Copyright Board Canada

Commission du droit d’auteur Canada

 

Date of Decision	2009-04-08
Corrected version	2009-05-06
Citation	Files: Public Performance of Musical Works; Public Performance of Sound Recordings; Reproduction of Musical Works
Regime	Collective Administration of Performing Rights and of Communication Rights Collective Administration in Relation to Rights Under Sections 3, 15, 18 and 21 Copyright Act, subsections 68(3) and 70.15(1)
Members	Mr. Justice William J. Vancise Mr. Stephen J. Callary Mrs. Jacinthe Théberge
Proposed Tariffs Considered	SOCAN (2005-2009), NRCC (2007-2010), CSI (2006-2009)

statement of royalties to be collected by socan, nrcc and csi in respect of multi-channel subscription satellite radio services

Reasons for decision

I. INTRODUCTION

[1] The Society of Composers, Authors and Music Publishers of Canada (SOCAN), the Neighbouring Rights Collective of Canada (NRCC) and CMRRA/SODRAC Inc. (CSI) filed tariffs for the use of their repertoire by Multi-Channel Subscription Satellite Radio Services. The proposed statements of royalties were filed pursuant to subsection 67.1(1) and 70.13(1) of the Copyright Act (the “Act”).

[2] SOCAN’s proposed statements for 2005, 2006, 2007, 2008 and 2009 were published in the Canada Gazette on May 1, 2004, May 14, 2005, May 20, 2006, June 23, 2007 and June 14, 2008. NRCC’s statement for the years 2007 to 2010 was published on May 6, 2006. CSI’s statements for 2006 to 2009 and for 2008-2009 were published on April 30, 2005 and May 19, 2007. [1] Potential users and their representatives were advised of their right to object. Sirius Satellite Radio (Sirius) and Canadian Satellite Radio Inc. (CSR), collectively the “Satellite Radio Services” did so.

[3] The examination of these three matters was consolidated at the request of the Satellite Radio Services, for the reasons set out in the Board’s order of July 21, 2006. The hearings took place over eleven days, from November 26 to December 11, 2007. The record of the proceedings was closed on February 13, 2008 with the parties’ last filing of supplementary materials.

A. Position of Parties

[4] SOCAN initially proposed a rate of 25 per cent of total gross income. That proposal was however revised downward substantially at close to 15 per cent, and then to about 13 per cent following the parties’ replies.

[5] NRCC initially proposed a rate of 17 per cent of gross income, with a minimum fee of $1.50 per subscriber per month. NRCC later revised its rate proposal to 4.4 per cent and then 4 per cent, without specifying any minimum fees.

[6] CSI initially proposed a rate of 5 per cent of gross income, with a minimum fee of $0.50 per subscriber per month. CSI’s new proposed tariff for 2008 and 2009 included rates of 5 per cent with a minimum fee of $0.50 per subscriber per month for a service that does not authorize subscribers to reproduce musical works, and 10 per cent with a minimum fee of $1 for a service that does. These rates were revised to 2.9 per cent ($0.29) and 5.8 per cent ($0.58).

[7] CSR did not propose any specific rates, although it was of the opinion that it should be lower than the rate paid by commercial radio stations. Sirius proposed SOCAN rates of between 2.0 and 2.6 per cent of income, NRCC rates of between 0.5 per cent and 0.7 per cent, and CSI rates of between 0.48 per cent and 1.14 per cent.

B. Description of the Multi-Channel Subscription Satellite Radio Services

[8] The satellite radio services industry originated in the United States. XM Satellite Radio (XM) launched its operation on September 25, 2001 and Sirius Satellite Radio (Sirius U.S.) launched its operation on July 1, 2002. They were the first and remain the largest operators in the world.

[9] XM uses two high-powered geostationary satellites that rotate in synchronization with the earth and provide blanket coverage of the entire U.S. mainland and southern Canada.

[10] Sirius uses three satellites that move around the earth in an elliptical orbit. These satellites are called geosynchronous and orbit above the equator for 16 hours a day and below the equator for 8 hours permitting the satellite to sleep and conserve energy.

[11] The multiplex signal sent by satellite to the mobile receivers is encrypted so that only those receivers equipped with a decryption key which permits the unscrambling of the signal can receive and play the signal.

[12] The American services were able to expand into Canada by forming exclusive partnerships with Canadian corporations. On June 16, 2005, the Canadian Radio-television and Telecommunications Commission (CRTC) issued broadcasting licences to CSR and Sirius to offer satellite radio services across Canada. CSR (XM Canada) launched its operation on November 22, 2005 and Sirius on December 1, 2005.

[13] At the end of 2004, XM reported over 3.2 million subscribers, and Sirius U.S. had reached the one million subscriber level. At the time of the hearing, Sirius and CSR had 200,000 and 120,000 subscribers respectively. By the summer of 2008, those numbers had increased to 750,000 and 440,000. [2] Satellite radio services quickly penetrated the market. It took Sirius U.S. 3.6 years to have 5,000,000 units in the hands of American customers, while DVDs took 2.5 years to reach the same amount, MP3 players 4.8 years, cellular phones 10 years and satellite television 10.6 years.

[14] For our purposes, the infrastructure and operations of the two American services, on which the Canadian services rely, are fairly similar. In order to provide an uninterrupted radio broadcast service, the U.S. satellite services augment their satellite signal through the use of a network of ground transmitters. This technique, which is said to create “space diversity”, prevents signal dropouts. With this combined infrastructure, the satellite services are able to deliver all of their programming to all subscribers, regardless of their location in North America at the time of reception.

[15] In terms of programming content, although each satellite service has developed its own micro-niche programming, both offer a large selection of commercial-free music channels covering a wide range of genres as well as channels of news, children’s programs, sports, comedy, talk and traffic. Additionally, a subscription to the satellite services offers the following innovative features: text display providing artists’ name, songs’ title, scores, stock quotations, a tagging mechanism alerting listeners when a song or an artist is playing on another channel, temporary and permanent recording options, pause and replay of live radio content, Internet service delivery of some audio channels over the Web as a streaming service as well as allowing the receiver to be used as a MP3 player.

[16] Programming of the U.S. services is created and delivered using a content management system (CMS) located at their main broadcast studio. The objective of this system is to store once and deliver many times.

[17] CSR uses a CMS provided by Delat Digital Media System. CSR produces and delivers 12 channels originating in Canada at studios located in Toronto and Montreal. Music directors at these two sites select the music to be used which is then injected in the system using functions of the Delat workstations located in these two cities. These workstations are directly connected to the main Delat CMS located in Washington, D.C. by a fibre optic line (OC3 line).

[18] Sirius uses a CMS called Nex Gen but does not produce any programming in Canada. All Sirius’ content in Canada is produced by Canadian third-party content providers. These providers generate and deliver the content to the Sirius master control centre located in New York City.

[19] XM has a complex of 82 studios in Washington as well as studios in New York, Nashville and Chicago. Sirius U.S. is based out of New York and has other studios in Los Angeles and Memphis. Programming is not typically delivered live, with the obvious exception of live sporting events. Essentially, before programming can be uplinked to the satellites for delivery, programming directors must store a copy of all music and audio files required onto the main server. These files are compressed, encoded and combined to complete the process commonly referred to as “multiplexing”. Selection and scheduling of programming content are done using specialized software that instructs the main server when and in what order it must play the various music or audio files. The server also serves the alternative delivery channels, including Internet and cellular phone streaming services.

[20] Although the Canadian satellite services rely heavily on their U.S. partner’s programming, the terms of their CRTC licence require them to include in their subscription package a minimum of content produced in Canada. Accordingly, out of the 130 channels CSR offers, 13 are produced in Canada while out of the 110 channels Sirius offers, 11 are produced in Canada. The Satellite Radio Services differ slightly in the way they create and deliver their Canadian content. It is useful, in light of the legal issues raised, to highlight the distinctive features of each Service.

[21] CSR creates its own programming. A digital communication link from the Canadian offices to the U.S. infrastructure allows the work stations in Canada to send instructions directly to the servers and the scheduling software sitting in U.S. headquarters in Washington. Thus, CSR programming is conceived and controlled in Canada but produced from Washington.

[22] CSR receives audio content in the form of CDs or through DMDS-Musicrypt [3] service provided by the sound recording industry. When dealing with a new CD, the production team makes a copy directly on the server in the U.S. using the digital connection, without making any back up or archival copies. New music obtained through DMDS-Musicrypt is received as digital audio files from a server that sits in Canada. In this case, an intermediary copy of the file is stored on a work station located in Canada. If the Canadian production team selects the song, then that file is “transferred” onto the main server in Washington via the digital communication link.

[23] When it comes time to scheduling program content, the programming director instructs the U.S. scheduling software to play specific songs and recorded voice elements in a certain order and at the appropriate time; the Washington server plays them off its local hard drives, combining the Canadian channels with the American ones into the common multiplex signal that is sent up to the satellite.

[24] Unlike CSR, Sirius does not produce any programming itself; it acquires all of its Canadian content from Canadian third-party content providers. Standard Radio Inc. provides Sirius with a Canadian rock music channel called Iceberg 95 created in studios located in Toronto. The content is available in CD and DMDS-Musicrypt. The music is scheduled from Toronto, loaded onto the Sirius master server where it is encoded and digitized for delivery to the server’s master control centre in New York City. Astral Media provides Sirius with two Canadian rock music channels, Rock Velours and Énergie, pursuant to a subcontract with Standard Radio. The programming is created in Montreal using the same technology used by Standard Radio. The music is scheduled from Montreal on a six-hour loop for broadcast each day by a program called Music Master. Content providers store the music files and create the programming on a server located in their respective broadcast studio. Again, if musical works are provided on a CD, a digital copy is made on the content provider’s server. If musical works are provided through DMDS-Musicrypt, a digital link to that service is used to copy that file onto the Canadian server. Sirius’ Canadian content providers do not make archival copies of musical works.

[25] Sirius’ content providers use a specialized scheduling software that is part of their server complex to determine which songs and other recorded voice elements will be played and when. When it is time for a show to air, the scheduling system automatically plays it off the copies on the Canadian servers. That output is linked by communication lines to the U.S. facility, combined with the other American channels and uplinked to the satellites. The content used on the Canadian originated signals is never actually stored on the Sirius U.S. server.

[26] In both cases, once the programming has been multiplexed and uplinked to the satellites, programming is delivered to the subscribers’ respective receivers in Canada and the U.S. The Satellite Services’ management system tells the Canadian receivers which channels a subscriber is entitled to receive and the U.S. satellite services’ management system does the same for its American subscribers. Although the signal that Canadian subscribers receive holds all the channels offered by both the U.S. and Canadian Satellite Services, because the signal is encrypted, they will only have access to a subset of channels.

II. THE EVIDENCE

A. Joint Evidence of Parties

[27] The parties jointly commissioned Erin Research to document the amount of music on all-music channels and the amount of music on selected channels on Sirius and CSR to measure the total amount of music on channels that the parties identified as relevant, that is on “talk channels”; to report on the amount of music used in, for example, commercials; and to report on the amount of music used by each network and each channel. The results accepted by all parties are that for Sirius the amount of music for non-music channels is 31.7 to 33 per cent, and for CSR, 25.3 per cent. They also agree that the music channels use music 94 per cent of programming time.

B. The Collectives

[28] CSI and NRCC commissioned the services of Paul Audley and Associates in consultation with Mr. Benoît Gauthier of Circum Network Inc. to develop a methodology and common database to analyze the use of their repertoire by the Satellite Services.

[29] CSI analyzed a sample of 6,147 plays to determine the percentage of plays of sound recordings used that are eligible for remuneration. It matched the titles in the sample to their existing database and contacted their members to confirm the ownership or eligibility of additional titles.

[30] Mr. Audley contends that ideally, the adjustment to the satellite services rate for CSI should take into account the number of times subscribers listened to musical works as part of the repertoire. However, Mr. Audley was unable to apply this method by reason that he did not have access to the necessary data. Instead, he relied on a 2006 Canadian Listening Study provided by Sirius Canada, ranking subscribers’ preferred genres of music and channels, to conclude that 72 per cent of songs played by Sirius Canada are part of CSI’s repertoire. When weighted for songs listened to, the ratio increased to 78.6 per cent. Applying the same adjustments to CSR, Mr. Audley found that 67.9 per cent of songs played were part of the CSI repertoire and that the compensable rate increased to 74.1 per cent when the ratio of songs played to songs listened to is accounted for. Overall, he estimated a repertoire share of 76.9 per cent for both Satellite Services.

