`Apple v Sightsound Technologies
`CBM2013-00023
`Page 00001
`
`
`
`FILE SHARING: CREATIVE DESTRUCTION
`
`OR JUST PLAIN DESTRUCTION?*
`
`STAN J. LIEBOWITZ
`University of Texas at Dallas
`
`ABSTRACT
`
`The sharing of sound recordings over the Internet is the newest controversy in a
`long-running battle between copyright owners and copying technologies. In order to
`provide some context, perspective, and background, this paper examines the short
`history of file sharing, the longer history of record sales, various explanations for
`the change in record sales, and some analysis of the economics of copying. Although
`file sharing has been imperfectly and inconsistently measured, it nevertheless appears
`to reveal a fairly close linkage between changes in file sharing and changes in record
`sales. Explanations, other than file sharing, for the recent decline in record sales seem
`to have little or no support. Because economic theories of the impacts of copying
`hold out little hope for a benign impact of file sharing, these results should not be
`surprising. These findings reinforce the econometric results from most of an ex-
`panding literature.
`
`I.
`
`INTRODUCTION
`
`TEN years ago the term “file sharing” was unknown. Then Napster arrived,
`and both file sharing and Napster quickly became etched into the public’s
`consciousness. Although Napster was effectively shut down as an unau-
`thorized file-sharing service within 2 years of its birth, its progeny live on,
`as do new habits of music listeners. These dramatic changes have given us
`the now familiar additions to the lexicon such as “ripping” files from CDs,
`listening to MP3s on iPods, and, of course, downloading files online using
`programs such as Kazaa or Grokster.
`Yet the file-sharing saga and the controversy surrounding it might appear
`a mere replaying of a narrative we have encountered several times before.
`The photocopier, introduced by Xerox in 1959, allowed individuals to cheaply
`and conveniently copy printed pages. Audiotaping, which became popular
`in the 1970s, made it easy and inexpensive for individuals to copy sound
`
`* I would like to thank Steve Margolis, seminar participants at the University of Montreal,
`and, for financial support, the Center for the Analysis of Property Rights and Innovation. I
`would also like to thank Russ Crupnik from NPD, Eric Garland from BigChampagne, and
`Graham Mudd from comScore Media Metrix for their generous help in providing data and
`insight.
`
`[Journal of law and Economics, vol. XLIX (April 2006)]
`© 2006 by The University of Chicago, All rights reserved. 0022-2186/2006/4901-0001$01.50
`
`1
`
`This content downloaded from 198.65.204.101 on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00002
`
`Page 00002
`
`
`
`2
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`recordings, with dual cassette decks intended for high-speed copying becom-
`ing commonplace. Videotaping, which became popular in the 19805, allowed
`individuals to copy broadcasts and prerecorded movies.
`The copyright industries reacted negatively when each of these copying
`technologies appeared. The publishing industry complained about photocop-
`ying, although an analysis by Liebowitz (1985) concluded that photocopying
`was beneficial to the industry. The movie and television industries brought
`suit to stop the video recorder, but after the Supreme Court’s ruling (Sony
`Corp. ofAmerica v. Universal City Studios, 464 U.S. 417 [1984]) went against
`these industries, a new market emerged—prerecorded video—which, al-
`though largely unanticipated, now provides the movie industry with revenues
`far in excess of box office revenues.1
`
`The sound-recording industry had an equally negative response to copying
`technology. No less a luminary than Alan Greenspan, prior to his becoming
`chairman of the Federal Reserve Bank, stated, “At present .
`.
`. severe eco-
`nomic damage [is being done] to the property rights of owners of copyrights
`in sound recordings and musical compositions.
`.
`.
`. [U]nder present and
`emerging conditions, the industry simply has no out. .
`.
`. Unless something
`meaningful is done to respond to the .
`.
`. problem, the industry itself is at
`risk” (U.S. Senate 1983). Sales of sound recordings began a decade-long
`expansion not long after this testimony (ending a 4-year decline), once again
`making the claims of concern by the copyright industry appear unwarranted.
`Nevertheless, as explained in Section IV, the role of these older copying
`technologies as economic precedents is limited both because there are im-
`portant differences between file sharing and these prior copying technologies
`and because the impacts of these technologies were not so clear-cut.
`Each of the previous copying technologies brought forth some work by
`economists on the economics of copying, although the focus was more on
`theory than empirics (for surveys, see Varian 2005; Watt 2004). Among the
`problems with conducting empirical analyses of these older technologies was
`the difficulty in measuring the extent of unauthorized copying. One of the
`incidental benefits of the new digital copying technology should be to provide
`better data, although a great deal of imprecision remains.
