007, Pandora has had a blanket license from the PRO SESAC for the right to publicly perform musical compositions in the SESAC repertoire. Pandora and SESAC each have the option to terminate the license each year, but neither has exercised that option.

SESAC is the smallest of the three PROs. It is an invitation-only organization. SESAC proclaims that it is a selective organization, taking pride in having a repertory based on “quality, rather than quantity.”

Pandora’s SESAC license rate increases annually by the greater of [REDACTED] percent or an amount tied to the percent increase in [REDACTED]. The annual rate escalation has been justified as a mechanism to account for SESAC’s growing repertoire. The rate in the SESAC license started at [REDACTED] of Pandora’s revenue in 2007, and the rates from the period of 2011 to 2015 (assuming the [REDACTED] increase) therefore start at [REDACTED] in 2011 and escalate to [REDACTED] in 2015. [REDACTED].

Calculating the implied rate applicable to ASCAP depends on what SESAC’s market share in compositions is — a figure that is impossible to know with certainty. SE-SAC does not publicly report its revenue or its catalogue of compositions. There is no public consensus as to what share of the total number of musical compositions are in the SESAC repertoire. Pandora’s Kennedy testified that based on SESAC’s representations during negotiations, he understood the SESAC PRO share to be 10% in 2007.⁵⁵ Pandora had no ability to confirm that number, but it is a number that has appeared in court decisions. See, e.g., MobiTV, Inc., 712 F.Supp.2d at 221; United States v. ASCAP (Application of Youtube, et al), 616 F.Supp.2d 447, 453 (S.D.N.Y.2009). Using the 10% figure for SESAC and a 45.6% figure for ASCAP, and making no adjustments for an increase in SE-SAC’s share and a concomitant decline in ASCAP’s share, the parties calculate an implied ASCAP rate of [REDACTED] in 2011, [REDACTED] in 2012, [REDACTED] in 2013, [REDACTED] in 2014, and [REDACTED] in 2015.

B. Apple’s iTunes Radio Licenses with Publishers and PROs

In September of 2013, Apple launched its iTunes Radio service. From a user perspective, iTunes Radio operates like Pandora. Both are customized radio. Users seed an iTunes Radio station by identifying a song, artist, or genre. They provide feedback by signaling “Play More Like This” and “Never Play This Song”. And like Pandora, iTunes Radio uses an algorithm to select songs that users are likely to enjoy in light of their initial selection and feedback.

Apple’s iTunes Radio is available in a free, advertising-supported format, and through a program called “iTunes Match,” in which users pay an annual fee for a bundle of services, one of which is access to iTunes Radio without advertising. Through a subscription to iTunes Match, subscribers have access to their entire personal music library or “locker”,⁵⁶ allowing them to stream music wherever they are. iTunes Radio is only available within the Apple ecosystem.

Pandora considers iTunes Radio a major competitor. Upon its launch, Pandora tracked the impact of iTunes Radio on Pandora closely. While it appears that the launch of iTunes Radio in the Fall of 2013 had a measurable (albeit relatively small) impact on Pandora, after a short period of time that impact appeared to decline. Pandora has continued to grow despite the presence of iTunes Radio. This may be due to several reasons, including unique characteristics of Pandora’s service, the availability of a Pandora app on Apple devices,⁵⁷ and the fact that iTunes Radio is only available on Apple devices.

Apple negotiated licenses for its iTunes Radio service in order to announce the launch of the service at its Worldwide Developers Conference in June 2013. Its license agreement with Sony is dated June 6, 2013, and is identical in its material terms to the other licenses that Apple entered with publishers. The whereas clauses of the agreement with Sony emphasize the complementary relationship between iTunes Radio and the sale of music through the iTunes Store. The clauses explain inter alia that the parties to the agreement wish to deter piracy and compensate songwriters appropriately for the digital distribution of their compositions, and that Apple wants to create an advertising-supported internet radio service to enhance “recommendations features of the iTunes Store for the purpose of promoting sales of eMasters.”⁵⁸

Apple simultaneously negotiated a license agreement for the public performance rights to the ASCAP repertoire. The ASCAP license covers a [REDACTED], Based on an industry-wide rate of 10%, the parties agreed that Apple would pay AS-CAP a share of [REDACTED] of the iTunes Radio advertising revenue, as that term was defined in their agreement.

The revenue base for the Apple license fee includes none of the subscription revenue from the iTunes Match service. In addition, the revenue base does not include any contribution from the sale of Apple products promoted through iTunes Radio, including the sale of music tracks sold through the iTunes Store. The revenue base does not include any attributed value for that advertising on iTunes Radio that promotes Apple’s music products.⁵⁹ In the license agreement, however, Apple committed that it would “use good faith, commercially reasonable efforts to sell the Advertising to third parties,” and also agreed to attribute revenue at a reasonable rate for the advertising of Apple non-music products that appeared on iTunes Radio.

For several reasons, it would be a difficult task to make the necessary adjustments to the terms of the Apple license to calculate an equivalent rate for an ASCAP license issued to Pandora, and none of the trial experts attempted to do so.⁶⁰ Because iTunes Radio launched only a short time before trial, data about the service is scarce. Moreover, the differences in the revenue bases; the use of iTunes Radio to promote sales of Apple products, which has no equivalent for Pandora; and the absence any means of capturing imputed revenue for the advertising of Apple’s music products, all add to the difficulty of the task.⁶¹

The ASCAP license had other features as well. The parties agreed to a minimum fee amount of [REDACTED] in the event that iTunes Radio failed, or ran primarily Apple advertisements. The agreement contained an advertising deduction of [REDACTED] for external advertising expenses only. Finally, the license provided for a “Most Favored Nations” clause for the benefit of ASCAP.⁶²

Using this same industry-wide rate of 10%, Apple also negotiated direct licenses with publishers Sony, Warner/Chappell, EMI, and BMG. Unlike Pandora’s agreements with music copyright holders, which only included the right to publicly perform works in their repertories, the Apple licenses with the publishers provided Apple with both the right to publicly perform the compositions in the publishers’ repertories, as well as the right to “[e]neode, reproduce, and otherwise use the Publisher Materials solely to the extent reasonably necessary to effectuate, implement and facilitate the foregoing Performances of Publisher Materials.”

CONCLUSIONS OF LAW

Pandora requests that this rate court set a fee for its license with ASCAP. Section IX of AFJ2 requires the rate court to set a “reasonable” fee for a requested license, but that term is not defined in AFJ2. Governing precedent, dictates, however, that in determining the reasonableness of a licensing fee, a court “must attempt to approximate the ‘fair market value’ of a license — what a license applicant would pay in an arm’s length transaction.” MobiTV, Inc., 681 F.3d at 82, “In so doing, the rate-setting court must take into account the fact that AS-CAP, as a monopolist, exercises market-distorting power in negotiations for the use of its music.” Id. The Second Circuit has recognized that, because music performance rights are largely aggregated in the PROs which operate under consent decrees, “there is no competitive market in music rights.” ASCAP v. Showtime/The Movie Channel, 912 F.2d 563, (2d Cir.1990). Consequently, fair market value is a “hypothetical” matter. Id. at 569. In such circumstances, “the appropriate analysis ordinarily seeks to define a rate or range of rates that approximates the rates that would be set in a competitive market.” Id. at 576.

Helpfully, both ASCAP and Pandora have endorsed the same definition of “fair market value,” drawn from a recent textbook:

A widely used description of fair market value is the cash equivalent value at which a willing and unrelated buyer would agree to buy and a willing and unrelated seller would agree to sell ... when neither party is compelled to act, and when both parties have reasonable knowledge of the relevant available information. ... Neither party being compelled to act suggests a time-frame context — that is, the time frame for the parties to identify and negotiate with each other is such that, whatever it happens to be, it does not affect the price at which a transaction would take place.... The definition also indicates the importance of the availability of information — that is, the value is based on an information set that is assumed to contain all relevant and available information.

Robert W. Holthausen & Mark E. Zmijew-ski, Corporate Valuation 4-5 (2014).

In rate court proceedings, a determination of the fair market value “is often facilitated by the use of a benchmark— that is, reasoning by analogy to an agreement reached after arm’s length negotiation between similarly situated parties.” United States v. BMI (In re Application of Music Choice), 316 F.3d 189, 194 (2d Cir.2003) (“Music Choice II”). Rate courts have been provided with guidance in their analysis of the parties’ proposed benchmarks:

In choosing a benchmark and determining how it should be adjusted, a rate court must determine the degree of comparability of the negotiating parties to the parties contending in the rate proceeding, the comparability of the rights in question, and the similarity of the economic circumstances affecting the earlier negotiators and the current litigants, as well as the degree to which the assertedly analogous market under examination reflects an adequate degree of competition to justify reliance on agreements that it has spawned.

United States v. BMI (In re Application of Music Choice), 426 F.3d 91, 95 (2d Cir.2005) (“Music Choice IV”) (citation omitted); accord DMX Inc., 683 F.3d at 45.

“[T]he burden of proof [is] on ASCAP to establish the reasonableness of the fee it seeks.” AFJ2 § IX(B). “Should ASCAP not establish that the fee it requested is reasonable, then the Court shall determine a reasonable fee based upon all the evidence.” AFJ2 § IX(D).

ASCAP and Pandora have each proposed a set of benchmarks for assessing the appropriate rate for an ASCAP license to Pandora. Interestingly, they both agree that the Pandora license with EMI is a valid benchmark. Their sets of proposed benchmarks share no other common element.