[31] Ms. Nancy Smith of NextMedia was commissioned by the Collectives to prepare a report on the methods used by the Satellite Services to market their receivers, their subscriber services and programming content. She concluded that without music, there would be no viable business strategy for satellite radio. Although talk, news and sports are used to differentiate the CSR and Sirius brands, i.e., Shock Jock Howard Stern for Sirius and the NHL for XM, the main attraction for satellite radio remains commercial-free music.

[32] Mr. Benoît Gauthier conducted a survey on behalf of the Collectives to determine the relative importance of various programming and non-programming features of satellite radio on a consumer’s decision to subscribe or not to a satellite radio service. Since the Collectives did not have access to the Satellite Services’ subscriber lists, a sample of subscribers was obtained by placing computer-generated automated calls to a random sample of households to identify those who were satellite subscribers. Of the 4,201 persons identified as such, 1,000 personal interviews were conducted and 302 individuals completed a web-based questionnaire. Even though it is the best information that the Collectives could obtain in the circumstances, Mr. Gauthier concedes the sample is skewed towards Western Canada, is predominantly male and under-represents French speaking Canadians.

[33] Sixty-five per cent of respondents stated that they made the decision to subscribe to satellite radio, while 34 per cent received the service as a gift or as part of a promotion. Eighty per cent of those questioned were paying for the subscription, 16 per cent were receiving the service as part of a promotion, but were considering whether they would continue with the service, and 3 per cent had purchased a lifetime subscription.

[34] The most frequently cited reasons for subscribing to satellite radio were, in order of preference: geographic coverage by satellite radio (36 per cent); variety of programming (20 per cent); and music programming (15 per cent). The factors which were most critical to the decision of consumers to subscribe or continue to subscribe were: music programming (66 per cent); sports programming (18 per cent); and talk & entertainment (17 per cent).

[35] Eighty-six per cent of respondents stated that they would not have subscribed to satellite radio in the absence of music programming. The second most important reason for subscribing was talk & entertainment at 34 per cent.

[36] The respondents were asked how or if subscribing to satellite radio had modified their behaviour. The responses showed a 37 per cent decrease in the purchase of prerecorded CDs, a 65 per cent reduction in the purchase of music downloads, a 57 per cent reduction in the listening to conventional radio, and a 13 per cent reduction in watching news on television.

[37] The respondents also indicated a willingness-to-pay for recording capabilities on satellite radio receivers such as the ability to record music and to be able to rewind current programs. However, the study concluded that more advanced recording features increased the number of customers drawn to the recording product but not the amount such customers were willing to pay for the more sophisticated features.

[38] The Collectives engaged Dr. Ajay K. Agrawal, Professor of Economics at the University of Toronto and Dr. John McHale, Professor of Economics at Queens University to develop a methodology to value music as an input to satellite radio’s subscription services. Dr. Agrawal signed the initial economic report filed as Exhibit Collectives-4 and both Drs. Agrawal and McHale signed the reply to the objectors’ experts, Dr. David Reitman, Exhibit Collectives-5; both of them testified and defended the economic theories and approaches referred to in the various reports. We will therefore refer to both of them when discussing their evidence and the reports. With regard to the choice of the appropriate proxy, they favour digital pay audio (DPA) services. They propose however three other economic approaches based on fairness, efficiency and consistency, which are described later. All four approaches arrived at similar results, which the authors contend leads to greater confidence in the tariff they proposed.

C. CSI

[39] CSI, in addition to accepting the reports prepared by Drs. Agrawal and McHale, commissioned Mr. Paul Audley and Dr. Douglas Hyatt, to prepare a report determining the amount that should be paid by the Satellite Services for the right to reproduce and authorize the reproduction of musical works in CSI’s repertoire. Messrs. Audley and Hyatt agree with Drs. Agrawal and McHale’s proposal of a 37.5 per cent rate as the overall value of music use (including both the communication and the reproduction rights of authors, performers and makers of sound recordings). With respect to the appropriate proxy, Messrs. Audley and Hyatt are of the opinion that the tariff for limited downloads, subject to certain adjustments, is the closest proxy to satellite radio.

D. The Satellite Radio Services

[40] Dr. David Reitman, Principal, Charles River Associates International, was commissioned by the Satellite Services to review and critique the economic report prepared by Drs. Agrawal and McHale. He criticized their work for relying too much on theoretical models (for both the Shapley Value and the Efficiency Approaches) and for placing too little emphasis on the traditional methods using relevant market information. In his opinion, it is more economically sound to start with a market rate for alternative music services that already exists and then adjust for the differences between those services and satellite radio services.

[41] The Satellite Services also commissioned Mr. Kenneth Goldstein, media consultant, to propose a methodology for establishing the tariffs. Having compared DPA to satellite radio and commercial radio, he concluded that commercial radio is the more appropriate proxy. He also criticized Circum Network’s attempt to value music as a retail experience rather than valuing the repertoire at the wholesale level. His analysis will also be considered in greater detail.

E. Sirius

[42] Sirius commissioned Angus Reid to do a survey of its subscribers to compare the results obtained by Circum Network. That survey used the actual list of Sirius subscribers. Angus Reid conducted a total of 400 telephone interviews of a representative sample of urban and rural as well as regional subscribers. The survey identified the benefits to consumers as:

• Programming packaging – the presentation and breadth of that content;
• Programming content – music, talk, etc.;
• Reception (geographic); and
• Commercial-free.

[43] Mr. Reid concluded that there was no one specific benefit that drove the consumer to subscribe. At best, 50 per cent of subscribers made their decision to subscribe based on the commercial-free music channels. Accordingly, Mr. Reid concluded that the Circum Network study exaggerated the value of music. Mr. Gauthier examined the Angus Reid Survey and concluded that the questions asked were either biased or unclear and for that reason, in his opinion, the survey results are misleading and flawed.

III. LEGAL ISSUES

[44] All parties agree that SOCAN and NRCC are entitled to a royalty, with the result that there are no legal issues to be resolved with respect to them.

[45] The same cannot be said with respect to CSI. Its claim with respect to the right to reproduce music in its repertoire raises four legal issues. First, can CSI claim royalties in respect of reproductions made directly on a server located in the U.S.? Second, does the 4 to 6 second buffer constitute a reproduction of a substantial part of a work within the meaning of subsection 3(1) of the Act? [4] Third, does the Internet-based streaming of the satellite services’ signals involve such a reproduction? Fourth, do the extended buffer, radio replay, pause and rewind feature and sampling of programming content for promotional purposes by retail outlets involve such reproductions?

A. Reproductions Made in the U.S.

[46] The issue is whether the reproductions made in the U.S. by American third-parties for both Sirius and CSR as well as those made by CSR itself are subject to Canadian copyright law.

[47] The parties generally agree that the programming copies are reproductions within the meaning of subsection 3(1) of the Act. The parties also agree that Sirius authorizes the reproduction made by Canadian third-parties for the reproduction of its Canadian channels which are delivered directly to its U.S. partner.

[48] The Satellite Services agree that they are liable for programming reproductions made in Canada. Such copies are limited to those made by Sirius’ Canadian content providers who store the programming copies on their respective servers and to the temporary copies CSR makes of musical works received through DMDS-Musicrypt. Thus, the issue is restricted to whether the Satellite Services are liable for copies made by CSR in the U.S. (from Canada) and by American third-party suppliers.

[49] Programming is an essential and critical part of this application for a tariff. The issue of where the programming takes place is critical to the right to be granted a tariff for the reproduction right.

[50] The Satellite Services contend that they are not liable for copies made in the U.S. because the Act is not intended to and does not have extraterritorial effect. They cite and rely on an earlier decision of the Board in support of their position that a tariff for reproduction could not extend to acts occurring abroad. The Board stated that it “does not understand how the act of reproduction, when performed abroad, could trigger liability under the Canadian legislation on copyright.” [5] They rely on a number of cases to support their argument on the applicability of the territorial principle to the Act, including Def Lepp Music v. Stuart-Brown [6] where the English Court stated:

It is therefore clear that copyright under the English Act is strictly defined in terms of territory. The intangible right which is copyright is merely a right to do certain acts exclusively in the United Kingdom: only acts in the United Kingdom constitute infringement either direct or indirect of such a right […] [7]

[51] They also rely on the decision of Society of Composers, Authors and Music Publishers of Canada v. Canadian Association of Internet Providers [8] where the Supreme Court, citing as authority Professor Vaver, stated:

Copyright law respects the territorial principle, reflecting the implementation of a “web of interlinking international treaties” based on the principle of national treatment. [9]

[52] The Satellite Services also argue they have not authorized any foreign entities to make programming copies for use as part of the Canadian satellite radio services and even if they did, such authorization would escape the application of the Act by reason that the reproduction itself does not occur in Canada. In their submission, the situs of the primary infringing act determines whether or not liability under the Act for authorizing the infringing act occurs. The situs of the authorization is irrelevant. They rely on ABKCO Music and Records Inc. v. Music Collection International Ltd. [10] as authority for the argument that: “the requirements of territoriality are satisfied by the need for the act authorized to have been done within the United Kingdom.” [11] It bears noting that the facts at bar are the reverse to those surrounding that case; CSI is looking to engage the Satellite Services’ liability under the Act for authorizing in Canada, infringement abroad.

[53] In Laddie’s The Modern Law of Copyright and Designs, [12] the leading U.K. textbook on copyright, the authors state:

[A] person who “authorizes” another to do an “act restricted by the copyright” may himself be liable; and in this case the territorial exclusion does not apply, in that it is possible to infringe copyright by authorizing abroad, although of course the act which is authorized to be done must itself be a restricted act, i.e., an act to be performed in the UK. [13]

[54] The Satellite Services rely on American authorities including Subafilms v. MGM-Pathe Communications [14] wherein the Ninth Circuit Court of Appeal held that a person could not be held responsible under the U.S. Copyright Act for primary acts that occur entirely overseas.

[55] The Satellite Services reject the application of the real and substantial test to reproductions that are made wholly within the U.S. or to acts that authorize such reproductions because unlike a communication, a reproduction does not occur concurrently “here and there” and there is therefore no need to localize the acts of infringement. They also contend that the “connecting factors” discussed in SOCAN v. CAIP (SCC) [15] are not relevant.

[56] The Satellite Services contend it is not an infringement to authorize or cause a person to do something the person has a right to do. [16] They argue that CSI provided no evidence that Sirius U.S. or XM U.S. breached any copyright law in the U.S. Even if there was evidence that the reproductions by their U.S. counterparts infringed copyright it would be a violation of the U.S. Copyright Act and not the Canadian Act.

[57] They also rely on Théberge v. Galerie d’Art du Petit Champlain Inc. [17] to support their contention that it is neither contrary to subsections 30.8(4) or (5) nor 30.9(4) or (5) of the Act, to store digital copies of a work on a server; simply using or possessing a work without any associated copying is not an infringement. [18]

[58] Finally, the Satellite Services claim that CSI abandoned its right to make any claim for copies of music made in the U.S. in the Responses to Interrogatories. [19]

[59] CSI contends that the Satellite Services authorize in Canada the reproduction of programming both in Canada and the U.S. It also argues that XM U.S. reproduces work that was previously reproduced by CSR to create Canadian programming.

[60] CSI argues that the Satellite Services authorize American reproductions because the restricted act of authorizing occurs in Canada with the result that a licence is required to provide the authorization. In its submission, implicit authorization is established by the fact that the Satellite Services could not operate their service without contracting with American third-parties to produce and program American content on their behalf in lieu of taking the necessary technical and practical steps to assemble and deliver the musical works to Canadian subscribers themselves. [20]

[61] The Satellite Services pay a fee to use the American programming in their own services. CSI assumes that the fee takes into account the programming reproductions, because they are essential to the operation of the Satellite Services. In its submission, there is no difference between the authorization the Satellite Services give to their American partners to make reproductions in the U.S. and the authorization Sirius gives to its Canadian subcontractors (for which the Satellite Services concede they are liable). CSI relies on the decision of the Federal Court of Appeal in Society of Composers, Authors and Music Publishers of Canada v. Canadian Association of Internet Providers [21] to argue that authorization can be inferred from a party’s conduct. [22]

[62] CSI also contends that the location of a third party to whom authorization is granted is not relevant. In its submission, the situs of the authorization is essential, and SOCAN v. CAIP (FCA) stands for the proposition that the location of the authorization may be the determining factor when liability is engaged for having authorized infringement. [23] CSI also emphasizes that the right to authorize a reproduction and the act of making the reproduction are separate rights reserved to the copyright owner with the result that the Board’s comments in SODRAC Video-copies [24] are not pertinent.