`This current copying technology of file sharing was personified by Shawn
`Fanning, who created Napster with the purpose of allowing music files to
`be shared among strangers. Napster began operations in mid-1999 and quickly
`rose to international prominence. The sound-recording industry experienced
`a dramatic swoon in sales beginning the next year, continuing unabated (with
`one informative exception) through 2005. The industry has blamed this sales
`decline on the rapid growth of file sharing and, in an attempt to stem the
`decline, has sued thousands of individuals heavily engaged in file sharing
`
`‘ See Liebowitz (2004a), where prerecorded video revenues were estimated to be twice as
`high as revenues from theatrical showings.
`
`This content downloaded from 198.65.204.101 on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00003
`
`Page 00003
`
`
`
`FILE SHARING
`
`3
`
`(as well as suing the file-sharing services).2 These lawsuits have led to a
`heated debate, often uninformed by facts. We are fortunate to have in this
`issue a detailed study by Bhattacharjee and colleagues (2006) that examines
`the impacts of those lawsuits on the amount of file sharing.
`File sharing has generated far more attention than earlier copying tech-
`nologies.3 It is easy to dismiss the intense media coverage, as evidenced by
`Time magazine’s putting Fanning on its cover (October 2, 2000), as just
`another case of romanticizing the impact of a new technology. Yet Napster
`truly began a revolution in music listening with a still uncertain impact on
`the entire economic model that has been used by the sound-recording industry
`for much of the last century. Copyright industries also tend to attract more
`attention than might appear warranted by their share of gross domestic prod-
`uct (GDP) alone.4 This is most likely due to the fact that consuming these
`products occupies a majority of the time that individuals spend on leisure
`activities, with the average American watching 4.5 hours of television and
`listening to more than 3 hours of music each day (US. Census Bureau 2003).
`Thus the impact of these industries on the collective consciousness is very
`large.
`Naturally, the current concern over the impacts of file sharing brought
`forth among economists a renewed interest in the economic consequences
`of copying. Recent econometric studies include, but are not limited to, the
`two very fine empirical examinations found in this issue—one by Rafael
`Rob and Joel Waldfogel (2006) and the other by Alejandro Zentner (2006;
`see also Blackburn 2004; Hong 2004; Michel 2005; Oberholzer—Gee and
`Strumpf 2005; Peitz and Waelbroeck 2004; Zentner 2005). Although neither
`of these two papers attempts to measure the impact of file sharing on the
`full U.S. sound-recording market, which is my focus in this paper, a recent
`econometric study (Liebowitz 2006) examines that specific question. All of
`the papers of which I am aware, except one, find that file sharing brings
`about some degree of harm to copyright owners.S
`
`1 According to the Recording Industry Association of America (RIAA) Web site, the re-
`cording industry announced plans to bring lawsuits against file sharers on June 25, 2003 (RIAA
`2003c). On September 8, 2003, the RIAA brought what they referred to as the “first wave”
`of lawsuits against 261 individuals (RIAA 2003b).
`3 Besides the economic studies discussed in this paper, there are papers of a more philo-
`sophical legal bent, particularly from some very vocal critics who have voiced their unhappiness
`with copyright law and the entertainment industry. These copyright critics, sometimes asso-
`ciated with the concept of the “creative commons” and the Electronic Frontier Foundation,
`argue that copyright laws are being used by the sound recording, movie, and software industries
`to thwart innovative forces that would otherwise open up the market to new competition. See,
`for example, Lessig (2004).
`4 It is estimated to be between 5 and 7 percent of the gross domestic product (GDP), according
`to a report produced for a copyright trade association (Siwek 2002).
`5 The one paper that does not find file sharing to harm record sales is Oberholzer—Gee and
`Strumpf (2005). Of the other papers, which cover different countries, different time periods,
`and different approaches, some find results that could be classified as consistent with the
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00004
`
`Page 00004
`
`
`
`4
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`Econometric studies are usually self-contained, but they do not, or should
`not, occur in a vacuum. In order to make an informed judgment about the
`impact of file sharing it is useful to understand the industry background. This
`should include examining the nature and size of file sharing as best we know
`it and investigating the history of the sound-recording market to gain some
`perspective on the current decline. It is also useful to examine the economic
`theories of copying that have been suggested. Finally, no judgment would
`be complete without examining the range of possible sources of evidence,
`particularly sources that might not be amenable to inclusion in econometric
`studies. This paper attempts to perform some of these tasks.