As already noted, ASCAP relies principally on the three direct licenses negotiated between Pandora and EMI, Sony, and UMPG in the wake of the April 2011 Compendium modification. ASCAP arrives at proposed rates of 1.85% for 2011-2012 (the Pandora-EMI license rate), 2.50% for 2013, and 3.00% for 2014-2015. This is the first time that ASCAP has sought a license rate of over 1.85% from any non-interactive internet music service.

Pandora recognizes the Pandora-EMI license agreement as a suitable benchmark, as well as the historical ASCAP-Pandora license rate of 1.85% under the 5.0 License. But, in addition to its analysis of appropriate benchmarks, Pandora argues that it is “similarly situated” to the RMLC licensees and is accordingly entitled by the terms of AFJ2 to the RMLC 1.70% rate.

In summary, ASCAP has carried its burden of demonstrating that its rate proposal of 1.85% is reasonable for the years 2011 and 2012. It has failed to carry its burden of demonstrating that its rate proposals of 2.50% and 3.00% for the years 2013 and 2014-2015, respectively, are reasonable. Pandora has failed to show that it is entitled to the 1.70% RMLC rate as the result of being similarly situated, within the meaning of AFJ2, to the RMLC member radio stations.

In conducting an independent inquiry into a reasonable rate, this Court is guided by the following parameters. First, having determined a reasonable rate for the first years of the five-year license period, there is a presumption that that rate will continue to be a reasonable rate for the entire license period. Second, the historical division between interactive and non-interactive internet music services requires that Pandora be licensed well below the 3.0% rate at which ASCAP licenses interactive music services. Third, the circumstances under which Sony imposed upon Pandora an implied ASCAP headline rate of 2.28% confirm that any reasonable rate for an ASCAP-Pandora license is below 2.28% by a measurable margin. For these and the other reasons described below, the 1.85% license rate is the reasonable rate for the entirety of the five year term of the ASCAP-Pandora license.

I. ASCAP’s Rate Proposal of 1.85% for 2011 and 2012

For the years 2011 and 2012, ASCAP proposes a rate of 1.85%. ASCAP’s benchmark for this proposal is the Pandora-EMI license (which is for the years 2012 and 2013), which provided for a headline rate of 1.85%. For confirmation that 1.85% is a reasonable rate, ASCAP relies on the fact that it is the same rate under which Pandora was licensed under the 5.0 License from 2005 to 2010.

Pandora agrees. It admits that a headline rate of 1.85% is within a range of reasonable rates in the event that Pandora is not entitled to the 1.70% rate in the ASCAP-RMLC license. According to Pandora, the 1.85% rate is the “upper bound of a range of reasonable rates for Pandora.” Since AFJ2 only requires AS-CAP to demonstrate that its rate proposal is “reasonable,” Pandora’s concession makes further discussion unnecessary.⁶³

II. ASCAP’s Rate Proposal of 2.50% for 2018 and 3.00% for 201k and 2015

ASCAP proposes a rate of 2.50% for 2013, and 3.00% for 2014 and 2015. AS-CAP predicates these rates principally on the Pandora-Sony license, which covers the year 2013 and yields an industry wide rate of 5.0% and an ASCAP implied rate of 2.28%; and on the Pandora-UMPG license, which covered the six month period from July 1 to December 31, 2013, and yielded an industry wide rate of 7.50% and an implied ASCAP rate of 3.42%. ASCAP also puts forth, as confirmatory benchmarks, the SESAC-Pandora license and Apple’s licenses with the PROs and publishers in connection with its iTunes Radio service.⁶⁴ In addition to these benchmarks there have been other justifications offered at trial for a license rate that exceeds the 1.85% rate dictated by the 5.0 License and the EMI license. ASCAP has offered a two-part theoretical argument in the form of Dr. Murphy’s opinions regarding increasing competition among internet radio providers and the need for variety in music. There were also two other theoretical arguments raised in the record and at trial (although not initially put forward by AS-CAP) for an elevated rate: the potential for cannibalization of music sales by Pandora, and the gap between what Pandora pays record labels for sound recording rights and what it pays the PROs and publishers for composition rights. Finally, Pandora’s success is a factor that has been present, implicitly, throughout the trial.

ASCAP has not carried its burden of showing that its proposed rates for 2013, 2014, and 2015 are reasonable. To begin with, rate court precedent and ASCAP’s own licensing history establish a presumption that a five-year license should have a single rate. ASCAP has not rebutted that presumption. ASCAP has also failed to demonstrate that Pandora’s direct licenses with Sony and UMPG constitute fair market benchmarks. The infirmities in these proposed benchmarks are not overcome by reliance on either the Pandora-SESAC license or the Apple licenses for its iTunes Radio service. Finally, none of ASCAP’s theoretical arguments support an upward departure from the 1.85% rate to the 2.50% and 3.00% rates that ASCAP also seeks.

A. Presumption of a Single Rate

Having accepted ASCAP’s proposal of a rate of 1.85% as a reasonable rate for the first two years of the Pandora license (2011 and 2012), there is a strong basis to recognize a presumption that the rate of 1.85% would also be a reasonable rate for the last three years of the Pandora license (2013 through 2015). Indeed, the two benchmarks that support adoption of a headline rate of 1.85% continued beyond the year 2012. The successor form license to ASCAP’s 5.0 License still has the headline rate of 1.85%, and there was no evidence offered at trial to suggest that AS-CAP is planning to alter that rate for non-interactive new media music services. As for the EMI license, it was for the period 2012 and 2013. Since ASCAP agreed that the EMI rate was reasonable for the year 2012, it is presumptively reasonable for 2013 as well. Indeed, ASCAP agrees that it remains a reasonable benchmark.

Also, adoption of an escalating rate over the term of a five year license would be out of step with historical practice. AS-CAP has never negotiated nor issued a five year license with an escalating rate, and rate court jurisprudence is devoid of any example of an escalating ASCAP rate for a single license term. The sole example in this record of an escalating rate is the SESAC license with Pandora. In that case, however, SESAC’s escalating rate was justified by a mutual assumption that SESAC’s market share would increase over the term of the license. Even accepting that the SESAC license, with its unique features, could be informative about a reasonable rate for an ASCAP license, the justification for an escalating rate for SESAC suggests that the ASCAP rate should be a declining rate since SE-SAC’s growth would come at the expense of ASCAP and BMI.

There appear to be good reasons why ASCAP and the industry generally adopt a single rate for the term of a license. Absent some unusual circumstances, the value of music to a user is assumed to remain constant through the term of a license. And an escalating rate is not necessary for the licensor to share in the success of the licensee: with a single rate as a percentage of revenue a joint interest is created between the parties in the growth of the licensee’s business. Adoption of a single rate facilitates business planning, encourages reliance on historical data, and discourages resort to contested projections. Likely for these reasons, and others, there is a well developed practice that supports the adoption here of a headline rate of 1.85% for not just the first two years, but also for the last three years of the license.

ASCAP has failed to overcome any presumption that exists in favor of a unitary rate. But, even without such a presumption it has not carried its burden to establish that the rates of 2.50% and 3.00% are reasonable.

B. Pandora’s Direct Licenses with Sony and UMPG

ASCAP has not shown that either the Pandora-Sony or the Pandora-UMPG licenses are good benchmarks for its license with Pandora. Sony and UMPG each exercised their considerable market power to extract supra-competitive prices. The UMPG agreement is a particularly flawed benchmark, for the several reasons discussed below. In addition, the evidence at trial revealed troubling coordination between Sony, UMPG, and ASCAP, which implicates a core antitrust concern underlying AF J2 and casts doubt on the proposition that the “market under examination reflects an adequate degree of competition to justify reliance on agreements that it has spawned.” Music Choice IV, 426 F.3d at 95 (citation omitted).

1. ASCAP and Publisher Coordination

Pandora has shown that the Sony and UMPG licenses were the product of, at the very least, coordination between and among these major music publishers and ASCAP. Sony and UMPG justified their withdrawal of new media rights from AS-CAP by promising to create higher benchmarks for a Pandora-ASCAP license and purposefully set out to do just that. They also interfered with the ASCAP-Pandora license negotiations at the end of 2012. UMPG pressured ASCAP to reject the Pandora license ASCAP’s executives had negotiated, and Sony threatened to sue ASCAP if it entered into a license with Pandora. With only a, few business days remaining in the year 2012, ASCAP refused to provide Pandora with the list of Sony works without Sony’s consent, which Sony refused to give. Without that list, Pandora’s options were stark. It could shut down its service, infringe Sony’s rights, or execute an agreement with Sony on Sony’s terms. Then, despite executing a confidentiality agreement with Pandora, Sony made sure that UMPG learned of all of the critical terms of the Sony-Pandora license. And LoFrumento admitted at trial that ASCAP expected to learn the terms of any direct license that any music publisher negotiated with Pandora in much the same way.

There is no need to explore which if any of these actions was wrongful or legitimate. Nor is there any reason to explore here the several justifications that ASCAP, Sony, and UMPG have given for at least some of this conduct.⁶⁵ What is important is that ASCAP, Sony, and UMPG did not act as if they were competitors with each other in their negotiations with Pandora. Because their interests were aligned against Pandora, and they coordinated their activities with respect to Pandora, the very considerable market power that each of them holds individually was magnified. But, since the UMPG and Sony license agreements constitute poor benchmarks even in the absence of coordination, it is not necessary to engage more deeply with the implications of this evidence.

2. The Pandotor-Sony License

The Pandora-Sony license was for the year 2013. It was premised on a 5% industry-wide rate, which implies a 2.28% headline rate for an ASCAP license. AS-CAP has not shown that this rate reflects the fair market value of an ASCAP license with Pandora.