[63] CSI further contends that the “real and substantial connection” test should be used to determine the choice of law with respect to the act of authorization. In its submission, there is no reason that the test applied in SOCAN v. CAIP (FCA) in the context of a communication right could not also apply to a reproduction right. [25]

[64] For the reasons that follow, we agree with the Satellite Services that the Canadian Copyright Act does not extend to the acts of reproduction made in the U.S. with the result that we do not have jurisdiction to grant a tariff for such acts.

[65] The actual location of the copies is not in issue. All parties agree that except for the copies made by Sirius Canadian content providers and the ones temporarily made by CSR at the content selection stages, the programming copies are located and made on the master servers in New York or Washington. What the parties do not agree on is the reach of the Canadian Act. In other words, can the Board certify a tariff extending to acts that occur outside Canadian borders. Factually, some reproductions are initiated in Canada by CSR but they are actually created directly on the master servers in the U.S. Other reproductions are made by the American content providers directly onto the American master server from which they are broadcasted to both American and Canadian subscribers.

[66] There are two different and distinct legal issues that arise from this factual situation:

• CSR copies initiated in Canada but made in the U.S.;

• copies initiated and made in the U.S. by American third-parties.

We will deal with these in order.

B. CSR copies initiated in Canada but made in the U.S.

[67] The issue of which law applies is different from the issue of jurisdiction. The B.C. Court of Appeal stated in Tolofson v. Jensen, [26] the “[c]hoice of law refers to the conflicts rules which have developed, through legislation or jurisprudence, in order to determine which system of substantive law the forum court will apply in respect of a legal matter having connection with other jurisdictions. There are different choices of law rules for different areas of law.” [27]

[68] Walker, in his text on Canadian Conflict of Laws [28] states that Canadian laws dealing with intellectual property are territorial [29] and are limited to the national territory. [30] Canada is also bound by certain international conventions that provide for, among other things, national treatment. [31]

[69] Jurisdiction on the other hand is the power of the court to hear a particular matter. The test for jurisdiction in Canada is whether there is a real and substantial connection between the forum and the litigation. A court may have jurisdiction to hear a matter because of the presence of one of the parties to the dispute in the jurisdiction but the real question is whether the court will “exercise” that jurisdiction. The choice of law, while not determinative, may influence the court’s discretion to exercise its jurisdiction.

[70] In the context of copyright law, there are two rules that apply to the choice of law conflict: (a) territoriality; and (b) the real and substantial connection test.

[71] In our opinion, the rule of territoriality applies to the reproductions at issue and consequentially it is the American Copyright Act that applies. Rightholders abroad rely on equivalent legislation to protect their copyrights. The Berne Convention and other international agreements impose minimal standards to which countries generally adhere. [32] The reproductions here lack the characteristics common to those actions which have been found to occur “here and there” under the real and substantial connection test, including such indicia as international communication, international transactions, or torts based on negligence. All of these “actions” are the product of discrete elements that can occur across national borders and all require more than one participant. Those characteristics are generally not associated with the act of reproduction.

[72] The act of reproduction occurs when a copy is created. In this case, the reproduction occurs when a digital copy of a work is put on the server located in the U.S. This occurs when the electronic mechanism activated by the production team in Canada creates the file on the American master server. Without the stored copy, there is no reproduction. In our opinion, the fact that the location of the output is different from that of the person pressing the button to create the copy is not important.

[73] The situation is not unlike what would occur if an Ottawa resident went to an Internet café while in New York to change the settings of the programmable thermostat located in his residence to activate the furnace to change the temperature of the house. The furnace would be turned on in Ottawa, irrespective of where the “enter” key that is used to turn on the furnace is located. Similarly, one can envision a situation where the director of production of a radio station with its head office in Windsor, Ontario calls his production manager working in the broadcast studio in Detroit and instructs him to copy a new single of a certain artist on the U.S. server in order to incorporate it into the upcoming program. In that case, even though the instructions were given from Canada, the copies were clearly made in the U.S. and the U.S. Copyright Act would govern the reproduction of the musical work.

[74] In light of our finding that the territoriality principle applies, it is not necessary to consider the substantial connection test. The territoriality principle dictates that the U.S. Copyright Act applies in this instance. Therefore, as the Copyright Board of Canada only has jurisdiction to set tariffs under the Canadian Act, the Board has no jurisdiction to set a tariff with respect to these copies.

C. Copies initiated and made in the U.S. by American third-parties

[75] The copies in question are those made in American broadcast studios for the programming of American channels, a number of which are included in the Canadian subscription package. The U.S. originated channels received by Canadian subscribers are prepared and programmed from studios in New York and Washington that are owned and operated by the Satellite Services respective American counterparts. These American third-parties make the reproductions for the purposes of programming the American channels.

[76] The evidence is unequivocal that the Satellite Services benefit and depend on the reproductions made by third-parties in the U.S. but have no involvement in their making. Liability of the Satellite Services for these copies can only occur if we find both: that authorizing from Canada an act that occurs abroad is an authorization “in Canada”; and that, in the absence of express authorization, authorization can be inferred by either the degree of control exercised by the Satellite Services or the nature of their relationship with their respective American counterparts.

[77] Once again we must choose which law applies to the authorization of an “infringing” act that occurs in the U.S. when the authorization is given in Canada. In our opinion, it is the American statute that applies for the following reasons.

[78] Briefly stated, the choice of law is determined by territoriality and cannot be based upon the “substantial connection test” when dealing with programming copies initiated in the U.S. by American third-parties. The evidence does not establish that these copies occur “here and there” or across borders. Thus, the reproductions lack the characteristics common to those actions which have been found to occur “here and there” in the jurisprudence cited by CSI.

[79] The making of the reproductions does not involve any action taken in Canada with the result that there can be no “injury” sustained under Canadian legislation. Therefore, the comments of the Supreme Court in Moran [33] that the wrongful act of the tortfeasor must have occurred where the injury took place are of no assistance to CSI. Equally, the comments in Tolofson (SCC) regarding the fact that “people expect their activities to be governed by the law of the place where [these activities] happen” [34] are not helpful. All of the activity involved in making the copies takes place in the U.S. and is therefore governed by U.S. legislation. The act of authorizing infringement abroad is covered by the act governing the primary infringement.

[80] CSI rightly points out that the act of authorizing is distinct from the act of doing. That being said, the Board has previously explained the interrelationship that exists between the two acts:

It is generally accepted that all acts protected in subsection 3(1), including the Authorization Right, are distinct. Even though it exists separately, the Authorization Right is, in the Board’s view, of a different nature since it is directly linked to the act being authorized. For one thing, the Authorization Right exists only if the act being authorized is itself protected by copyright. For example, the Public Performance Right can exist without the Communication Right being protected, but the right to authorize a communication cannot exist if the Communication Right is not protected. [35]

[81] If making a copy in the U.S. on instructions from Canada is subject to U.S. law, then, a fortiori, any authorization of the making of a copy in the U.S. from the U.S. is subject to U.S. law.

[82] It follows that the mere act of authorizing in Canada is not actionable under the Canadian Copyright Act where the primary infringement occurs abroad. There is therefore no need to consider whether or not the Satellite Services have in effect authorized the programming copies of the American channels. In any event, they have not.

D. The 4 to 6 Second Buffer

[83] Every satellite radio receiver stores 4 to 6 seconds of the Satellite Services’ multiplex signal at all times in its random access memory (RAM). The principle of first in first out is used to describe buffering – that is, the first byte of information received is the first byte out after 4 to 6 seconds. According to Mr. Greg Nease, Sirius’ technical expert, the information contained in the buffer is in a “native signal transmission”, which he described as “byte soup” containing “a lot of slicing, interleaving, coding for transmission so that they’re totally unrecognizable in terms of audio files”. [36] In his opinion, the information is inaccessible and completely invisible to users.

[84] Mr. Paul Marko, an expert called by CSR, explained that in the technology used by CSR the information is transmitted from the satellites to the CSR receivers’ by a process that spreads the signal out over 4 to 6 seconds and once received by the receivers’ RAM, the receivers perform a despreading function rendering it audible.

[85] The issue is whether the 4 to 6 second buffer – which stores in the temporary memory of the satellite radio user’s receiver a “rolling” four seconds of the signal received from the satellite or repeater which provides time diversity – is a reproduction of a substantial part of a work. To so conclude, we must find that: (1) buffers are in effect a copy of the protected work; (2) the copying is substantial; and (3) the resulting copy is in a material form.

[86] The Satellite Services contend that the buffers are not reproductions within subsection 3(1) of the Act because no new copy of a work in a material form is created. They also contend that the sequential transitory storage of small fragments of a musical work that is transmitted to facilitate time diversity fails to meet the substantiality requirement of a reproduction. In the U.S., the transient buffer is considered to be de minimis and therefore not infringing. [37] They cite as authority Australian Video Retailers Association Ltd. v. Warner Home Video Pty Ltd. [38] and Stevens v. Kabushiki Kaisha Sony Computer Entertainment. [39] They argue, as well, that the 4 to 6 second buffer does not meet the material form requirement that is defined by the case law that is “fixed in some concrete or non-evanescent form, having a more or less permanent endurance”. [40]

[87] Alternatively, the Satellite Services submit that the 4 to 6 second buffer benefits from the protection of paragraph 64.1(1)(d) of the Act, as it is essentially a method or principle of manufacture and therefore, should not be actionable under Canadian law. In support of this contention, the Satellite Services merely refer to SOCAN v. CAIP (SCC) [41] and Euro-Excellence Inc. v. Kraft Canada Inc. [42]

[88] CSI argues that all buffer copies are reproductions and attract copyright liability because aggregating the sequential content of the buffer will generate a copy of the whole work. The 4 to 6 second buffer is in a form that is sufficiently palpable, tangible and perceptible as to be fixed in a medium from which the works can be reproduced or otherwise communicated. [43] It relies on Twentieth Century Fox Film Corp. v. Cablevision Systems Corp. [44] which found, among other things, that buffer copies are capable of being reproduced and that the buffer copies in the aggregate comprise the whole of the plaintiff’s programming.

[89] CSI adds that the buffer copy is similar to a streaming copy which, according to the Board, contains all the constituent elements of copying, substantiality and material form. [45] CSI also submits that the 4 to 6 second buffer cannot be equated with the cache copy in SOCAN v. CAIP (SCC) [46] because the exception provided for in paragraph 2.4(1)(b) of the Act only concerns the communication right, not the reproduction right.

[90] We conclude that the 4 to 6 second buffer is not a protected reproduction within the meaning of the Act because it does not meet all of the three requirements we set out above.

[91] The 4 to 6 second buffer is a reproduction of the protected work, just as downloads and streams targeted in the CSI – Online Music Services Tariff [47] are copies. The buffer stores information, if temporarily, before it can be used. Thus, buffering like streaming involves an act of copying, which in our view satisfies the first requirement.

[92] The next issue is whether the 4 to 6 second buffer is a “substantial” part of a work copied. The Satellite Services rely on Copinger and Skone James to conclude that it does not: in commenting on a film being copied in the RAM of a machine as it was being played, the authors conclude that “each fragment is so small that at any one time there cannot be said to be a ‘copy’ of the film or a substantial part of the film in RAM and so no copying of the film will have taken place for the purposes of the Act. […] in the case of the fragmentary copy into RAM at no point does there exist anything which can realistically be said to be a copy.” [48] The Satellite Services fail to point out that the authors concede that the “restricted act” is the “copying” of the work, not the making of a copy such that this streaming process “could be a situation in which the principle of little and often could be adopted.” The authors do however conclude that if a copy is not produced at some stage of the process it seems unrealistic to say that copying has taken place. [49]

[93] The Satellite Radio Services also turn to Professor Paul Goldstein and argue that in the U.S. transient data buffer is considered to be de minimis and that copyright law sets the de minimis hurdle to infringement relatively low. [50] The Satellite Services rely on the two Australian cases, AVRA v. Warner and Stevens v. Kabushiki in support of their argument that sequential fragments do not amount to a substantial part of a musical work. It should be noted, however, that there are substantial differences between the Australian legislation and subsection 3(1) of the Act and that consequently, the cases are of limited assistance.

[94] CSI argues that by aggregating the sequential content of the buffer there will be a copy of the entire work. It relies on the decision of the Board in SOCAN 22.A which dealt with a communication right and not a reproduction right. CSI also relies on U.S. Copyright Office’s DMCA Section 106 Report. This report, in our opinion, is non-binding and, more importantly, it is inconclusive.

[95] Finally, CSI refers to Cablevision. There is no need to address this aspect of the argument, by reason that the decision has since been reversed, vacated and remanded by the United States Court of Appeal for the Second Circuit.