When Pandora inaugurated its service in 2005, it obtained a blanket license from ASCAP and fully expected to continue to be able to do so throughout the life of its business. It was entitled under that AS-CAP license to full access to the ASCAP repertoire and to use any composition in the ASCAP repertoire as frequently as it wished. This included compositions to which Sony held public performance rights. Pandora had no incentive therefore to identify Sony works or to steer its listeners toward or away from those works.

Once the Compendium modification had been adopted, and Sony had withdrawn new media rights from ASCAP effective January 1, 2013, however, the identity of the Sony works suddenly became significant to Pandora.⁶⁶ Because of the nature of its music service, Pandora had more of an ability to substitute one work for another than many other music services. It certainly had more flexibility than an on-demand service, which needed to play virtually any composition its listeners demanded. It even had, at least theoretically, more flexibility than many programmed radio services. For instance, it would be difficult for a terrestrial Top 40 radio station to thrive without access to each week’s top 40 hits. Thus, with a list of the Sony works, Pandora would have information necessary to remove Sony works from its service or steer listeners away from Sony works, or at least to threaten to do so.⁶⁷

Both Pandora and Sony treated knowledge of Sony’s catalogue as a significant bargaining chip in their license negotiations. Pandora repeatedly asked for it, orally and in writing, and Sony pointedly ignored those requests and stopped AS-CAP from providing the list to Pandora. Sony knew that it held the upper hand, as it acknowledged when it conveyed to Pandora that it was not its intention to “shut down” Pandora.

By withholding the list, Sony deprived Pandora of significant leverage in their negotiations.⁶⁸ Pandora was faced with three options: shut down its business, face crippling copyright infringement liability,⁶⁹ or agree to Sony’s terms. Accordingly, the agreement fails the parties’ agreed-upon definition of fair market value: that neither party to the negotiation be “compelled to act.”

Dr. Murphy argues that the Sony license nonetheless can constitute a competitive price because in a competitive marketplace, a copyright owner would not be likely to offer Pandora information that would enable it to operate without the publisher’s works. Dr. Murphy reasons that, given this expectation of common business practice, Sony’s refusal to provide the list of works to Pandora should not be read as constituting undue compulsion such that the Sony 2.28% rate is not a fair market rate. This analysis is too divorced from the world in which Pandora and Sony were actually functioning to be helpful.

First, their marketplace is not the “atomistic” marketplace from which Dr. Murphy’s theoretical framework is, derived. In a competitive, atomistic market, if one of many rights 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.

Second, Pandora had built its business with the understanding that it could obtain a blanket license from ASCAP. It had already, therefore, incorporated the Sony repertoire into its MGP. Unlike a new entrant into a market, it was not free (at least, without a list of the Sony works) to attempt to create a business model that made no or more limited use of Sony music.

For any economic model to be useful here, it must account for the circumstances that created Pandora’s need for Sony’s repertoire information. Even if Sony had provided the list of its works to Pandora, Sony would have retained enormous bargaining power;⁷⁰ 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.

ASCAP next argues that, even if Pandora’s possession of the Sony list of works was necessary to create a valid benchmark, Pandora could have obtained this information from sources other than AS-CAP and Sony. But, Sony did not act in 2012 as if Pandora had a reliable alternative source of information available to it (other than ASCAP), and ASCAP failed at trial to prove that such an alternative existed.

ASCAP also asserts that Pandora did not actually require a list of Sony works to negotiate a fair market rate license with Sony since it had no list of EMI works when it negotiated the EMI license, and the EMI license is endorsed by both parties to this litigation as a suitable benchmark. The negotiations that Pandora conducted with EMI are not, comparable. Pandora did not need a list of EMI works since it learned in its first substantive meeting with EMI that EMI was not seeking an increase in Pandora’s license rate. EMI immediately offered to let Pandora pay for an EMI license at the 1.85% headline rate in the ASCAP 5.0 License.

ASCAP makes two additional arguments in support of its contention that the Pandora-Sony license reflects a fair market value for a Pandora license. First, as evidence that the license was the product of meaningful give and take between the parties, ASCAP contends that Sony made a meaningful concession regarding the advertising deduction when it allowed Pandora to deduct internal advertising costs. There was little evidence offered at trial on negotiations over this term of the license. ASCAP certainly did not show that Sony resisted it or was reluctant to agree to it. After all, EMI had already agreed to such a deduction in the event another major publisher or a PRO accepted it. Given this record, this single modification of a draft agreement in the course of a four-day negotiation period does not alter the conclusion that Pandora was compelled to enter into a license on Sony’s terms.

ASCAP also argues that Pandora’s witnesses have admitted that the effective rate for Pandora of the Sony license, after the deduction of internal advertising costs is taken into account, was only a “modest” increase over the effective rate of the AS-CAP license and was not “out of control.”⁷¹ But even if this were true, it would not cure the primary concern with the Sony license as a benchmark, which is the coercive process by which it was negotiated. In any event, some of Pandora’s characterizations of the Sony rate were made to contrast it with the exorbitant UMPG license rate, and not as an independent assessment of its reasonableness.

In sum, the combination of the looming January 1, 2013 deadline and the lack of information about the Sony catalogue meant that Pandora was compelled to conclude a licensing agreement with Sony at the end of 2012. The presence of such compulsion renders the 2.28% rate a poor benchmark. Since Sony achieved the 2.28% rate in such circumstances, it is reasonable to infer that the fair market value for Pandora’s license is materially lower than 2.28%.

3. The Pandorctr-UMPG License

UMPG and Pandora executed a six-month license for the last half of 2013 that had an industry-wide rate of 7.5%, and an implied ASCAP rate of 3.42%. ASCAP has failed to show that this license is a useful benchmark for an ASCAP license with Pandora.

As already described, there were virtually no meaningful negotiations between Pandora and UMPG because UMPG, controlling roughly 20% of the music market, began with and insisted upon a demand that bore no relation to the then-existing market price. One of Pandora’s principal competitors was covered by the RMLC at a rate of 1.70%; Pandora had been covered under the 5.0 License and had recently executed a license with EMI that encompassed the year 2013 at a 1.85% rate; and Sony had obtained a hike to an implied ASCAP rate of 2.28%. But, UMPG’s 7.5% industry-wide rate implied an ASCAP rate of 3.42%. This was even higher than the ASCAP rate for interactive music services, which was set at 3.00%. If there was one principle regarding rate structure on which the parties agreed at trial it was that the rate for customized radio should be set below the rate for on-demand interactive services.

UMPG’s leap in rate, demanded within a matter of weeks following the Sony negotiations, was so astounding that it drove Pandora to buy a radio station and to file a summary judgment motion challenging the legality of the Compendium modification. Given UMPG’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.⁷² In such circumstances, this license rate cannot be said to represent a bargain arrived at by a willing buyer and seller.

Moreover, UMPG not only demanded an extraordinarily steep increase above the prevailing market rate, but also deprived Pandora, as Sony had, of critical leverage in their negotiations. Although UMPG provided Pandora with a list of its works, UMPG insisted on doing so under the umbrella of an NDA. The NDA accompanying the list would appear to any reasonable reader to prohibit Pandora from using the list to take down UMPG works from its service.⁷³ At the very least, the NDA raised the specter that UMPG would sue Pandora if it used the list to do so. For this additional reason, the UMPG license rate is not a useful benchmark.⁷⁴

C. ASCAP’s Secondary Benchmarks: the SESAC and Apple Licenses

ASCAP has offered the Pandora license with SESAC and the Apple iTunes Radio licenses as confirmatory of the Sony and UMPG license rates. Without the Sony and UMPG license agreements as benchmarks, however, these confirmatory benchmarks lose their utility. Nonetheless, since the parties devoted attention to these licenses, they will be discussed. In short, there is insufficient data about the SESAC repertoire and the Apple iTunes Radio business model to make the adjustments required to support an increase of rate above 1.85%.

1. The Pandora-SESAC license

The Pandora license with SESAC had an escalating rate that runs from an implied rate for ASCAP of [REDACTED] to [REDACTED], assuming that SESAC has a 10% market share and ASCAP has a 45.6% market share.⁷⁵ Of course, even at its upper range, the SESAC license rate is lower than the implied Sony rate of 2.28% and implied UMPG rate of 3.42% for AS-CAP licenses to Pandora. There are several reasons, however, that the SESAC license terms provide minimal guidance here.

The SESAC license has historically been a benchmark of limited value because the public knows little about the size of the SESAC repertoire. See MobiTV, 712 F.Supp.2d at 254. As such, it is difficult to adjust a SESAC license rate to arrive with confidence at an implied ASCAP rate. This problem is exacerbated by the fact that SESAC’s small size, when compared to ASCAP and BMI, not only amplifies any error in a projection, but also reduces the incentive to resist SESAC’s rate requests. While the cost associated with resistance may not be justified when a license fee is relatively small, the willingness to incur those costs will necessarily grow with the size of the anticipated payments.

Second, Pandora’s contemporaneous understanding of the SESAC repertoire and the rate undercut any suggestion that it supports an ASCAP rate above 1.85%. SE-SAC argued that an escalating rate in the SESAC license was appropriate to account for SESAC’s anticipated growth in market share.⁷⁶ There was no evidence presented at trial to suggest that the ASCAP market share is growing. Indeed, to the extent the SESAC share is growing, ASCAP’s share would be presumed to be declining. Thus, if this principle were applied evenhandedly, for the latter years of the AS-CAP-Pandora license, the ASCAP rate might move below 1.85%.

In addition, Pandora believed that the initial SESAC rate reflected an approximate SESAC market share of [REDACTED], despite SESAC’s representation that its market share figure was 10%. The escalating rate was intended to recapture the initial underpayment over time.

Therefore, for each of these reasons, the SESAC license with Pandora is of limited utility in assessing the appropriate rate for a Pandora-ASCAP license. If it were to be used at all, it does not suggest a rate above 1.85% for an ASCAP license.

2. The Apple Licenses

As it was announcing the inauguration of iTunes Radio, Apple entered into a set of licenses with ASCAP and music publishers premised on an industry-wide rate of 10%. This implies, an ASCAP rate of [REDACTED]. This rate is substantially in excess of the 1.85% rate of the Pandora-EMI license, as well as the implied rates of 2.28% and 3.42% for the Sony and UMPG licenses with Pandora, respectively, There are at least two reasons why the Apple licenses for its iTunes Radio service provide little guidance for an ASCAP-Pan-dora license.

First, iTunes Radio is a service offered by Apple to complement its iTunes Store and iTunes Match. Through these latter services, Apple sells digital downloads and operates a locker system so that subscribers may access their music from any of their Apple devices. The integration of these services, seamlessly, within the Apple ecosystem generates synergies. As a consequence, Apple conducted negotiations for its licenses for the public performance of compositions within the context of a business model that has no analogue for Pandora.

Second, the Apple revenue base for its licenses has several exclusions that may be important. Although Apple advertises its music offerings over iTunes Radio, none of that imputed advertising revenue is captured in the revenue base. The revenue base also excludes any contribution from the iTunes Match subscription fees. None of the revenue from the sales of downloads, purchases that can be made by clicking on a buy button while listening to iTunes Radio, is captured. And the list goes on. And because iTunes Radio launched only a short time before trial, data about the service is scarce and no one was in a position to undertake the exceedingly difficult task of making adjustments to the terms of the Apple license to calculate an equivalent rate for an ASCAP license issued to Pandora. Consequently, the Apple licenses do not provide a basis to find with any confidence that a rate above 1.85% is a fair market rate for an ASCAP license issued to Pandora.

D. ASCAP’s Theoretical Arguments and Motivations

In addition to offering benchmarks, either ASCAP or its witnesses presented five arguments in support of either a higher rate for a Pandora license than it had historically paid, or an escalating rate within a single Pandora license. These theoretical arguments seek to justify AS-CAP’s request for an otherwise hard-to-explain sharp rate increase from 1.85% in 2011 and 2012 to 2.50% and 3.00% in the years between 2013 and 2015. First, Dr. Murphy identified increased competition among internet music users, and listeners’ preference for variety in music, to support ASCAP’s license request. Second, UMPG and Sony both justified their demands for a higher rate from Pandora because of the extreme gap between the size of payments made by Pandora for rights to the public performance of compositions and sound recordings. Third, there is theoretical and historical support for imposing a higher rate on a music service that cannibalizes the sale of music, and this justification arose at trial. Fourth, ASCAP has emphasized a purported difference in the intensity of music use between internet music services like Pandora and terrestrial radio services. Finally, although unstated, the publishers and ASCAP appear to believe that Pandora’s license rate should be increased because Pandora is currently a successful internet radio service. Each of these reasons for a hike in rates will be discussed in turn.

1. An Increase in Competition

Dr. Murphy posits, on behalf of ASCAP, that an increase in the demand for public performance rights in musical works on the internet would lead to an increase in market prices in a competitive market. Dr. Murphy and ASCAP list a number of recent customized radio services which have emerged in support of the relevance of this theory here. They tender this observation to support an increase in the Pandora licensing rate from 1.85% in 2012 to 3.00% by 2015. This theory of economic behavior in a competitive market is so untethered to actual music industry market conditions and historical evidence that it provides minimal assistance when the task at hand is to set the rate for this five year Pandora-ASCAP license.

First, Dr. Murphy does not grapple with the history of music on the internet and the licensing rates for that music, and the implications of that history for his theory. There has been a sizable and growing internet radio industry since the mid-1990s. At first, these were simulcast stations. Internet-only radio arrived shortly thereafter. Customized radio entered the arena in the late 1990s.⁷⁷ There has been increasing competition in the radio and internet music spheres for a long time without corresponding rises in licensing rates attributable to increased competition. AS-CAP adopted its 5.0 License for internet music in 2004. Since that year, right through until today, ASCAP has utilized a single rate — 1.85%—for the most music-intensive internet services. ASCAP licensed Pandora itself under the 1.85% rate for the entire period of 2005-2010, despite the purported increase in competition within the field of internet radio, including customized radio, over that period. EMI chose to adopt the 1.85% rate for Pandora for the years 2012 and 2013. As recently as 2012, ASCAP accepted a blended rate of 1.70% for terrestrial and internet radio.⁷⁸ This stability in rates over a decade in the internet music market is unexplained by Dr. Murphy’s observation, and indeed runs contrary to it. In particular, ASCAP has made no showing that competition within the internet music market increased so dramatically within the single year in which the Sony and UMPG licenses were negotiated, that an increase in rate from 1.85% (which ASCAP agrees is the correct rate for the year 2012) to 3.42% (the ASCAP rate implied by the UMPG license in 2013) was reasonable.

In addition, there are theoretical gaps in Dr. Murphy’s theory. For one, the laws of supply and demand teach us that the price of a commodity will increase as demand increases, but only to the extent that supply is held constant. Dr. Murphy did not attempt to address whether a growth in music supply may also have contributed to a stability in rates over many years, and whether any tendency to raise rates within a competitive market is tempered by the expectation that supply would increase in such a circumstance. Moreover, Dr. Murphy’s theory is also difficult to apply to a market in which the price of music is expressed as a percentage of revenue. In that market, an increase in demand does not necessarily result in an increase in rate. The rights holder participates in the growth of revenue by application of a stable rate to an expanding revenue base. In light of all of the above, Dr. Murphy’s competition theory does not persuasively suggest that the rate for Pandora’s license should rise above 1.85% in the final three years of its license with ASCAP, nor that the Sony or UMPG license rates are necessarily fair market rates.

2. Demand for Variety

Dr. Murphy offers a second theoretical argument in support of ASCAP’s requested fee structure. He contends that, all else being equal, “listeners prefer more variety to less.” From this observation, he concludes that the demand for variety increases the competitive market price of rights to publicly perform musical works and justifies an increased rate for a Pandora license.

As was true with Dr. Murphy’s first theoretical assumption, this theory comes undone when applied to the real world. Dr. Murphy’s claim that listeners prefer variety above all is unsupported and cannot form the basis for an upward departure from a rate of 1.85%. Dr. Murphy did not conduct any research or analysis into consumer listening behavior to arrive at his conclusion that listeners prefer variety above all. And it is likely that once a certain minimal variety threshold is reached listeners don’t actually prioritize extra variety. The record evidence suggests that, as a general matter, listeners are not so eclectic in their tastes that the addition of a song to a music service will necessarily provide added value. Listeners often like to hear music that they already know that they like, or music very similar to music they already like.

Confronted with this fact, Dr. Murphy emphasized instead the need a music service has for variety so that it can satisfy the many distinct individual tastes of its listeners. But, even if the price of a license were driven by the extent to which a music service requires a large library of compositions, ASCAP’s license fee request would not be supported. First, there was no change in the nature of the Pandora service between the years 2012 and 2013, when measured by an exhibited demand for variety in its repertoire, to support the shift in rate from 1.85% to 2.50% and higher, as proposed by ASCAP. This is especially notable in light of the fact that Pandora, as a blanket licensee, faced no marginal cost from adding songs from the ASCAP repertoire to its MGP and presumably would add every song in the AS-CAP repertoire to its service if demand for variety were a primary driver. Instead, its business model has allowed it to use just a fraction of the repertoire.

As significantly, even if Pandora were shown to demand variety above all, the “variety” metric certainly does not support a fee for Pandora’s service that is as high as the 3.0% fee ASCAP charges on-demand music users in its form license. And it is even unlikely to justify a higher rate than the fee ASCAP uses for programmed internet music.

By a large order of magnitude, customized music services like Pandora have lower demonstrated demand for an extensive library of music than do on-demand services. Although Pandora has access to the full repertoire of each of the PROs, its MGP includes only about one million compositions.⁷⁹ In contrast, on-demand services need to be able to play roughly twenty times as many compositions. Spotify has somewhere in the neighborhood of 20 million tracks. After all, on-demand services essentially promise their members that they can listen to any song they want anytime they want. Consequently, even if one accepted Dr. Murphy’s theory of variety, ASCAP’s rate would have to be set well below the 3.0% rate applicable to on-demand services under ASCAP’s form license.

Even when it comes to programmed radio stations, ASCAP did not show that those stations could succeed with fewer licenses or smaller libraries than Pandora. Take just one example: a Top 40 station. To be successful, a Top 40 station would need licenses from every holder of rights to Top 40 songs. This would in all likelihood necessitate deals with all three PROs as well as any major publishers with works outside of the PROs. Pandora, in contrast, has the ability to substitute songs.⁸⁰ Its MGP enables it to play songs with characteristics its listeners enjoy rather than each popular song in a category. As a theoretical matter, this flexibility in programming gives Pandora more flexibility in licensing negotiations.⁸¹ If the demand for variety were a driving factor in licensing negotiations, therefore, Pandora’s license rate should probably not be set higher than broadcast radio’s 1.70%, and should perhaps be lower.

Dr. Murphy’s theory also has very little utility in the world in which ASCAP and Pandora operate. It does not help to distinguish among types of music services in a concentrated industry, and where blanket licenses exist. To be successful, every music service needs a license with every PRO. And where their works are not covered by PRO licenses, every music service will ultimately need a license with both Sony and UMPG, which together control about half of the U.S. market. Thus, it is difficult to see how demand for variety could assist in distinguishing between rates for different types of music services when all types of music services must make the same licensing deals to survive. Moreover, with a blanket license a music user receives the right to every work in the repertoire, whether it avails itself of the opportunity to play all the works or not. As a result a music service with a high demand for variety can satisfy that demand by playing more of. the songs covered by its blanket license.