[96] The leading authority on how to assess substantiality remains U & R Tax Services Ltd. v. H & R Block Canada Inc., [51] in which Richard J. (as he then was) decided, among other things, that substantiality must be assessed by placing “more emphasis on the quality of what was taken from the original work rather than the quantity”. [52] Whether there has been a reproduction of a substantial part of a work is a question of fact which must be determined in light of all the circumstances in each case.

[97] Thus, the question is reduced to whether the 4 to 6 second buffer is a substantial part of an entire work. The rolling 4 to 6 seconds of a musical work is not an aggregate of an entire work. At no time does a subscriber possess a series of 4 to 6 second clips which when taken together would constitute a substantial part of the work. It matters not that over time the totality of all works transmitted are reproduced. We are dealing with a rolling buffer and at no time can we line up all of the fragmented copies amounting to one complete copy of a musical work. At no point in time can one extract from the RAM of the receiver more than 4 to 6 seconds of a song (or more accurately of a signal). More importantly, at no time is there a choice as to what goes in there or when it comes out.

[98] In our opinion, the 4 to 6 second buffer fails to satisfy the substantiality requirement. It is not a substantial part of the protected work. It is not, therefore, strictly necessary for us to deal with whether or not the buffer is stored in a material form. We will do so, however, to provide guidance in future cases.

[99] The Satellite Services contend that it is not in a material form because it is a fleeting copy and not perceptible to subscribers. It is byte soup stored in a “native signal transmission format”. They rely on U.S. and Australian case law which as we have noted is, because of differences in the respective copyright acts, of limited use.

[100] There is no definition of “material form” in the Act. In Canada, the case cited to determine whether a work is in “material form” is Canadian Admiral which stands for the proposition that “for a work to be protected by the Canadian Copyright Act, a work must be expressed in some material form capable of identification and having a more or less permanent endurance”. [53]

[101] In EROS-Équipe, the Federal Court defined “material form” based on its ordinary meaning of something that was palpable, tangible and perceptible. [54] The Court concluded that the use of software which showed forms owned by the plaintiff on computer screens infringed copyright because they were not ephemeral and they reproduced the plaintiff’s work in a material form within the meaning of the section. The Court relied on Bookmakers’ Afternoon Greyhound Services Ltd. et al v. Wilf Gilbert (Staffordshire) Ltd. [55] that found mere posting on a television screen constitutes material reproduction. The Court further added that the generation of a printed version of the on-screen version was also an infringing reproduction.

[102] The Board has recognized various ephemeral copies including streaming, caching, and permanent and temporary downloads as being in material form. In our opinion, RAM copies created by digital technology meet the requirement of material form. [56] It would be contrary to the case law to find that in order to be in a material form, the copy must be capable of being seen or retrieved. Even if it were so, similarly to the circumstances in EROS-Équipe, [57] the user who owns a receiver equipped with the extended buffer or recording features could potentially retrieve a more permanent ephemeral copy.

[103] In our opinion, the 4 to 6 second buffer does meet the third requirement. Having said that, it matters not given our conclusion on substantiality.

E. Internet Streaming

[104] The Satellite Services offer, as an additional feature, a free online service of a subset of satellite radio channels through an Internet-based streaming service using a specially designed media player. The content for the media player is delivered at virtually the same time as it is uplifted to the satellite. The streamed content is extracted from the same process by which radio satellite services are programmed and delivered. The stream is copied temporarily in the buffer of a computer to make it play without interruption. The buffer, which according to Dr. Michael J. Murphy never exceeds ten seconds, delays the playback just long enough to ensure smooth playback to the listener.

[105] A number of cellular phone service providers have entered into agreements with the Satellite Services to permit their subscribers to connect to the streamed service. The streaming technology used to deliver this service is identical to the one used on the Internet. The Satellite Services contend that apart from the “so-called copy” resulting from the buffering process, no other reproductions are created by this service. [58]

[106] The Satellite Services contend the Board has not yet decided whether the reception of a pure stream on a computer or other device involves the reproduction right. Any comments made by the Board to date are, in the submission of the Satellite Services, obiter and more importantly contradict recent decisions of the Supreme Court of Canada. [59] The Satellite Services contend this is the first time the Board is squarely faced with making a decision as to whether the transmission of information to a device in the course of a streaming service involves the making of a reproduction.

[107] CSI, on the other hand, argues that streams are reproductions within the meaning of subsection 3(1) of the Act and relies on the Board’s comments in CSI – Online Music Services for the Years 2005 to 2007. [60] In CSI’s submission, the evidence before the Board is that there is no difference between a stream in the context of the delivery of a satellite radio service and a stream in the context of a delivery of an online music service. It also relies on the comments made by the Board in SOCAN 22.A. [61] The Board did not decide whether or not the temporarily stored copy was a reproduction within the meaning of the Act in SOCAN 22.A.

[108] The Board decided in CSI – Online Music Services, [62] as we do here, that the Internet cellular phone streaming of the objectors’ services require the making of copies. The issue here, however, is whether or not these copies are sufficiently substantial to attract liability. In our opinion, they are not. The uncontradicted evidence of Dr. Murphy is that the buffer never stores more than 10 seconds of information. Based on our earlier comments with respect to the 4 to 6 second buffer, the same finding must be made, that is, they are not sufficiently “substantial” and therefore do not constitute a protected reproduction within the meaning of subsection 3(1) of the Act.

F. Extended Buffers

[109] A small percentage of receivers contain an extended buffer feature. This enables the subscriber to block record programming, to pause and replay, to radio replay which once enabled permits subscribers to record programming content for subsequent enjoyment, and to pause and rewind. All these features are user selectable: any decision to employ them is made by the user, not by the Satellite Services.

[110] The parties agree that these copies are reproductions within the meaning of the Act and that users, not the Satellite Services, make them. What is not agreed is whether the Satellite Services authorize their subscribers to make the copies in question.

[111] The Satellite Services note that all these features are used and initiated by the user, who alone determines what, if anything, is stored. In their submission, the services do not authorize or purport to grant the right to make copies. They rely on Muzak Corp. v. Composers, Authors and Publishers Association of Canada Ltd., [63] Vigneux v. Canadian Performing Rights Society Ltd. [64] and CBS Songs Ltd. v. Amstrad Consumer Electronics Plc. [65] Succinctly stated, their argument is that they do not have or exercise the necessary control to be deemed to have authorized or purported to authorize the reproductions. In their submission, based upon De Tervagne v. Beloeil [66] “an act is not authorised by somebody who merely enables or possibly assists or even encourages another to do an act, but does not purport to have any authority which he can grant to justify the doing of the act.” [67] They also rely on Muzak, [68] CCH [69] and SOCAN v. CAIP (SCC) [70] to argue that they are entitled to presume that the means they provide will be used in accordance with the law.

[112] CSI contends that on the evidence the Satellite Services “sanction, approve and countenance” the programming activities of their subscribers. In its submission, the Satellite Services exercise a sufficient level of control to infer authorization. [71]

[113] In our opinion, the Satellite Services have authorized a reproduction in the present circumstances. All the recording options contained in the “Stiletto” and similar receivers sold by CSR are dependent on the subscriber’s decision to use those features. The Satellite Services’ contention that they authorize the mere use of equipment that may or may not be used to infringe copyright which entitles them to presume that subscribers use the device in accordance with the law is not in accord with the evidence in this case. Here the Satellite Services are not passive. They control the programming sent to the subscribers by encrypting the signal, and by decrypting it they grant to their subscriber the right to access the full programming including the right to use all of those services. The Satellite Services can program their receivers to permit or prevent copying. With respect to block copying, pause and replay and other features, access to the content copied in the extended buffers is controlled by the Satellite Services. Subscribers who stop paying for the service no longer have access to the content stored in their receivers. In addition, some end-user licence agreements contemplate the possibility that subscribers will use the receiver software to copy content programming or even individual songs based on which a subscriber could presume that the Satellite Services purport to have the authority to allow private copying.

[114] Finally, the Satellite Services commercial interests are tied to the ability of subscribers to use the extended buffers. While to date the number of subscribers using the more sophisticated devices is quite small, the fact is that the Satellite Services rely on these features to promote themselves as a better alternative to other substitute services. For example, they market themselves as an alternative to iPods or MP3 players. They prominently advertise the reproduction features of their radio receivers. For example, CSR describes one of its receivers as a substitute for the iPod-iTunes combination proclaiming “another Pod? Hardly. This is the mother ship.”

[115] In SOCAN v. CAIP (SCC) the Supreme Court, in deciding whether or not Internet Service Providers could be deemed to authorize the communication of protected works, pointed out “that copyright liability may well attach if the activities of the Internet Service Provider cease to be content neutral, e.g. if it has notice that a content provider has posted infringing material on its system and fails to take remedial action.” [72] Based on the evidence before us, the Satellite Services’ activities are not content neutral. This finding is consistent with the decision of the Federal Court of Appeal in CCTA where the Board found the cable television services were liable for the transmitters’ public performances of music over their network on the grounds that they either authorized the performances or were joint performers. The Court stated:

However, if one wants to lay with the subscriber the ultimate responsibility for the materialization of the public performance and therefore the infringement of the copyrights, there is no doubt that, upon a plain or a constructive meaning of the word “authorization”, the appellant authorizes such materialization by its customers. I think the learned Trial Division judge correctly summarized the state of the law with regard to the appellant’s actions when he wrote:

I believe a cable television system which provides electromagnetic signals to a subscriber, under a contract which clearly contemplates that the sole use of the cable company’s service is to be the production of audible and visual messages from the subscriber’s television set connected to that cable, must be taken to have authorized that ultimate performance. [73]

[116] In our opinion, those comments apply with equal force to the present case.

[117] That brings us to the final issue and that is whether or not the Satellite Services can rely on the fair dealing right to escape liability for authorizing their affiliated retailers to reproduce a sample of programming content. We note that initially the Satellite Services sought to invoke the private copying regime to evade liability but appear to have resiled from that position. The Satellite Services now rely on the concept of fair dealing with respect to the programming content recorded for retail purposes. The Satellite Services further contend that even if the Board concludes that they authorized the copies, they cannot be liable for such reproductions by virtue of the fair dealing exception for research or private study purposes. In our opinion, that argument is a long reach and one which, as will be seen, is easily rejected. The Satellite Services contend that the samples recorded on the devices provided to affiliated retailers should come within the fair dealing right because the purpose is to facilitate private research of potential subscribers.

[118] The fair dealing right for the purposes of research or private study is governed by section 29 of the Act which states:

29. Fair dealing for the purpose of research or private study does not infringe copyright.

[119] The Supreme Court of Canada examined the issue at some length in CCH. [74] Chief Justice McLachlin, writing for the Court, stated that fair dealing is not an exception (as has previously been thought) but a right. In order to show that the dealing was fair, the defendant must prove on a balance of probabilities that the dealing was for research or private study and that it was fair. The Chief Justice described what is required to prove fair dealing in the following terms:

[60] To conclude, the purpose of the dealing, the character of the dealing, the amount of the dealing, the nature of the work, available alternatives to the dealing and the effect of the dealing on the work are all factors that could help determine whether or not a dealing is fair. These factors may be more or less relevant to assessing the fairness of a dealing depending on the factual context of the allegedly infringing dealing. In some contexts, there may be factors other than those listed here that may help a court decide whether the dealing was fair.

[120] In CCH, the Supreme Court recognized that the right equally applies to facilitators of research and private study. The Court stated unequivocally that the onus of establishing fair dealing lies with the person seeking to invoke it. In this regard, the Court stated:

[48] […] Procedurally, a defendant is required to prove that his or her dealing with a work has been fair; however, the fair dealing exception is perhaps more properly understood as an integral part of the Copyright Act than simply a defence. Any act falling within the fair dealing exception will not be an infringement of copyright. The fair dealing exception, like other exceptions in the Copyright Act, is a user’s right. In order to maintain the proper balance between the rights of a copyright owner and users’ interests, it must not be interpreted restrictively. As Professor Vaver, supra, has explained, at p. 171: “User rights are not just loopholes. Both owner rights and user rights should therefore be given the fair and balanced reading that befits remedial legislation.”

[121] The Satellite Services argue that in the context of a proceeding before the Board the burden of proof is shifted to the Collectives because they have the burden of establishing the reasonableness of their proposed tariff. This submission is untenable. The information respecting the intentional purpose is strictly within the control of the user. The onus of establishing fair dealing should remain, as the Supreme Court stated, with the person asserting the right. The Satellite Services have not provided any evidence to establish that the subscribers using the extended buffer were doing so for the purposes of research or private study. There is no aspect of the recording features that suggests that potential purchasers use the prerecorded messages for research or private study. Quite the contrary, the evidence demonstrates that retailers use this feature to promote the sale of the service.