In sum, Dr. Murphy’s contention that a demand for variety in music can explain the upward adjustments in the Pandora-ASCAP license rates must be rejected. It does not fit the world in which Pandora operates or the Pandora business model. If anything, resort to this theory undercuts ASCAP’s request for an increase in a license rate for Pandora.

It is worth observing that there is no evidence in the record that any of the licensing negotiations in this industry have been driven by either of the rationales proffered by Dr. Murphy. There was no evidence that ASCAP, EMI, Sony, UMPG, or any other licensor negotiated with any music user on the ground that their service required a larger catalogue or that there had been a recent surge in competition in the market.⁸² But, there was ample evidence of the actual driving force behind the Sony and UMPG withdrawal of new media rights from ASCAP and their negotiations with Pandora. That driving force was the music publishers’ envy at the rate their sound recording brethren had extracted from Pandora through proceedings before another rate setting body, the CRB.

ASCAP has not offered any theoretical support for raising the rate for public performance of a composition by a comparison to the rate set for sound recording rights. There may be several reasons for this, but first and foremost is the statutory prohibition on considering sound recording rates in setting a rate for a license for public performance of a musical work. See 17 U.S.C. § 114(i) (“License fees payable for the public performance of sound recordings ... shall not be taken into account in any ... proceeding to set or adjust the royalties payable to copyright owners of musical works for the public performance of their works.”). Thus, this Court may not take the rates set by the CRB into account in determining the fair market rate for a public performance license from ASCAP to Pandora.

Despite this statutory prohibition, one observation may be safely made. Unhappiness about the gap between what Pandora pays record companies and what it pays the PROs drove the modification to the ASCAP Compendium, the publishers’ withdrawals from ASCAP, and the Sony and UMPG negotiations with Pandora. The corporate rivalries over digital age revenues explain a great deal of this history. In any event, the record is devoid of any principled explanation given by either Sony or UMPG to Pandora why the rate for sound recording rights should dictate any change in the rate for composition rights.

4. Cannibalization of Music Sales

There is agreement between the parties that it is appropriate to require a higher licensing fee from a music service that acts as a substitute for the sale of a musical work, when compared to one that does not. To the extent that a music service is a replacement for sales, it is said to cannibalize the sales; to the extent it encourages sales, it is said to be promotional.

The reasons for this distinction arise, at least in part, from a separate stream of rights belonging to composers. Composers have a copyright interest in the reproduction and distribution of musical works, an interest that is referred to as “mechanical rights.” 17 U.S.C. §§ 106(1) & (3); 17 U.S.C. § 115. The licensing regime for mechanical rights is complex, and its details need not be described here. It is sufficient to observe that when hard copies (e.g., vinyl records, CDs) or digital downloads of compositions are sold, the composers receive mechanical rights payments. On-demand services, as well, are required to pay mechanical rights. As described earlier, Spotify pays a 10.5% fee for both mechanical rights and the right to publicly perform a musical work.

The parties have argued about the extent to which Pandora and services like it are promotional or cannibalistic. There is apparently no industry consensus on this question. It is worth noting, however, that what evidence was presented at trial suggests that Pandora is promotional.

To begin with, radio has traditionally been considered promotional. The record industry has long sought to have its music played on radio stations.⁸³ Pandora is no exception. Record labels have taken advantage of Pandora Premieres to feature new work in advance of release, with the hope that that exposure will engender sales. Pandora itself has buy buttons that permit listeners to buy digital downloads from Amazon and Apple, and they use them with some frequency.⁸⁴ There is no evidence that artists have taken steps to prevent Pandora from playing the artist’s work. As significantly, one of Pandora’s principal competitors — iTunes Radio — was created to complement Apple’s iTunes Store and promote sales in that digital store.⁸⁵

In contrast, on-demand streaming services like Spotify are widely considered cannibalistic and are licensed at a higher rate accordingly. After all, a listener has no need to purchase a digital download when the listener has any song that she wants to hear instantaneously available through Spotify. For this very reason, some prominent performers have acted to prevent Spotify from playing their recordings. In sum, while this metric — whether a service is promotional or cannibalistic— could justify a differentiation of rates between services, ASCAP failed to show that Pandora is anything other than promotional of sales.⁸⁶

5. Music Intensity

ASCAP argued at trial that Pandora’s licensing fee should exceed the RMLC rate of 1.70% because its channels use music more intensively than terrestrial radio stations. Music intensive broadcast stations play, on average, 11 songs ■ per hour; Pandora’s stations play 15 or so. This difference is attributable at least in part to the difficulty of placing advertising on internet radio, which is a challenge that Pandora is addressing through its substantial investment in an in-house advertising department. In any event, ASCAP has not shown -that this current differential justifies any increase in the last three years of the Pandora license above the 1.85% rate it has requested for the first two years.

First, assuming that the purported music intensity differential justifies an upward departure for Pandora’s rate, AS-CAP has already created that differential. Its 5.0 License rate for internet music services like Pandora is 1.85% whereas its RMLC license rate for Music Format stations is 1.70%. And the form license rate for new media services escalated from the prior ASCAP form license rate of 1.615% to reflect “the maturation of the new media marketplace and ASCAP’s observation that many of these services were using a large amount of ASCAP [mjusic.”

Second, it is not clear, in all events, that any rate differential is justified on the basis of music use. As described above, the habits of terrestrial radio listeners and the presence of music in advertising may make any differential largely illusory. Moreover, ASCAP does not have a history of fine-tuning its rates in the way suggested here. The RMLC license rate for music intensive services covers a far broader range of music usage than the differential between 11 and 15 songs per hour.

Finally, ASCAP agrees that Pandora’s rate for 2011 and 2012 should be 1.85%. And it has presented no evidence that the music intensity of Pandora’s services will change in any material way for the last three years of the license term. For this reason, as well, the music intensity metric cannot provide a basis to justify the hike in rates to 2.50% and 3.00% that ASCAP seeks.

6. Pandora’s Success

There is one final motivation for AS-CAP’s requested rates that must be acknowledged. The backdrop for this rate court dispute is the arrival of the digital age in the music industry, the resulting disruption to the business models of the music industry, and Pandora’s current success in the digital radio market.

A rights holder is, of course, entitled to a fee that reflects the fair value of its contribution to a commercial enterprise. It is not entitled, however, to an increased fee simply because an enterprise has found success through its adoption of an innovative business model, its investment in technology, or its creative use of other resources. It appears that Sony, UMPG, and ASCAP (largely because of the pressure exerted on ASCAP by Sony and UMPG) have targeted Pandora at least in part because its commercial success has made it an appealing target.

Pandora has shown that its considerable success in bringing radio to the internet is attributable not just to the music it plays (which is available as well to all of its competitors), but also to its creation of the MGP and its considerable investment in the development and maintenance of that innovation. These investments by Pandora, which make it less dependent on the purchase of any individual work of music than at least some of its competitors, do not entitle ASCAP to any increase in the rate it charges for the public performance of music. To the extent Pandora prospers because of its innovations and because of its separate investment in an initiative to develop advertising revenue, ASCAP and its members will prosper through the increased revenue stream that is generated by the application of an appropriate rate to Pandora’s revenue base.

Moreover, market share or revenue metrics are poor foundations on which to construct a reasonable fee. Internet radio remains in its infancy. There is little likelihood that the landscape of today will remain unaltered. Indeed, remarkable changes occur with lightning speed in the digital age.

As of today, ASCAP has two sets of rates for internet radio: those in the RMLC license (which are available to owners of broadcast radio stations) and those in its form licenses. Neither set of rates has a separate schedule for customized radio. ASCAP may or may not wish to explore the creation of a separate rate structure for customized radio services like Pandora. It is unnecessary to reach the issue of whether such a separate rate structure would be justified, much less what the spread in rates should be between programmed radio and customized internet radio. Suffice it to say that AS-CAP has not shown that Pandora’s particular success in expanding its . audience and revenue justifies in any way an increase in rate from the 1.85% rate which ASCAP seeks for the years 2011 and 2012 to the higher rates ASCAP seeks in the three succeeding years.

III. Whether Pandora is Entitled to the RMLC 1.70% Rate

Before concluding that the rate for an ASCAP license to Pandora for the five years from 2011 through 2015 should be 1.85%, it is necessary to address Pandora’s contention that it is similarly situated to the RMLC licensees and entitled to the RMLC license rate of 1.70% under the anti-discrimination provisions of AFJ2. See AFJ2 §§ IV(C), IX(G). Pandora has not shown that a ruling in this Opinion that requires ASCAP to license Pandora at a rate of 1.85% from 2011 to 2015 will violate the anti-discrimination provisions of AFJ2.

There are several provisions in AFJ2 that concern similarly situated licensees and impose upon ASCAP the duty to treat them in a non-discriminatory manner. AFJ2 § IV(C) enjoins and restrains AS-CAP from “[ejntering into, recognizing, enforcing or claiming any rights under any license for rights of public performance which discriminates in license fees or other terms and conditions between licensees similarly situated.” Section IX(G) of AFJ2 further provides that “[wjhen a reasonable fee has been determined by the Court,” ASCAP must “offer a license at a comparable fee to all other similarly situated music users who shall thereafter request a license of ASCAP.”⁸⁷

AFJ2 defines “similarly situated” licensees as “music users or licensees in the same industry that perform ASCAP music and that operate similar businesses and use music in similar ways and with similar frequency.” AFJ2 § II(R). It lists the factors that are relevant to a determination of whether licensees are similarly situated as including, but not limited to, “the nature and frequency of musical performances, ASCAP’s cost of administering licenses, whether the music users or licensees compete with one another, and the amount and source of the music users’ revenue.” Id.

The licensees to which Pandora seeks comparison are those covered by the RMLC license. The current RMLC license was approved in early 2012 and runs through the year 2016. As described in more detail above, it allows RMLC members to pay at a rate of 1.70% of the revenue derived from radio stations that principally play music. The RMLC members own commercial radio stations. With the advent of the internet, many of the RMLC members have simulcast their programming for their terrestrial stations on the internet. Some have also created internet-only radio stations. Of the top 20 internet radio services, as measured by the Triton Digital firm, 16 of the 19 services that are not Pandora are owned by RMLC entities. Clear Channel’s internet radio offering, known as iHeartRadio, includes a customized internet radio service called Create Station. Create Station began operation in 2011 and was initially run as a commercial free service. By late 2012, it was running with advertising and generating revenue. The Create Station feature is a direct competitor with Pandora and is a growing component of the iHeartRadio offering.

Although Pandora contends that it is similarly situated to all RMLC licensees, it emphasizes its similarity to Clear Channel’s iHeartRadio generally, and more specifically to customized radio offerings by RMLC members Clear Channel and CBS, the Create Station and Last.fm services, respectively.⁸⁸ Pandora has shown that its service is indistinguishable for licensing purposes from these components of Clear Channel and CBS.

More generally, Pandora has shown that it is radio and competes with programmed radio, including terrestrial radio, for listeners and advertising dollars. Its most direct competitors within the radio industry are other internet radio services, especially customized radio services. Although terrestrial radio stations generally play about four fewer songs per hour than Pandora, Pandora has shown that this difference is not material given the broad categories of music use in the ASCAP-RMLC license. In any event, ASCAP has not offered any evidence to suggest that the internet radio stations of RMLC members use music less intensively than Pandora. Indeed, for that period of time in which Clear Channel ran its Create Station feature as an ad-free service, it was a more music intensive service than Pandora. And ASCAP has not suggested that there is any distinction among any of these services in terms of ASCAP’s cost of administering their licenses.

Of the factors which AFJ2 lists as pertinent to an analysis of similarity, the only factor which may provide a basis for distinguishing between Pandora and other customized internet radio services run by RMLC members is the amount of revenue. But, while it appears that Pandora is earning significantly more revenue than the RMLC customized radio services, all of the internet music services at issue here are run as commercial stations.

ASCAP emphasizes the degree to which Pandora markets itself as an improvement on traditional radio. It is true that the digital delivery of music has permitted the creation of customized radio stations that are unique to individual listeners. But, despite that development, customized radio retains the essential characteristics of radio. The radio service programs and delivers the music and the listener does not know which song will be played next. For purposes of analyzing the non-discrimination prohibitions of AFJ2, of course, what matters is the degree to which the RMLC licensees are similarly situated to Pandora, and it is uncontested that RMLC licensees provide both programmed and customized internet radio services.

In light of these similarities, the question that is fairly presented by Pandora’s application is whether it is entitled by AFJ2 to the RMLC rate.⁸⁹ The answer to that question, while close, is no. Pandora is not entitled to the 1.70% RMLC rate for at least three reasons.

First, the RMLC rate applies to a large-scale license agreement that binds a variety of licensees in both the terrestrial and the internet radio sphere. Moreover, the revenues from terrestrial radio swamp those from the internet services. Second, while Pandora’s service is, for the purpose of this analysis, identical to services offered by some RMLC members, AFJ2 forbids discrimination among licensees, and Pandora has not shown that it is similarly situated to any RMLC licensee. Pandora relies heavily on comparison with Clear Channel’s iHeartRadio’s customized Create Station feature. But Clear Channel is the licensee, and the Create Station feature constitutes a very small part of Clear Channel’s business at present. Third, Pandora is as similarly situated to internet music services covered by the 5.0 License at the rate of 1.85%. Since this Opinion sets the rate for the Pandora license at 1.85%, it is difficult, if not impossible, to find that there is any violation of AFJ2 due to a discrimination in rates.

What this discussion may underscore is a lack of coherence in the present rate structure of ASCAP licenses. This is understandable given the evolving nature of the radio market. Any change in rate structure (for instance, to create a rate structure for customized music services) would have to be made with care based on a thorough understanding of the market and the uses of music in the market, informed by a desire not to discriminate among similarly situated licensees or between similar services simply because of a difference in the mode of distribution.⁹⁰ After all, if there is no commercially legitimate reason for a distinction in rates, then the distinction would not survive in a competitive market.

IV. Publisher Concerns Regarding the Consent Decree and the Rate Court

There is one remaining issue to address. ASCAP, Sony, and UMPG witnesses expressed frustration with the Consent Decree and the rate court process, both in their communications with each other and in their trial testimony. LoFrumento explained that this frustration arrived with the digital age and reflects a fear that the record industry will grab all of the available revenue from the digital transmission of music. According to ASCAP, AFJ2 and its processes, in particular the requirement that ASCAP issue a license to any applicant, hamper ASCAP’s ability to negotiate a fair market rate. Sony and UMPG witnesses asserted that they had to withdraw their licensing rights from ASCAP in order to negotiate effectively with Pandora and achieve appropriate parity with sound recording licensing rates. They expressed skepticism that the rate court proceedings could determine a fair market value for a Pandora license.

The Court is sensitive to ASCAP’s concerns and understands that the unique characteristics of the market for music licensing and the Consent Decree regime produce challenges for all parties. But, for the reasons already discussed, ASCAP did not show that the upshot of the negotiations conducted by either Sony or UMPG with Pandora was a competitive, fair market rate.

CONCLUSION

The headline rate for the ASCAP-Pan-dora license for the years 2011 through 2015 is set at 1.85% of revenue for every year of the license term. Pandora is entitled to take a deduction for any direct payments to publishers made following their partial withdrawals from ASCAP.

SO ORDERED.

1 . New media is defined below, but the term refers generally to internet transmissions.

2 . ASCAP initially submitted testimony from a fourth expert, Timothy Hanlon, on the issue of a proper advertising costs deduction, but the parties settled this issue before trial.

3 . The witnesses who had excerpts of their depositions offered by ASCAP were Tom Conrad, Simon Fleming-Wood, Michael Herring, John Kennedy, John Trimble, and Timothy Westergren.

4 . The witnesses who had excerpts of their depositions offered by Pandora were Orley Ashenfelter, Martin Bandier, Peter Boyle, Peter Brodsky, Vincent Candilora, Matthew De-Filippis, Roger Faxon, Wayland Holyfield, John LoFrumento, Seth Saltzman, Raymond Schwind, and Paul Williams.

5 . Those were the deposition designations for Richard Conlon, J.D. Connell, Zach Horowitz, Robert Rosenbloum, and William Velez.

6 . A music publisher is an entity which coordinates licensing and other logistics pertaining to copyrighted compositions. Music publishers are distinguished from record labels, which coordinate licensing of the sound recordings of performances of copyrighted compositions.

7 . There was reference at trial to ongoing antitrust litigation concerning SESAC.

8 . For background discussion of AFJ2, see generally Meredith Corp. v. SESAC LLC, 09 Civ. 9177(PAE), 1 F.Supp.3d 180, 196-98, 2014 WL 812795, at *11-12 (S.D.N.Y. Mar. 3, 2014).

9 . Satellite radio permits coast-to-coast nationwide programming. It is primarily directed at the automotive market.

10 . High Definition radio — a digital radio technology which piggybacks on existing AM/FM signals but cannot be received by traditional radios — was approved by the FCC in 2002.

11 . With the digital age, the music world has transitioned from one in which music must be purchased in physical form, whether a vinyl LP or CD, to a digital world in which digital downloads and digital music streaming are major forces. Apple’s iTunes Store was launched in 2003 and established a mainstream market for the purchase of digital music files.

12 .These services have been acquired, respectively, by MTV, Yahoo! and by CBS.

13 . Spotify Radio and iHeartRadio's “Create Station’’ use personalization technology from a company called The Echo Nest. CBS’s Last.fm uses technology called "scrobbling.’’ Although not part of the trial record, it has been recently reported that Spotify has acquired The Echo Nest.

14 . Despite the introduction of a customized radio feature in late 2011, Spotify remains an overwhelmingly on-demand service. Under Spotify’s licensing agreement with ASCAP, which covers the period of [REDACTED], Spotify pays between [REDACTED] and [REDACTED] of its revenue, depending on certain conditions not otherwise relevant here. This licensing fee has marginal relevance to this trial, however, since that fee is a component of a federal regulatory license rate of 10.5%, which Spotify must pay to obtain both public performance rights and mechanical rights under 17 U.S.C. §■ 115. See 37 C.F.R. § 385.12(b)(2). Any fluctuation in the public performance licensing fee has no impact on the overall 10.5% rate which a service like Spotify is required to pay.

15 .The RMLC allocated the fees among individual radio stations according to a formula that it developed.

16 . The category "Music Format Station” is broad. A station is defined as a Music Format Station if it has a featured performance of ASCAP music in more than 90 of its "Weighted Program Periods” in a given week. A Weighted Program Period is of 15 minutes duration, and there are 318 Weighted Program Periods in a week. So, a station is a Music Format Station whether it plays ASCAP music anywhere between approximately 28% to 100% of the Weighted Program Periods. A station that does not fall within the Music Format Station definition because it uses AS-CAP music less frequently pays at a rate starting at .0296% of revenue, plus a supplemental fee.