[122] A sample of the content contained on devices for retail purposes and for the benefit of on-site sampling is distinguishable from the previews described in SOCAN 22.A. [75] The receivers sold in retail stores hold a prerecorded sample of the various channels in their memory. When the device is turned on by a potential buyer, it will play the prerecorded channel. The potential consumer has no choice in selecting channels other than what is offered by the Satellite Services. After the device is purchased, the subscriber must contact the satellite service and pay the subscription fee before the Satellite Services’ respective management systems will send a signal informing the receiver that the subscriber is entitled to the full subscription and to playback the recorded content. [76]

[123] Contrary to the facts in SOCAN 22.A, [77] here the consumer does not request a sample from the Satellite Services. On the contrary, it is the Satellite Services who initiate and make the copy. In our opinion, the Satellite Services cannot be considered facilitators of research. The prerecorded content offers a sample of the channels offered to incite customers to purchase their services.

[124] The Satellite Services have not provided sufficient evidence to discharge the onus of proof that this activity is fair dealing within section 29 of the Act. Thus, the Satellite Services cannot rely on fair dealing to escape liability for the reproduction of the sample programming.

IV. ECONOMIC ANALYSIS

[125] In the narration of the facts we described briefly the approaches advocated by the economic experts called by the various parties. Mr. Glen Bloom, when arguing the Collectives’ case, stressed that there is a problem with the Board continuing to use the benchmark approach by reason that it builds upon prior proxies. In his submission, a newer and more innovative approach is necessary to ensure that the certified tariff is a reflection of the true value of music to the industry. We have examined the alternative approaches developed by Drs. Agrawal and McHale and while we find them interesting and believe that they could become useful as the market matures, we are not persuaded to use them at the present time.

[126] We essentially agree with Dr. Reitman that the benchmark approach, even with the specific adjustments that are required, permits the Board to certify rates that are fair and equitable for satellite radio. Once a tariff has been established by the benchmark, it is not a simple reflection of the initial proxy, but rather a tariff arrived at as the result of an independent analysis of the specific sector to which the tariff applies.

[127] We will therefore use the benchmark approach to determine the fair and equitable tariff. However, we will briefly describe the alternative approaches proposed by Drs. Agrawal and McHale and will provide reasons why we cannot rely on them at this time. We offer these comments in the hope that they will provide guidance for the parties in developing new economic models in the future. In our opinion the development and presentation to the Board of such alternative economic models can, in the long term, play a crucial role in the determination of fair and equitable tariffs.

[128] We begin by examining the economic models proposed by Drs. Agrawal and McHale based on fairness, efficiency and consistency.

A. Industry’s Willingness-to-Pay

[129] Drs. Agrawal and McHale propose to use sample agreements to estimate the share of revenues the Satellite Services are willing to pay to the automobile industry to obtain subscribers, and apply this share to the subscribers that are attracted to the service because of the music. They claim this leads to a fair tariff since the Satellite Services should be willing to pay the same share of their revenues for all the subscribers, whether attracted by music or otherwise.

[130] Drs. Agrawal and McHale include every financial transfer between the Satellite Services and the car manufacturer. They find that the Satellite Services incur two costs from their agreements with the automobile industry: a hardware subsidy and a share of subscription revenues. In exchange, the car manufacturer pays for a trial subscription for a brief period when the car is sold. They also consider the percentage of the consumers that will retain the service once the trial period paid for by the car manufacturer is over.

[131] The authors conclude that, when including the trial subscription and other amounts, the Satellite Services are willing to pay approximately 45 per cent of revenues earned from automotive subscribers to car manufacturers.

[132] Drs. Agrawal and McHale also note that, according to the Circum Network survey, 86 per cent of current subscribers would not continue to subscribe to the satellite radio service if music was removed from the programming. Applying this to the proportion of 45 per cent leads the authors to conclude that the Satellite Services should be willing to pay 39 per cent of their revenues for music.

[133] The authors recognize, however, that music rights owners might not be able to capture all of the Satellite Services’ willingness-to-pay. On the other hand, they also recognize that the Satellite Services might be “willing to pay” even more to car manufacturers. They conclude that it may be unsustainable for the Satellite Services to pay at the maximum they are willing to pay for each input.

[134] Dr. Reitman disagrees with the calculation with respect to the subscription fee for trial months paid for by the car manufacturer. Rather than include trial months paid by car manufacturers in the revenues of the Satellite Services, he suggests they should net out the payments the Satellite Services make to car manufacturers to subsidize the equipment and share subscription revenues. He argues that although this treatment of trial months should make no difference in terms of business decisions for the Satellite Services and car manufacturers, it has an important impact on the cost-benefit ratio. Dr. Reitman also proposes certain changes to the exchange rate calculation.

[135] The result is that he calculates an estimated willingness-to-pay of approximately 10 per cent, much lower than the 39 per cent estimated by Drs. Agrawal and McHale. Moreover, in Dr. Reitman’s opinion the willingness-to-pay corresponds to payments for all music costs including music programming and only a proportion of it should constitute the tariff.

[136] In addition, Dr. Reitman notes that the Satellite Services would pay out more than their revenues if every input was remunerated at the maximum willingness-to-pay rate.

[137] Finally, Dr. Reitman calculates that, using the retail channel from which 75 per cent of customers originate, the Satellite Services are only paying 2-3 per cent of their revenues for a new subscriber-month. He argues that the Satellite Services are willing to pay more when it comes to car manufacturers because it provides brand equity to the service.

[138] The willingness-to-pay approach proposed by Drs. Agrawal and McHale is based on the assumption that the Satellite Services should be willing to pay the same amount for a new subscriber-month, whether it is attracted by the music offerings of the service or from an agreement with a car manufacturer. In our opinion, the evidence does not support this assumption. On the contrary, the evidence dealing with the retail channel is that willingness-to-pay can differ substantially between various inputs. Even the two car manufacturers do not obtain the same share of revenues from their agreements with the Satellite Services. Thus, the willingness-to-pay for a particular input does not appear to be a reliable proxy for the willingness-to-pay of another.

[139] When they began the negotiations with the car manufacturers, the Satellite Services had already incurred many of the costs required for providing the service and as a result, they were willing to pay a high rate to attract new customers, as long as the benefit exceeded the incremental cost. Drs. Agrawal and McHale recognize that one reason the car manufacturers might have been able to negotiate such a high share of revenues is because they arrived later in the development of the business. When negotiating with content providers, the Satellite Services were able to demonstrate they had to incur important start-up costs to deliver the service and as a result did not have a high profit margin to share. Once the service is in place and certain costs are unavoidable, the bargaining positions change. In theory, if a car manufacturer requested 99 per cent of revenues from every new subscriber they sign up, it could still be accepted by the Satellite Services.

[140] For these reasons, the willingness-to-pay model is, is in our opinion, fundamentally unsuited for the setting of a tariff. In general, the order of arrival of a particular input in a business should not make a difference to the tariff.

B. Shapley Value

[141] To address the problem of the order of arrival of a particular input, Drs. Agrawal and McHale propose to use the Shapley value approach, a methodology that models the Satellite Services as the result of an agreement between five members of a potential coalition: the Satellite Services and four content providers (music, news, sports and talk & entertainment).

[142] Drs. Agrawal and McHale examine the value of the five members of the coalition using data from Circum Network about the share of current subscribers that would unsubscribe if one or more members of the coalition were to withdraw from the agreement. They start with the proposition that nobody would subscribe to satellite radio if the Satellite Services were not part of the coalition, because there would be no service to provide the content; or if the Satellite Services were the only members of the coalition, they would have no content to offer subscribers. The survey asked respondents about the 32 possible coalitions, ranging from all the members in the coalition joining (where all of the current subscribers would subscribe) to no members joining (where none of the current subscribers would subscribe).

[143] Based on the 120 different ordering scenarios of the five members of the coalition, the authors determined the marginal value of each member, or what they also call the Shapley value of the Satellite Services and the four content providers. This approach led Drs. Agrawal and McHale to estimate that music should receive 34 per cent of revenues, compared with 44 per cent for the Services, 5 per cent for news content, 7 per cent for sports content and 9 per cent for talk & entertainment content.

[144] In Dr. Reitman’s opinion this approach should be adjusted to take into account the operating costs of the Satellite Services, since they reflect their own spending to attract and retain customers. Using projections of future Satellite Services costs, he estimates that operating costs constitute 84.2 per cent of revenues. Therefore, the Shapley approach is used to share the remaining 15.8 per cent of revenues between the five members of the coalition.

[145] Based on the data on listening time and on the reasons for the decision to subscribe, he obtains a value for music ranging from 5.5 to 6 per cent of revenues, using Sirius and XM data respectively. His results also show that the Services should be getting between 88 to 90 per cent, news content about 1 per cent, sports content between 1.1 and 1.3 per cent and talk & entertainment content between 3.1 and 4 per cent. Subtracting music programming costs (3.6 and 5.5 per cent for Sirius and XM, respectively), Dr. Reitman obtains a tariff proposal of 0.5 to 1.9 per cent.

[146] In our opinion, the Shapley approach is interesting by reason that it provides information on the fundamental value of music for Satellite Services. However, it relies heavily on data from a survey where respondents were questioned on hypothetical scenarios. Unfortunately, we do not have enough information to be able to test the variations and the stability of this model. Under these circumstances, we cannot use this approach. The parties could eventually further develop and better utilize it if they agreed on the model and the data collection methodology. This would permit the Board to analyse and validate the numbers.

i. Cannibalization

[147] Drs. Agrawal and McHale argue that the presence of music on satellite radio has consequences on music revenues in other markets, notably CD sales and music downloads. In their opinion, if it were not for the compulsory licence, the Collectives would not accept to licence the use of music if the revenues resulting from the Satellite Services were inferior to the loss of revenues in other markets. Thus, the minimum tariff the Collectives would accept without the compulsory licence is a rate that would offset lost revenues in other markets.

[148] Using data that Circum Network obtained from current subscribers of the Satellite Services about the impact of their subscription on their purchase of CDs and music downloads, Drs. Agrawal and McHale estimate that the equivalent of $2.58 per subscriber-month is lost as a result of the Satellite Services.

[149] Dr. Reitman opines that the surveys do not provide accurate answers in this case. In his opinion, to add a cannibalization effect to the tariff constitutes a surcharge, and a fair and equitable tariff does not require such an adjustment.

[150] Dr. Reitman also states that the results from Drs. Agrawal and McHale are substantially higher than other evidence. Using U.S. data and applying the same method as Drs. Agrawal and McHale leads to a cannibalization effect of $1.10 per subscriber-month.

[151] In our opinion, the cannibalization effect is not a useful measure. If it is possible that the Collectives would refuse to be part of the Satellite Services if the revenues were inferior to the resulting losses in other markets, it is also possible that they would accept to be part of it if, for instance, they expected satellite radio to grow stronger in the future. The cannibalization effect, even if we could obtain a reliable measure, should not be added to the tariff, and should not be used as an indication of a minimum value either.

C. Efficiency Approach

[152] In this approach, the tariff is determined as the rate that maximizes the value of music for society. Drs. Agrawal and McHale believe that the establishment of a tariff implies a trade-off between the loss of subscribers arising from higher subscription costs and the increase of music production resulting from increased revenues for the industry. Therefore, the “optimal tariff” is determined by the rate where the incremental dynamic efficiency gain from additional music content offsets the incremental static efficiency loss from a reduced number of subscribers.

[153] In order to evaluate the static efficiency loss, the authors need to obtain an estimate of the elasticity of demand of subscribers. They used data from the Circum Network survey and found that subscribers would collectively accept an average increase of $1.44 and that 73 per cent of them would retain their subscription if the price was increased by one dollar. They also found that, as economic theory predicts, there is a bigger sensitivity to the price of the firm people subscribe to, than to the price of the competing firm.

[154] Drs. Agrawal and McHale need to estimate how music revenue gain is translated into increased music content and how increased music content increases social surplus for society to evaluate the dynamic efficiency gain.

[155] Drs. Agrawal and McHale use data from a 2006 Canadian Heritage publication about annual revenues for the music industry and on production of musical works. Using data from 2001 to 2004, for French and English music, Drs. Agrawal and McHale find that it costs about $55,000 to generate an album. They argue this result is validated by Sellaband, a music corporation that collects money from consumers and holds these “micro-investments” in escrow until an artist raises $50,000 allowing Sellaband to produce an album for that artist.

[156] In addition, they use data from Canadian Heritage to estimate that the average lifetime revenue of a new album is $109,827. Applying basic economic theory, the authors calculate that half of that amount, or $54,914, corresponds to the social surplus. Therefore, every additional dollar of revenue for the music industry generates approximately one additional dollar of dynamic efficiency gain (or social surplus).

[157] Drs. Agrawal and McHale are then able to resolve their model, since they have estimated how a tariff would induce the loss of subscribers and how that same tariff would induce additional music revenues from music sales. The maximization of these two contrasting effects leads them to an optimal tariff of 38 per cent of revenues ($6.47 per month per subscriber). Drs. Agrawal and McHale state that this rate is optimal because it balances the costs and benefits from a tariff and maximizes the social welfare of Canadians.

[158] In Dr. Reitman’s opinion, the model proposed by Drs. Agrawal and McHale is not appropriate since it does not take into account the dynamic loss incurred by an increased tariff. In his opinion, as music becomes a higher cost input, potential new channels or new music services are dissuaded from entering the market.

[159] Dr. Reitman also criticizes Drs. Agrawal and McHale for not including subscriber acquisition costs as a variable cost of operating the satellite services. This unique but important adjustment to the methodology leads Dr. Reitman to an optimal tariff of 5.2 per cent of revenues, compared with 38 per cent for Drs. Agrawal and McHale.

[160] Finally, Dr. Reitman notes that the model does not include language as a variable. He argues that this is a problem since revenues of the English music industry appear to be much higher than the French music industry.

[161] Dr. Reitman also finds that adding a language variable leads the model to predict that it would take a much higher level of additional revenue for the industry to lead to a new album, than the $55,000 obtained by Drs. Agrawal and McHale. This would result in an optimal tariff of $0.

[162] The relation between industry revenues and album production is essential to the model since it measures the dynamic efficiency gain resulting from the tariff. In our opinion, having only six observations and not being able to control for relevant variables like language is problematic for such an important component of the model. This problem with the methodology makes the approach very difficult to adopt.

[163] Once again, while we find the conceptual approach interesting, it is essential that it be based on high quality data, leading to results that are reliable and stable. That is not the case in this instance but, as noted, we think there are possibilities for such an approach in the future if there is reliable underlying data.

D. Benchmark Approach

[164] Having rejected the conceptual approaches proposed by the Collectives, we will now turn to how the parties propose to use the benchmark approach to establish the tariffs and then describe our approach to set the tariffs in this case.

i. Collectives

[165] Drs. Agrawal and McHale propose a benchmark methodology that establishes the total royalties that should be paid by Satellite Radio Services, and then proposes rules to divide the total among the three Collectives.

[166] Drs. Agrawal and McHale consider that DPA services are similar in nature to satellite radio services since both are subscription-based models and both focus on narrowly defined musical preferences. As the starting point, they use the rate of 26 per cent the Board used for DPA, before an adjustment for repertoire. A ratio of 1 to 3.2 (reproduction to communication ratio for commercial radio) is then applied to the 26 per cent rate to account for the value of the reproduction rate of both musical works and sound recordings. This leads to a total reproduction rate of about 8.1 per cent, and a total royalty rate of 34.1 per cent.

[167] After having initially applied a further correction to reflect higher value arising from portability offered by satellite radio, which DPA does not offer, Drs. Agrawal and McHale subsequently agreed with Dr. Reitman that the higher value from portability is already reflected in the subscription prices, and that a percentage tariff takes this higher value into consideration.

[168] Drs. Agrawal and McHale then examine possible adjustments that take into account the non-music content of satellite radio. They argue that contrary to DPA, not all satellite channels are music; about 60 per cent are. This in itself would justify a downward adjustment. However, it is compensated by the fact that satellite radio offers more music channels than DPA. To resolve this, they propose an adjustment to reflect the fact that not all subscribers to satellite radio subscribe to the service for music. Drs. Agrawal and McHale use the percentage of subscribers that would stop subscribing if music was removed from satellite radio, i.e., 86 per cent to make the last adjustment which results in an overall royalty rate of 29 per cent.

[169] SOCAN and NRCC argue that based on past decisions of the Board, they should get equal value, and CSI should get about one-third of SOCAN’s rate. If one applies this calculation to the 29 per cent rate calculated by Drs. Agrawal and McHale, it results in a SOCAN (or NRCC) rate of 12.5 per cent and a CSI rate of 4.0 per cent. By adjusting the 12.5 per cent rate to account for the repertoire use (91.19 per cent and 27 per cent of programming time for SOCAN and NRCC, respectively) results in final rates of 11.4 per cent for SOCAN and 3.4 per cent for NRCC.

[170] CSI contends there are two types of reproduction activities involved in satellite radio. First, the reproduction right is used in the operation of the service, which includes reproductions as part of programming activities, the 4 to 6 second buffer, the pause and replay feature and the Internet streaming services. Second, reproductions are also initiated by the subscribers (and authorized by the service).

[171] Mr. Audley and Dr. Hyatt argue on behalf of CSI that neither the commercial radio nor the DPA tariff can be used as a proxy for the tariff, because neither of them includes a value for the authorization right. They argue the best proxy is the rate certified by the Board on limited downloads of online music services because of the recording features on some receivers. Even though these recording features are not currently used to a significant degree, what is important is that subscribers have the option of buying receivers with these features and have the option of making these reproductions.

a. Satellite Radio Services

[172] Dr. Reitman agrees with Drs. Agrawal and McHale regarding the adjustment to include the reproduction right in the tariff. He disagrees, however, with the adjustment used by Drs. Agrawal and McHale concerning non-music content, which is based on the percentage of subscribers that would cancel their subscription if music was subtracted from the programming. Dr. Reitman believes that, because the sum of these percentages for all the various content providers is much higher than 100 per cent (172 per cent), the adjustment should be closer to 50 per cent or 40 per cent.

[173] In addition, Dr. Reitman argues that Drs. Agrawal and McHale omitted an important adjustment. DPA services are provided at the wholesale level and if one applies the unadjusted DPA rate to satellite radio services, which are provided at the retail level, this results in royalties far higher than would otherwise be the case by using the DPA benchmark. After applying the appropriate adjustments, Dr. Reitman suggests the appropriate rate should be in the range of 1.6 to 6.5 per cent.

[174] Mr. Goldstein, Sirius’ expert, argues that commercial radio, rather than DPA, should be used as the benchmark. He contends that commercial and satellite radios are close markets by reason that they are competitive substitutes, listeners use both types of radio in similar ways, and the use of music is similar between the two (and more so than between satellite and pay audio). He thus uses the rates of 4.4 per cent (music stations) and 1.5 per cent (low-use stations) as starting points.

[175] Mr. Goldstein makes a first adjustment to account for the higher use of music by the music satellite radio channels. He starts from the music use rate of 94 per cent for music channels, as agreed among the parties. He then revises this rate by including from the list of non-music channels, six additional ones that, according to the percentage of sound recordings in programming time, should be classified as music channels. He obtains a revised estimate of music use by music channels of 90.2 per cent of the programming time. Comparing this percentage to the percentage of music use by commercial radio, 76.1 per cent, leads to an adjusted rate of 5.2 per cent (= 4.4 × 90.2 ÷ 76.1).

[176] Using this rate of 5.2 per cent for the 71 music channels that are part of Sirius offerings, and the rate of 1.5 per cent for the 38 non-music channels, Mr. Goldstein obtains a weighted average rate of 3.9 per cent. A further correction is made to account for the use of SOCAN’s repertoire, at 0.9119. This brings the final SOCAN rate to 3.57 per cent.

[177] The rate for NRCC is obtained by adjusting the 3.9 per cent rate to take into account NRCC’s repertoire use of 0.27. This leads to a NRCC rate of 1.06 per cent. Finally, for CSI, Mr. Goldstein first takes a proportion of 1 to 3.2 of the SOCAN rate of 3.9 per cent and then applies a repertoire adjustment of 0.705. This leads to a final CSI rate of 0.86 per cent.

[178] Mr. Goldstein argues that these rates should be applied to revenues from programming activities and that operations similar to broadcast distribution undertakings should not generate royalties to rights owners. While he first estimated that BDU-like activities constituted about 50 per cent of all satellite radio activities, additional analysis brought him to conclude that programming activities represented only 30 per cent of all revenues.

[179] Finally, Ms. Sherry Kerr, Vice-President and General Counsel of Sirius Canada Inc., filed a report that proposes a benchmark approach to set the CSI rate for the reproduction right. Ms. Kerr analysed the various types of reproduction activities that satellite radio undertakes.

[180] With regard to reproductions in the programming of the service, Ms. Kerr assumes that Canadian satellite radio and commercial radio stations make a similar use. Thus, the proxy to apply to this type of reproduction is the CSI rate for commercial radio. She proposes to adjust this rate to take into account the higher music use of the music channels. She applies a further reduction of 50 per cent to account for the wholesale/retail difference. This leads to a rate of 0.47 per cent on gross income above the first $1.25 million (0.16 and 0.31 per cent of the first two tranches of $625,000). In addition, she proposes that the rate apply only to Canadian produced channels and that the revenue base consist of gross income of the satellite operator multiplied by the listening share of the channel.

[181] For extended buffer and radio replay functions, Ms. Kerr starts from the 5.2 per cent rate obtained by Mr. Goldstein. She also uses Mr. Goldstein’s initial assumption that only 50 per cent of the rate base should be subjected to royalties. Hence, her starting point is a rate of 2.6 per cent. Ms. Kerr uses the rates certified by the Board for Internet streaming as an indication of the ratio to use to obtain a reproduction rate. The rate certified for reproduction is 4.6, which represents 38 per cent of the total (of communication and reproduction rate) of 12.2 per cent. The adjusted rate is thus 0.988 (= 2.6 per cent × 0.38).

[182] Ms. Kerr makes two further adjustments to take into account the share of music in the total value, estimated by Dr. Reitman to be 55 per cent, and the use of CSI’s repertoire. For the latter, she adjusts for the difference between the use of the repertoire by satellite radio (70 per cent) and commercial radio (80 per cent). The final rate is 0.4755 per cent (= 0.988 × 0.55 × (70 ÷ 80)).

[183] For reproductions initiated by subscribers, Ms. Kerr argues that the appropriate proxy is limited downloads rather than streaming. The rates certified by the Board for limited downloads are 6.3 per cent for communication and 5.9 per cent for reproduction. The adjusted rate is thus 1.3 per cent (= 2.6 × 5.9 ÷ 12.2).

[184] Ms. Kerr further distinguishes between two types of reproductions initiated by subscribers: the storing of blocks of programming, and of individual songs. For the former, Ms. Kerr suggests making the same adjustments as before, resulting in a final rate of 0.626 per cent (= 1.3 per cent × 0.55 × 70 ÷ 80). For the latter, Ms. Kerr suggests that no downward adjustment be applied since subscribers may only store individual songs. However, she adds that given the uncertainty as to how many reproductions are being made, the Board might still consider a discount. Thus, for individual songs, she suggests a rate of 1.14 per cent (= 1.3 × 70 ÷ 80).

b. Approach Used [78]

[185] We agree with the Collectives’ use of DPA as the proxy subject to certain adjustments. The mix of satellite radio programming that brings together “music” and “non-music” channels has some similarity to the programming offered by commercial radio, but the music channel component of the satellite radio service is much closer to DPA in terms of the prevalence, importance and use of music. Satellite music channels are based on strong genre segmentation, a feature that is very similar to DPA. And finally, both business models are subscriber-based, even though DPA services are not directly “observable” by the consumers. Therefore, we will use DPA as the proxy for the rates to be established for the various collectives.

SOCAN

[186] Our starting point for SOCAN is the rate established by the Board for DPA, that is, 13 per cent – with a number of adjustments. First, we must adjust for the fact that the DPA rate was set at the “wholesale” level, while the rate for satellite radio is at the “retail” level.

[187] Dr. Reitman proposes to use the payments made to content providers to calculate the appropriate tariff. In our opinion, these payments underestimate the true “wholesale value” of the music programming. The affiliation revenues DPA services receive from cable operators pay for a variety of expenses of which programming is only one. DPA services incur other expenses: technical resources, sales, promotion and administration services. The equivalent expenses incurred by the Services are higher than just the payment for content. In addition, the main business of the content providers is not selling programming to the Services. In our opinion, this is a reason to find that the prices negotiated with the Services are less than their true competitive value.

[188] Mr. Goldstein examined the various activities performed by Satellite Radio Services and concluded that there are essentially two functions: the first is programming; the second is all other, non-programming activities such as the cost of using the satellite, the cost of subsidizing the receivers and the cost of providing customer services.

[189] Mr. Goldstein equates these latter functions to a broadcast distribution undertaking (BDU). He argues that several indicators point to these functions being shared equally in total expenses. He assumes, therefore, that the BDU function represents about 50 per cent of all activities of satellite radio.