17 . The agreement provided for a deduction of 12% for gross revenues from "Radio Broadcasting” and a 25% deduction for gross revenues from "New Media Transmissions," subject to a cap.

18 . Pandora provides streaming internet music services in the United States, New Zea-land, and Australia.

19 . The use of human beings to classify each composition is unique to Pandora and has the advantage of ameliorating what the industry recognizes as the “cold start” problem. A cold start problem exists when the recommendation system cannot draw on adequate inferences for a composition because the item is new or obscure, Because of this limitation, the service may play the composition for listeners who do not like it or fail to play it for those who might.

20 . While there is general agreement about these numbers, they may not be an altogether accurate description of the difference in a listener's experience of music. Listeners to broadcast radio stations frequently switch stations in search of more music when an ad begins. Moreover, even ads contain music.

21 . The music industry's use of the term "interactive” is explained below.

22 .Pandora points out a cost advantage that broadcast and satellite radio have over Pandora. Broadcast radio pays no royalties for terrestrial broadcasts of sound recordings; satellite radio pays 9% of revenue for satellite transmission of sound recordings; and Pandora paid 55.9% of its revenue for internet transmission of sound recordings in 2012.

23 . ASCAP created a “New Media” department in 1995 to address the licensing of music over the internet. Although the meaning of the term new media may depend on the context, ASCAP generally considers new media to include any music user that operates "primarily over the Internet, through wireless devices, or through other emerging digital technologies.”

24 . The Rhapsody on-demand service was inaugurated in 2001.

25 . The 5.0 License defined non-interactive services as "site[s] ... from which 'Users' may not download or otherwise select particular musical compositions, unless such compositions are sixty (60) seconds or less in duration.”

26 . The 1.615% rate that ASCAP had applied to internet radio before its adoption of the 5.0 License was ASCAP's rate for a terrestrial radio license in the era before the RMLC license adopted a flat rate.

27 . Interactive services are defined as those which "transmit[] and/or provided access to transmissions of content comprising or containing music to ‘Users’ at their request or direction.”

28 . For example, if a customer used a licensed service once for forty minutes and once for fifteen minutes, that was measured as two sessions. Similarly, if a customer used the service for 61 minutes, that was counted as two sessions.

29 . As a consequence, Pandora and ASCAP agreed that Pandora could use a sample of usage to arrive at a reasonable estimate of the amount it owed.

30 . Publishers lobbied for this provision in Congress because they were concerned that the sound recording rates would be set below the public performance rates for compositions and drag down the latter. ASCAP also supported the enactment of the provision, for the same reason.

31 . The references in these two communications to SoundExchange and the CRB are references to the entities that distribute fees from and set the rate for sound recording rights licenses.

32 . The [REDACTED] and [REDACTED] emails addressed as well their concern with the direct licensing activities of Sony, exemplified by its license with DMX. See In re THP Capstar Acquisition Corp., 756 F.Supp.2d 516, 552 (S.D.N.Y.2010), aff'd sub nom. BMI v. DMX Inc., 683 F.3d 32 (2d Cir.2012). .

33 . The ASCAP submission to the DOJ focused on three issues which ASCAP believed might require amendments to AFJ2. The third of the issues it mentioned was the proposal to amend the Compendium to allow a publisher (the submission focused on EMI) to "reserve exclusively to itself the right to license particular on-line users.” In an oblique reference to the higher rates for a license for the public performance of a sound recording, the submission disclosed that EMI had concluded that "the consent decree is not giving it adequate value for its repertory, especially as compared to revenues they derived from other, similar rights.” The first two issues,' to which it devoted a far lengthier discussion, were problems created by the rise of "carve out” licenses, and other challenges associated with online licensing.

34 . With respect to the timing of a publisher's withdrawal of rights, the modified Compendium provided that:

A [new media withdrawal] ... will be effective on the first day following the last day of the calendar quarter in which the anniversary date of the Member’s election falls (the "Effective Date”), upon submission of an executed copy of such [withdrawal] to AS-CAP no more than nine months nor less than six months from the calendar quarter in which the anniversary date falls.

35 . At trial, the publishers agreed that the Compendium modification does not apply to digital downloads. The publishers and AS-CAP do not appear to have a position yet on whether it applies to satellite radio.

36 . By the Fall of 2012, there was increasing dissent within ASCAP about the wisdom of the 2011 Compendium modification. At a September Board meeting, the writer members of the Board urged that ASCAP reverse the modification and reject Sony’s pending request that the option to withdraw new media rights be further modified to allow a publisher to target its withdrawal and limit it to the rights needed by large music users only. The writers did not want ASCAP to assist Sony as it weakened ASCAP. Much of this anger was engendered by Sony’s direct licensing program generally, including the licensing of DMX.

In June of 2012, the Court of Appeals for the Second Circuit had affirmed rate court decisions that set rates for ASCAP and BMI licenses with DMX that were comparable to DMX’s direct license rates, with certain adjustments. DMX Inc., 683 F.3d at 47. The decision also required ASCAP to provide a blanket license that was subject to carve-outs to account for an applicant’s direct licensing program. Id. at 44. DMX is a commercial music service provider that supplies background and foreground music to public venues such as restaurants, frequently through the transmission of music via satellite transmission. To succeed in its direct licensing campaign with music composers and publishers, DMX decided that it was necessary to sign at least one major music publisher. In 2007, it entered into such a license with Sony by offering a substantial advance and an administrative payment. Id. at 38.

37 . During the negotiations over this Compendium amendment, ASCAP's counsel communicated to Sony that there were antitrust concerns with the carve-out proposal.

38 . EMI and ASCAP estimated ASCAP’s market share as 47%. The parties determined the revenue base against which the 1.85% would be applied by calculating Pandora’s revenue, multiplying it by the percentage of tracks played that embodied EMI's catalog, and multiplying that number by ASCAP’s estimated 47% market share. This calculation required EMI to provide Pandora with a list of its works, which it did monthly until Sony acquired EMI.

39 . [REDACTED],

40 . Sony’s attitude to a negotiated Pandora-ASCAP license had been clear for months. In an email of October 4, Sony (which by that time controlled EMI) refused Pandora’s request to disclose the terms of the Pandora-EMI license to ASCAP.

41 . Prior to that meeting, LoFrumento had discussed with a few committee members the terms on which Pandora and ASCAP had agreed in principle and which he intended to reject.

42 . Sony Corporation and other investors purchased EMI Music Publishing companies in June 2012. With that purchase, Sony/ATV undertook the administration of EMI, which remains a separate entity.

43 . EMI's Michael Abitbol was also a participant in the call.

44 . Costin wrote:

While we remain hopeful that we will reach mutually acceptable terms, we also find ourselves in a position where we must prepare for the possibility that we are unable to obtain a license prior to January 1, 2013, which is the date that has been signaled as the effective date of the Sony/ATV withdrawal of certain of its compositions for certain uses. I am writing, therefore, to request that Sony/ATV identify the specific musical compositions that it intends to withdraw from each of [the PRO’s] license authority effective as of January 1, 2013.

45 . Brodsky testified that Sony did not provide Pandora with a list of works because, when he contacted Rosenbloum regarding Ms. Cos-tin’s request, Rosenbloum replied that “there was no need for Sony/ATV to provide such a list of works because we were very close to finalizing a deal.” Rosenbloum denies ever telling Brodsky any such thing. Brodsky also testified that Rosenbloum did not make oral requests for the list of works in between the November 1 written request and the request during the breakfast meeting on November 30. While Rosenbloum was entirely credible in his testimony on these issues, Brodsky was not.

46 .ASCAP personnel shared their amusement with each other over Sony's decision to withhold the list from Pandora. In one email, DeFilippis asked ASCAP’s counsel Richard Reimer "Why didn’t Sony provide the list to Pandora,” to which Reimer replied "Ask me tomorrow,” to which DeFilippis responded "Right. With drink in hand.”

47 . Pandora needed the publishers’ list of works before it could take any steps to remove them from the Pandora service. Once Pandora had the list, it could quite quickly remove any song with an identical title and eliminate the copyright infringement risk. It would take Pandora more than a week, however, to identify which songs with identical titles but from other publishers could be reintroduced into the Pandora playlist.

48 . Although Brodsky denied knowing that anyone at Sony had leaked the terms of the license to the press, the evidence is that Sony did just that. Despite reporting dutifully that Sony had "declined” to comment on the terms of the deal, the articles referred to anonymous industry insiders as their source and quoted Bandier’s analysis of the deal. While Pandora had absolutely no interest in seeing the 25% hike in its rates known to other licensors, Sony hoped that its rate would be a jumping off point for the next publisher's negotiations with Pandora, and it was. Pandora had its attorneys call Sony to complain of the breach of their confidentiality agreement.

49 . Unlike Pandora, Spotify is an on-demand service. As an on-demand music service, its 10.5% rate is set by a licensing board and covers mechanical rights and a public performance right. The public performance right rate constitutes an offset from the overall 10.5% rate.

50 . Upset by Sony’s public disclosure of the Pandora license terms, Pandora requested an NDA that would bar UMPG from revealing the terms of any license. UMPG refused to include any such restriction -in the NDA.

51 . While Horowitz took the position at trial that he had believed in 2013 that Pandora was free to use the UMPG list of works to remove those works from the Pandora service, despite the requirement that Pandora execute the NDA, that testimony was not credible. Horowitz was intimately involved in these negotiations and is a strong executive. He had no desire to strengthen Pandora’s hand. If Horowitz had intended to give Pandora the ability to use the list of works to remove the works from its service, then there would have been no need for any NDA. The only confidential information described in the NDA was the list of works. Nor was the creation of the NDA a trivial matter. The negotiations over the NDA were carefully managed. At trial, Horowitz opined that UMPG had no legal obligation to provide a list of works to Pandora.