[190] Mr. Goldstein later revised his estimate of this proportion to something between 21 and 30 per cent, depending on whether satellite radio is being compared to DPA or to commercial radio. This revision was based on a detailed analysis of the cost structure of satellite radio. Ms. Colette Matteau’s cross-examination of Mr. Goldstein on this issue made it clear that his analysis of similarities between cost categories of satellite radio and DPA or commercial radio is at best very imprecise, and mostly reflected personal judgment. We therefore prefer to rely on Mr. Goldstein’s initial estimate of 50 per cent.

[191] The adjusted rate, after deducting the percentage attributable to the BDU function of the satellite radio service is 6.5 per cent.

[192] Next the proxy must be further adjusted to account for the difference in music use. The parties agree the music channels use music 94 per cent of programming time. The Erin Research study found that non-music channels use music 22.7 per cent of programming time. However, as pointed out by Mr. Goldstein, many of the “non-music” channels use music more than 20 per cent of programming time, the Board’s usual definition of low use.

[193] We use the Erin Research results to include as low-use channels, the ones that use music less than 20 per cent of their programming time. Of the “non-music” channels identified by Erin Research, 31 used music more than 20 per cent of program time. The average use was 36.5 per cent. This affects the average use of music for music channels. The result is that the average rate of music use for music channels is 83.2 per cent. The remaining non-music channels used music an average of 10.7 per cent. This results in increasing the percentage of music channels to 72 per cent of the total. The remainder, 28 per cent, represents the non-music channels.

[194] Since DPA services use music 100 per cent of program time, the 6.5 per cent rate obtained before must be adjusted by taking a proportion of 0.832. This results in a rate of 5.41 per cent.

[195] Finally, we must account for the amount of music used by the Satellite Radio Services that is in SOCAN’s repertoire. In its statement of case, SOCAN assumed that the equivalent of 4 out of 134 music channels used music in the public domain. This translates into 97 per cent of music that is part of SOCAN’s repertoire. The Satellite Radio Services do not contest this number and use it in their proposed rate calculations. The rate to apply to the music channels, adjusted for SOCAN’s repertoire, is thus 5.25 per cent.

[196] We now deal with low-use channels, which have no equivalent in DPA services programming. We will use commercial radio as the proxy, where the rate for low-use stations is 1.5 per cent. This rate needs to be adjusted to reflect the higher value of music that was used for the main rate. The use of DPA as the proxy for the main rate results in an overall value of music that is higher than for commercial radio (i.e., 5.25 per cent as opposed to an effective rate of 4.2 per cent). This higher value is the result of two factors: a higher value for music, and a higher use of music. Only the former must be applied to the low use rate.

[197] If we apply a rate of music use of 76.1 per cent (rather than 83.2 per cent) to adjust for music use, the final rate for SOCAN would be 4.8 per cent for music channels. The difference between 4.8 per cent and 4.2 per cent, is 14 per cent and reflects the difference in the value of music. If we apply this calculation to the low-use rate of 1.5 per cent, the adjusted rate is 1.71 per cent.

[198] There is no evidence that the Satellite Radio Services have the capability to allocate portions of their total revenues to music or non-music channels. These services are sold as a package for a single subscription fee. Thus, we certify a single rate of 4.26 per cent, which constitutes the weighted average of the rate set above for music channels, 5.25 per cent, and the rate set for low-use channels, 1.71 per cent.

NRCC

[199] The parties agree that 27 per cent of the music used by the satellite music channels consists of NRCC’s repertoire. The SOCAN blended rate of 4.26 per cent was adjusted to take into account the small percentage of the music that is not part of SOCAN’s repertoire. The uncorrected rate is 4.39 per cent (= 4.26 ÷ 0.97) which, when multiplied by 27 per cent produces a blended rate of 1.18 per cent, which is the rate we certify for NRCC.

CSI

[200] The approach proposed by Mr. Goldstein with respect to the CSI rate is not acceptable. Mr. Goldstein uses a ratio of 1 to 3.2 derived from the rates prevailing for commercial radio when the evidence demonstrates that the reproduction activities are much more important to the Satellite Radio Services than they are for commercial radio. Commercial radio does not have buffers, radio replays or reproductions initiated by users. These additional functions of the Satellite Services have to be taken into account in calculating the rate for CSI. The commercial radio proxy is simply inadequate.

[201] We also reject the proposal by Mr. Audley and Dr. Hyatt to use the rate set for CSI for limited downloads by online music services. The “limited downloads” services offered by the Satellite Services are available to about 3 per cent of all satellite subscribers. Mr. Audley and Dr. Hyatt contend that the tariff rate should reflect the value of this option, and every subscriber has the option of using this service. We do not agree. The percentage use must be part of the equation when setting a rate. For instance, a blanket licence provides broadcasters with the option of using music 100 per cent of program time, yet we adjust the rate to take into account the actual amount of music use. The same principle must apply to the limited download services of the Satellite Services.

[202] The overall approach proposed by Ms. Kerr is useful and reasonable, with some adjustments. For example, we agree that subscribers who use a more sophisticated receiver should attract higher royalties. However, we would limit the number of categories to three instead of the proposed four. Satellite Services will pay only for programming copies if a subscriber’s receiver cannot make copies. Receivers equipped with extended buffer and replay functions will trigger an additional royalty and those that can store blocks of programming or individual tracks yet another one. This approach establishes a direct correlation between the rate and the receiver’s functionality (something the services already know), rather than with the actual use made of each receiver (something the services cannot know).

[203] We agree with Ms. Kerr that the reproductions involved in the programming operations are similar to those made by commercial radio stations. For these specific reproductions, we could then simply apply the ratio of the reproduction to the communication right of commercial radio (1 ÷ 3.2, or 0.3125) to the SOCAN uncorrected blended rate of 4.39 per cent. This produces a rate of 1.33 per cent. However, the rate must reflect the fact that most programming copies are made outside the country and are not subject to the Act. Ten per cent of the channels each service offers are programmed in or from Canada. Therefore, a first adjustment for programming copies yields a rate of 0.13 per cent (= 1.33 per cent × 0.10).

[204] A further adjustment is required. There are two types of Canadian programming copies: assessment copies that Canadian programmers make when they upload a DMDS file onto their server, and “play” copies that Canadian programmers make onto a server to be played when the software so instructs the server. The adjusted rate of 0.14 per cent accounts for both types of copies. CSR never makes “play” copies in Canada; consequently, it should pay less for its programming copies. Nothing on the record would allow us to assess precisely the relative value of both types of programming copies. Yet it is clear that assessment copies are much less valuable than play copies, since assessment copies are temporary and concern only a fraction of all songs that programmers assess (no assessment copies are made of songs delivered on CD). Under the circumstances, we set at 95 per cent the discount for a service when no work is transmitted to subscribers using copies on a server located in Canada.

[205] For the extended buffer and the replay functions, Ms. Kerr suggests that the relative value of the reproduction and the communication right should be the same as the value the Board set for on-demand streaming by online music services. We agree since in both cases, the reproduction activities that are taking place are similar in nature. The rates set by the Board for on-demand streaming by online music services are 4.6 per cent for reproduction and 7.6 per cent for communication, a ratio of 0.61. [79] Applying this ratio to the uncorrected, 4.39 per cent SOCAN rate results in a rate of 2.66 per cent. In our opinion, this rate provides fair remuneration for the type of reproductions under discussion. This rate will apply only if a subscriber’s receiver offers at least one of the relevant features. At the time of the hearings, that proportion was about 30 per cent.

[206] Finally, there is the rate for the reproductions initiated by the subscribers. These can either take the form of storage of blocks of programming or of individual songs. We will follow the same approach used above but rely on the rates certified for reproduction and communication of limited downloads by online music services (5.9 per cent and 6.3 per cent, respectively) to establish the relevant rate for these reproduction activities. By applying a ratio of 0.94 (= 5.9 ÷ 6.3) to the SOCAN rate of 4.39 per cent, we arrive at a reproduction rate of 4.11 per cent. Again, this rate will apply only if a subscriber is able to make such reproductions. At the time of the hearing, about 3 per cent of subscribers could.

[207] A final adjustment is needed to account for the use of CSI’s repertoire by the Satellite Services. The use of the repertoire can be calculated using the number of copies on the server, the number of plays or listening patterns. The objectors propose the number of plays; the parties agree that this corresponds to a repertoire use of 70.5 per cent. CSI, relying on Mr. Audley and Dr. Hyatt, would rather base their calculations on listening patterns and proposes 76.9 per cent. In our opinion, listening patterns should not in general be used in the calculation of music use. Doing so attributes to music use some measure of value derived from the subscribers’ preferences. The number of plays is the relevant measure in this instance. Other measures that relate to value or the subscribers’ preferences should, at least implicitly, be reflected in the calculation of the value of the music, from which the rate is obtained. Adjusting this rate by a measure of repertoire use, itself at least partly based on value, would amount to some form of double counting.

[208] We do not have reliable evidence on the music use differences by types of reproduction and will therefore assume that the average music use of 70.5 per cent applies to all types of reproductions. This leads to final rates of 0.10 (or 0.005 when no work is transmitted to subscribers using copies on a server located in Canada), 1.87 and 2.90 per cent.

V. RATE BASE

[209] The Services want to pay royalties only on the share of subscription fees they actually receive and not on advertising revenues, receiver sales, paid commissions or activation and termination fees. They would exclude from the rate base any form of revenues that they do not currently collect such as membership, product placement, commissions and the like. They would allow the deduction of subscriber acquisition costs.

[210] The Collectives would include in the rate base anything that is paid for the service, irrespective of who pays or collects it, as well as all other possible sources of income. In their view, excluding revenues encourages the Services to inflate those revenues so as to reduce royalty payments.

[211] The Board is of the view that the rate base should include almost all forms of revenues generated by the service. We have already applied a deduction of 50 per cent to the rate to account for the fact that the DPA rate which is used as the proxy was set at the wholesale level while the rate being set here is at the retail level. In other words, this deduction amounts to a reduction of 50 per cent of all “retail income” to obtain an equivalent “wholesale income”. Applying an additional reduction to the rate base to deduct some forms of revenue, including advertising revenues, could amount to a double deduction.

[212] Consequently, the rate base shall include everything that is paid for the service. This includes activation, termination and access fees, which are often used to artificially lower the monthly subscription cost. It also includes most other potential forms of income, as long as it is reasonably possible that these might emerge in the market. The base will not exclude subscriber acquisition costs as those simply are costs of doing business, but will exclude revenues from the sale of receivers (since no other group of users pay royalties on the hardware subscribers require to receive the service) and agency commissions.

[213] It would appear however that the issue of defining the rate base is, at this time, largely theoretical. Advertising and other income currently represent a very small share of overall revenues.

VI. MINIMUM FEES

[214] While SOCAN did not request a minimum fee, both NRCC and CSI did in their proposed tariffs. NRCC proposed a minimum fee of $1.50 per subscriber per month while CSI proposed $0.50 per subscriber per month for services that do not authorize subscribers to reproduce musical works and $1 per subscriber per month for services that provide such authorization to subscribers.

[215] NRCC did not propose a minimum fee in its statement of case. CSI was the only collective society that presented evidence supporting minimum fees for this tariff. CSI based its proposed minimum on two-thirds of the $15 per month price of a subscription. Applying its proposed rate of 2.9 per cent for 2006-2007 and of 5.8 per cent for 2008 and 2009 yielded proposed minimums of $0.29 and $0.58 per subscriber per month, respectively.

[216] While the Objectors argued that this tariff did not justify any minimum fees because there are only two services and that no exemption at a lower revenue threshold is proposed, CSI stated that its request for a minimum payment is both reasonable and necessary to ensure fairness to the right holders it represents because there is evidence in the record of the use of promotional subscriptions, as well as evidence of so-called lifetime subscriptions, which, unless a minimum payment is specified, could continue for a lengthy period without payments for music.

[217] We agree with CSI and certify minimum fees per subscriber per month. Using the methodology proposed by CSI results in minimums of $0.43 for SOCAN, $0.12 for NRCC and $0.01 (or $0.0005 when no work is transmitted to subscribers using copies on a server located in Canada), $0.19 or $0.29 for CSI, depending on the type of reproductions a subscriber’s receiver does.

[218] SOCAN did not ask for a minimum. This does not prevent us from setting one. The Board can alter a tariff’s terms and conditions in any reasonable fashion. Moreover, the tariff we certify is much less than what SOCAN asked for; consequently, there is little risk that adding a minimum will result in any service being required to pay more than if we had certified SOCAN’s proposal without alterations.

VII. TOTAL ROYALTIES AND ABILITY TO PAY

[219] The Satellite Services have not yet made a profit from their operations. This is not unexpected in an industry which is in a start-up phase, and has large upfront fixed costs. However, the evidence clearly indicates a financial situation that is more difficult than for many other industries. The Satellite Services have incurred heavy losses since the beginning of their operations. The upfront fixed costs, which are estimated to be about 85 per cent of total operating costs, contribute to these difficulties in the short term. In addition, the Satellite Services are incurring very large subscriber acquisition costs. As a result, losses by both services at this time are many times higher than subscription revenues.

[220] As the number of satellite radio subscribers increases, the financial situation of the industry is expected to improve significantly. Indeed, the Satellite Services expect to be profitable by about 2010. That is however beyond the period of the tariffs being certified here.

[221] We believe that, as we have done in the past for other tariffs, an initial discount should be applied to these tariffs. However, for the reasons just mentioned, we believe that a larger discount than usual should be applied. We will thus apply a 25 per cent discount from 2005 to 2007, the latter being the first year where the three tariffs apply together. A discount of 10 per cent will apply for 2008 and 2009, after which we believe no discounts should be applied. Hence, we certify the full NRCC rate for 2010.

[222] The attached table shows the full rates and the rates we certify, reflecting the phase-in discounts for each of the Collectives. This table takes into account the different years for which individual Collectives’ tariffs are being certified.

[223] The rates we certify in this decision will generate about $1 million in royalties for all three Collectives in 2006, the first year for which data on revenues were available. However, at the time of the hearing, the Satellite Radio Services were expecting rapid growth such that in 2009, revenues could be multiplied by a factor of 10 compared to 2006, leading to a similar increase in royalties.

[224] The total rate that will apply when the full tariff is in effect will vary between 5.54 and 10.31 per cent. Were the proportion of receivers having the advanced reproduction functionalities to remain the same as in 2006, the effective total rate would be 6.2 per cent. For comparison purposes, it is useful to note that commercial radio stations pay 7.1 per cent for similar rights; in addition, satellite radio clears all rights associated with the functionality that makes it so different. The services U.S. counterparts currently pay 6.5 per cent only for the right to broadcast sound recordings; that rate is set to increase to 8 per cent in 2012. [80]

IX. TARIFF WORDING

[225] The following comments may help the reader to better understand the wording of the tariff. As is now the rule with any tariff of first impression, we consulted the parties on this matter before reaching a final decision.

A. Application Periods

[226] SOCAN filed five proposed tariffs for 2005 to 2009, NRCC one for 2007 to 2010 and CSI two for 2006 to 2009. We cannot certify a tariff for years in respect of which nothing was filed. Smaller collectives have clearly stated in the past that they can only afford to file proposed tariffs for periods of several years and that other collectives should not be allowed to dictate how long certified tariffs last by filing for shorter periods of time. Anything that helps to reduce the costs associated with getting a tariff certified should be encouraged. Consequently, the tariff will deal with all the years for which the Board was seized at the time of issuing the decision. The fact that this results in staggered application periods should not raise any significant difficulties.

B. Dealing With Subscribers Who Receive the Service For Free

[227] Does the notion of subscription imply that a payment is made for what is the object of the subscription [81] or do users commonly “subscribe” to a variety of services for free in the virtual and material worlds? [82] We are being asked to deal with this apparent contradiction.

[228] The Collectives do not want the notion of subscriber to be limited to those who provide valuable consideration. According to CSI, the Board’s statement in CSI – Online Music Services that subscription implies payment has created enforcement difficulties, with some arguing that free trial subscriptions do not attract royalties. The Collectives ask that free subscriptions trigger minimum royalties. The Services, on the other hand, maintain that a definition is no more needed than in CSI – Online Music Services and that if the term is defined, it should apply only to paying customers.

[229] Free subscriptions should attract the minimum royalties. Any promotional element involved is overridden by the fact that such a subscription involves a protected use. In this respect, the parallel with permanent downloads is interesting. Previews do not attract royalties, but free downloads attract the minimum fee. In order to avoid any confusion, the tariff so provides explicitly.

[230] CSI’s definition of subscriber requires that valuable consideration be received. This does not prevent us from applying the minimum fee to subscribers who get the service for free. Again, given that the tariff we certify is much less than what CSI asked, there is little risk that adding a minimum fee will result in any service being required to pay more royalties than if we had certified CSI’s proposal without alterations.

C. Counting Subscribers For the Purposes of the Tariff

[231] The manner in which the number of subscribers is determined in Board tariffs is not uniform. In some cases, the issue is not addressed. In others, subscribers are counted on the last day of the month. CSR asked that the number be the average in any given month. The Collectives countered that using averages is unnecessarily complex, adding that a subscriber that joins at the end of the month benefits from the tariff and should trigger the full amount of royalties.

[232] In the long run, this probably is a non-issue. A subscriber’s monthly payment is accounted for only once, irrespective of when it is paid. Therefore, in a stable market, the choice of date for counting subscribers probably matters little, if at all. This, however, is a tariff of first instance in a market where the number of subscribers changes significantly, even month-to-month. Consequently, for the purposes of this first tariff, the number of subscribers used to calculate royalties for any given month shall be the average number of subscribers in the reference month.

D. Use of Services by Commercial Subscribers

[233] CSI asked that the tariff target only direct reception by subscribers for private use and not commercial subscribers or subsequent protected uses of music in a signal, for example as background music in a public place. The assignments that CSI secures from its member publishers do not include the delivery of satellite radio to commercial subscribers. CSR agreed with this proposition. NRCC submitted that unless the tariff targets communications to commercial subscribers, it will not be entitled to collect royalties on account of those communications (as opposed to subsequent uses, to which other tariffs, such as NRCC Tariff 3, would apply). Neither SOCAN nor Sirius expressed any view on the matter.

[234] SOCAN’s proposed tariffs are not limited to private use. Those of CSI are; in any event, we cannot licence uses that a collective does not administer. NRCC’s proposed tariff also is limited to “direct reception by subscribers for their private use”. We cannot expand the ambit of the proposed tariff; by acting as it did, NRCC abandoned, for 2007 to 2010, its remuneration right with respect to the communication of the Satellite Services’ signals to subscribers who clearly receive a signal for non-private uses.

[235] Consequently, the tariff shall licence the Services’ uses of the repertoires only to the extent that it is for the subscriber’s private use. Furthermore, the tariff specifies that commercial subscribers are not to be included in the calculation of royalties: Services should not pay for what they do not get.

[236] This measure should not inconvenience the Services. They probably are free to avail themselves of SOCAN Tariff 16 and NRCC Tariff 3 with respect to commercial subscribers who use their signals as background music. The Services also need to licence separately the programming reproductions of musical works with respect to commercial subscribers.

E. Sound Recording and Musical Work Use Information

[237] The parties agree that music use information should be provided for seven days each month. They make a number of representations on other issues.

[238] CSI wishes to be informed of all music use, not only of sound recording use, so as to pay royalties to rights holders whose music is broadcast live or during live events. It disagrees that the Services should provide information only if available. Providing music use information is part of doing business in the music industry in Canada. The Board should apply its standing principle that if information is essential, users ought to provide it even if it means getting it from someone else.

[239] The Services ask that they provide only what is available to them; monitoring the use of live music is not possible and would not increase the reliability of the data.

[240] NRCC asks that the Services provide the running time of each sound recording. The Services counter that they do not always get that information.

[241] In CSI – Online Music Services, the Board stated that users must supply essential information even if it means getting it from someone else, but added that information would have to be supplied only if available during the application period of the first tariff. [83] In this instance, information will need to be supplied only if available to the Service or to a third party from whom the Service is entitled to get the information. Since only available information will have to be provided, adding a requirement to report on live music or to supply the running time will not add unduly to the Services’ reporting burden. Indeed, it will allow the Board and the Collectives to find out precisely how much information is available.

[242] The Services should expect to be required to supply all of the information that the Collectives need to effect distributions starting with the next tariffs. They should therefore adjust their data collecting practices as soon as possible, and require their programming partners to do the same.

[243] The Services are required to provide music use information for seven consecutive days every month. Other tariffs leave the choice of days to the Collectives. In this instance, the Services ask that the tariff provide that the information is supplied for the last seven days of each month. We grant this request.

[244] The tariff will provide that music use information must be filed in electronic format.

F. Audits

[245] The Services asked that audits be limited to one per year. We will not do so. No other tariff so limits audits. Collectives must be able to re-audit a user if a first audit shows that the user is not complying with the tariff.

[246] A limit of sorts is imposed in joint tariffs by requiring collectives to share audit information. NRCC argued that it may sometimes be inappropriate to do so. We can think of no such scenario.

G. Confidentiality

[247] Until recently, tariffs provided that a collective could share confidential information “if ordered by law or by a court of law”. In this tariff, as in the most recent retransmission tariffs, these words have been changed to “if required by law”. The previous language was unclear. Courts order but the law does not. “Required by law” is more usual English usage and includes a court order.

H. Transitional Provisions

[248] The Services ask for six months to provide the information pertaining to the period from January, 2005 to the publication of the tariff, to allow them to adapt their computer systems. This is a new tariff with significant retrospective effect, targeting only two companies, both of which must get the information from third parties. Data gathering will take time, especially with respect to the number of each receiver model. Sirius adds that it may not have the receiver activation information for the whole tariff application period.

[249] The 60 to 90 day period the Collectives are proposing is too short. The Services will have roughly four months from the certification of the tariff to report information and pay royalties relating to past periods.

[250] We are as surprised as CSI to hear that Sirius might find it difficult to specify the type of receiver used by each subscriber, given that this formula was advanced in the testimony of its own General Counsel, and given the assurances that Sirius had the data to make the distinction. Still, we have to account for that possibility by providing for the use of alternative information where actual data is unavailable. CSI’s proposal that the highest rate apply is unreasonable, given that the more sophisticated models occupied only a modest share of the market at the time of the hearings. Instead, the tariff will provide that if the number of subscribers using a particular model is not available for a month, royalties are calculated using the same information for the next following month for which the information is available. Receivers get more sophisticated with time, thereby attracting higher royalties. A later number necessarily will overestimate the number of more sophisticated receivers for a previous month. That should be encouragement enough for the Services to find and provide the relevant information.

[251] As has become our common practice for new tariffs, a table sets out multiplying factors to be used on sums owed, derived using the previous month-end Bank Rate. Interest is not compounded. The amount owed for a month is the amount of the approved tariff multiplied by the factor set out for that period.

Claude Majeau

Secretary General

TABLE / TABLEAU

Full and Certified Rates (as a Percentage of Total Revenues)
Pleins taux et taux homologués (en pourcentage des revenus totaux)

Full and Certified Rates (as a Percentage of Total Revenues)

	Full rates/ Pleins taux	Certified rates/Taux homologués
		2005	2006	2007	2008	2009	2010
SOCAN	4.26%	3.19%	3.19%	3.19%	3.83%	3.83%
NRCC/SCGDV	1.18%	-	-	0.89%	1.07%	1.07%	1.18%
CSI
Programming (with play copies)/Programmation (avec copies de diffusion)*	0.10%	-	0.07%	0.07%	0.09%	0.09%
Extended buffer and replay/Tampon prolongé et écoute différée	1.87%	-	1.40%	1.40%	1.69%	1.69%
Storing individual songs and block programming/Stockage de pistes individuelles et de blocs de programmation	2.90%	-	2.17%	2.17%	2.61%	2.61%
TOTAL Receiver with no copying functionality (with play copies)/Appareil ne pouvant copier (avec copies de diffusion)	5.54%	3.19%	3.26%	4.15%	4.99%	4.99%
Receiver with extended buffer and replay/Appareil avec tampon prolongé et écoute différée	7.41%	3.19%	4.66%	5.55%	6.68%	6.68%
MP3-like receiver/Appareil de type MP3**	10.31%	3.19%	6.83%	7.72%	9.29%	9.29%
TOTAL (Average/Moyen)***	6.19%	3.19%	3.75%	4.64%	5.58%	5.58%

* A 95% discount applies to this rate when no play copies are being made/Un escompte de 95 % s’applique à ce taux lorsque aucune copie de diffusion n’est effectuée.

** Assuming that the receiver is also enabled for extended buffer and replay/Si l’appareil possède aussi les fonctions de tampon prolongé et d’écoute différée.

*** Assuming that 30% of subscribers have receivers with buffer and replay functions, and 3% have MP3-like receivers/Si 30 % des abonnés possèdent un appareil équipé des fonctions tampon et écoute différée, et 3 % possèdent un appareil de type MP3.