52 .Pandora’s purchase of KXMZ-FM remains pending. ASCAP has petitioned the FCC to deny the transfer of the station’s FCC license to Pandora.

53 . On June 6, Horowitz wrote to Rosenbloum that he was “calling [about] Pandora. We haven’t heard anything ... There’s almost no time left. Very odd process.” And on June II, Horowitz wrote to Rosenbloum that he was "disappointed that Pandora has chosen not to respond to our proposal” and that "[w]e are confused by Pandora’s unwillingness to respond in any way to our repeated inquiries for direction, even if to simply advise us that it no longer desires to license our music.”

54 . After the Opinion was rendered, the publishers sought intervention nunc pro tunc for the sole purpose of appeal. The motion was granted but the publishers were limited in the arguments they could raise. See In re Pandora Media, Inc., 12 Civ. 8035(DLC), 2013 WL 6569872, at *10 (S.D.N.Y. Dec. 13, 2013).

55 . SESAC's vice president for new media licensing confirmed in his deposition that in the context of unrelated negotiations SESAC had used a 10% figure to describe its market share.

56 . A locker service stores a customer’s digital music in the cloud, and permits the customer to access the music through multiple devices.

57 . Approximately forty percent of Pandora’s listeners access Pandora through their Apple devices.

58 . eMasters are defined as sound recordings available for download from the iTunes Store.

⁵⁹

60 .ASCAP’s expert did attempt to make one “rough” adjustment to address the absence from the Apple revenue base of any subscription income from iTunes Match. Using the percentage of revenue derived from Pandora’s subscription service to estimate the amount of iTunes Match subscription revenue, Dr. Murphy concluded that even if Apple's share of subscriber hours was twice as large as Pandora’s, the Apple licensé rate implies an ASCAP rate for the Pandora license which exceeds what ASCAP is seeking here.

61 . There is a suggestion in the record that the Apple negotiations over a public performance licensing fee may have been influenced by its overall obligation to pay for music content, including its ability to negotiate more favorable rates with record labels. There was insufficient evidence, however, to permit any reliable finding in this regard.

⁶²

63 . Pandora's argument that it is entitled to the rate in the RMLC license is addressed below.

64 . At points in this litigation ASCAP also cited the Spotify-ASCAP license as a potential benchmark, but it did not press this benchmark at trial. In all events, the Spotify license is a manifestly poor benchmark because it is a license for an overwhelmingly on-demand service and the public performance rate need not be closely negotiated since it is simply a component of an overall 10.5% rate for mechanical rights. See 17 U.S.C. § 115; 37 C.F.R. § 385.12(b)(2).

65 . Among other things, ASCAP asserts that it was pure of heart. It points out that it denied UMPG the waiver from the notice period that UMPG sought when it gave notice of its intent to withdraw new media rights from ASCAP. LoFrumento also explained that, while he considered the pressure exerted by UMPG and Sony, he rejected the term sheet negotiated with Pandora because of his own independent judgment. And Sony’s Brodsky denied at trial being involved in or knowing who had leaked the confidential license terms that appeared in the early 2013 press reports.

66 . ASCAP suggests that Pandora could have begun to pester Sony for a list of its works as soon as Pandora learned of the Compendium modification. There is no reason to find that any business executive would have considered that wise.

67 . After it executed the license with Sony, Pandora had no incentive to steer listeners away from Sony works; Sony had demanded a sizable but recoupable advance.

68 . Since Sony controlled about 30% of the market (counting the EMI repertoire), it would have been difficult for Pandora to operate for any length of time without access to any Sony composition. Nonetheless, the threat of being removed, substantially removed, or even incrementally removed from a service as popular as Pandora would be a risk that Sony would need to weigh with care. Sony’s determined refusal to provide the list despite repeated requests over the license negotiation period is testament to the importance Sony itself placed on this bargaining chip.

69 . Statutory copyright damages are up to $150,000 per work. See 17 U.S.C. § 504(c)(2).

70 . This Opinion does not take a position on whether a fair market price would have resulted if Sony had provided a timely list of its works to Pandora in 2012. In judging the extent to which any future benchmark provides guidance about fair market value, the totality of the circumstances that surround the creation of that future license will have to be considered.

71 . While the implied ASCAP headline rate of the Sony-Pandora license is 2.28%, the net effective rate of the Sony-Pandora license, calculated in light of the [REDACTED] deduction for internal advertising costs, is [REDACTED],

72 . As described above, the UMPG agreement was contingent on the outcome of (1) Pandora's summary judgment motion that the publisher withdrawals had no effect on its ASCAP license application, and (2) a determination that Pandora is entitled to the RMLC rate as the result of its purchase of a terrestrial radio station. Pandora succeeded with the first motion in this Court, which means that the UMPG license will have no effect unless that decision is reversed on appeal.

73 . Pandora is correct that the NDA unambiguously prohibited Pandora from using the list to take down UMPG works from its service. "Whether a contract is ambiguous is a question of law.” VAM Check Cashing Corp. v. Fed. Ins. Co., 699 F.3d 727, 729 (2d Cir.2012) (citation omitted).

74 . ASCAP contends that Pandora’s failure to object to the NDA on this ground is evidence that the complaint about the NDA’s terms is an after-the-fact concoction. To the contrary, UMPG’s insistence on an NDA addressed exclusively to the list of works was reasonably taken by Pandora as further evidence that any efforts to negotiate with UMPG would be futile.

75 .ASCAP contends that the implied rate for ASCAP should be even higher. ASCAP uses a 7% figure rather than a 10% figure, relying on materials obtained in discovery, to show that SESAC's share of PRO revenue (as opposed to its share of compositions) in 2011 was 7%. Holding this measure of SESAC’s share steady, and again making no adjustments for any growth of SESAC’s share and concomitant decline in ASCAP’s, ASCAP calculates that the 7% figure yields an implied ASCAP rate of [REDACTED] in 2011, [REDACTED] in 2012, [REDACTED] in 2013, [REDACTED] in 2014, and [REDACTED] in 2015. But it is not the industry practice to use share of PRO revenue as the relevant number in calculating implied rates. Moreover, the relevant figure is what Pandora thought the SESAC market share constituted at the time it entered into the agreement.

76 . Through this discussion of the SESAC license, this Opinion should not be read to endorse a rate structure in which an increasing market share justifies an increase in rate.

77 . On-demand services have also existed since at least 2001, when Rhapsody was launched. While not the most direct competitors with radio, they are in the same general market.

78 . As one more example, for ten years, through two five year terms, the rate for mechanical rights for on-demand services has been set at the same level: 10.5%. See 37 C.F.R. § 385.12(c).

79 . There was evidence at trial that Pandora may recently have added as many as a million more compositions to its music service for some purposes.

80 . For example, Pandora conducted experiments in the Summer of 2013 called the "Label Experiments,” in which it determined that it could substitute away from certain record labels with little negative response from Pandora users.

81 .ASCAP did not attempt to show at trial that Pandora's need for an extensive library of compositions was greater than, to take an example, Clear Channel’s iHeartRadio, which ASCAP licenses at a rate of 1.70%. ASCAP did offer evidence from a surveyor of major commercial radio stations, which it used to argue that Pandora performs a wider variety of songs from a wider selection of artists than terrestrial radio. Given Pandora's ability to substitute songs within a genre, however, the survey may show little more than Pandora's flexibility.

82 . It is also worth, noting that there is no evidence that ASCAP has taken action to license customized radio at a higher rate than programmed radio in any context outside of Pandora.

83 . There is a well-documented history of record promoters going so far as to use bribes, or "payola,” to increase the number of times songs are played on a radio station.

84 . Pandora’s "buy button” resulted in over $3 million per month in music sales on Amazon and the iTunes Store during 2013.

85 .The preamble to Apple's licenses with publishers for its iTunes Radio service provides that "[w]hereas, Apple wishes to enter into this Agreement with Publisher to enable Apple to create an advertising supported Internet radio service, which will further enhance the music discovery and recommendations features of the iTunes Store for the purpose of promoting sales of eMasters.”

86 . ASCAP relies on an annual 2012 study by a firm called NPD which showed that Pandora users tend to purchase less music than do users of on-demand services like Spotify. But this does not show that Pandora is more cannibalistic of music sales than on-demand services (or that it is cannibalistic at all). Correlation does not equal causation, and the disparity may be fully explained by the self selection of music users into on-demand services versus customized or programmed radio services. Users of on-demand services tend to be music “super fans” who know what they want to listen to and use on-demand services as supplements to purchased music collections. Users of customized radio services like Pandora tend to be more casual, or “lean back” music listeners, who are less likely to purchase music for their own collections.

87 . See also AFJ2 §§ VIII(A) (prohibiting discrimination in the types of licenses offered); IX(F) (presumption of setting interim rates at same level as those between ASCAP and licensees similarly situated to an applicant).

88 . The parties debate the significance of dicta in In re Applications of Salem Media of California, et al., 981 F.Supp. 199, 201 (S.D.N.Y.1997), which questioned whether the comparison in a similarly-situated analysis in the context of a multi-member license should be to the median licensee or to any given licensee. Because of the conclusion drawn herein, it is unnecessary to resolve that question.

89 . As AFJ2 requires, nothing in this Opinion may be construed as affecting the present ASCAP-RMLC license. See AFJ2 § IX(G) ("[A]ny license agreement that has been executed between ASCAP and another similarly situated music user prior to such determination by the Court shall not be deemed to be in any way affected or altered by such determination for the term of such license agreement.'').

90 . If ASCAP revises its rate structure it will no doubt be attuned to the need to treat major competitors in a market fairly.