`
`II.
`
`THE BRIEF HISTORY OF FILE SHARING AND ITs MEASUREMENT
`
`File sharing, simply put, allows one computer on the Internet to search
`for and access files on the hard drives of other computers that are connected
`to the Internet. Any individual on a file-sharing network can make available
`any file on his or her hard drive to all other members of the file-sharing
`network.
`
`The term “file sharing” is actually something of a misnomer, however.
`Individuals do not “share” the files that move back and forth on the Internet.
`
`They do not experience these files together nor are they likely to ever meet
`or even know one another. Nor do they lend or trade the files among one
`another, since the files are not borrowed or given back. A more appropriate
`term might be “anonymous file copying,” since that reflects what actually
`occurs. The end result of file sharing is that individuals who do not own and
`have not purchased a particular song or movie can nevertheless obtain that
`song or movie from unknown third parties.
`Currently, file sharing encompasses sound recordings, films and television
`programs, computer software, various forms of pornography, and other prod-
`ucts that can be digitized. Because music files are easily compressed, rela-
`tively small, very popular, and the primary type of file downloaded, they
`appear to be the best candidate for assessing the impact of file sharing itself.6
`As Internet transmission speeds increase, file sharing is likely to focus more
`
`possibility that the entire decline might be due to file sharing (Blackburn 2004; Liebowitz
`2006; Zentner 2005), while others (Hong 2004) apportion only a part of the decline to file
`sharing, and yet others are difficult to classify in this manner. A detailed discussion of this
`literature can be found in Liebowitz (2005).
`6 IDATE (2003) claimed that the ratio of audio files to video files was 100:1 in an October
`28 report. Note, however, that the IDATE report seems of somewhat questionable value, as
`discussed below. Lyman and Varian (2003) report in their table 8.9 that although shared video
`files took up twice as much hard drive space as shared audio files, audio files were nevertheless
`10 times as common as video files in 2003. An Organisation for Economic Co-operation and
`Development (2004, figure 5) report using data from BigChampagne claimed that the number
`of audio files transferred was only twice the number of video files in 2003. This claim seems
`somewhat implausible (unless most of these are short clips of pornography), given the enormous
`size of movie files.
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00005
`
`Page 00005
`
`
`
`FILE SHARING
`
`5
`
`on full-length movies and computer programs. This threatens or promises,
`depending on your point of view, to do for the movie and software industries
`what it has done for the sound-recording industry, which is one reason it is
`so important to understand its impacts.
`Napster was, for all intents and purposes, shut down by a preliminary
`injunction granted to the recording industry in February 2001 (A&M Records
`v. Napster, 239 F.3d 1004 [9th Cir. 2001]). Into the void stepped numerous
`other file-sharing programs, particularly those that, unlike Napster, were not
`based on a central server. Figure 1 provides an estimate of the number of
`home-based file sharers in the United States during this transition. (Data are
`from Jupiter Media Metrix 2001; comScore Networks 2002.) The legal “vic-
`tory” of the recording industry over Napster is not in evidence in these
`statistics, since the number of file sharers continued its upward trend within
`months of Napster’s shutdown.
`The impact of file sharing is something of a moving target, so we should
`not expect a single-sized impact at all times and all places. When Napster
`first came into existence in 1999, most downloaders would not have had in
`place CD burners, and MP3 players did not yet exist. MP3 files, therefore,
`were not terribly good substitutes for music purchased on a CD since the
`downloaded music was tethered to the computer. Any negative impact of file
`sharing at that time should have been quite small. Since then, devices that
`can play MP3 files have become increasingly popular, CD writers have be-
`come ubiquitous, and Internet speeds have increased as broadband has be—
`come more common, shortening the time needed to download songs .7 Because
`file sharing currently produces files that are much better substitutes for pur-
`chased CDs than was the case at the time of Napster, any negative impact
`of file sharing per shared file should be greater now than it was in, say,
`2000.8
`
`The digital and public nature of file sharing would seem to make it more
`amenable to analysis than prior methods of copying. The reality of file-sharing
`measurements, however, does not yet live up to this promise.
`Although there have been numerous news stories reporting statistics on
`
`7 In June 2002, according to Ipsos/Tempo (Ipsos-Reid 2002), 53 percent of American file
`sharers had CD burners, which was more than twice as high as for the general population as
`a whole. Ipsos/Tempo also reported that the penetration of CD burners for the general population
`increased from 22 percent to 31 percent from the first quarter of 2002 to the first quarter of
`2004 (“Ownership of Digital Music Peripherals Trending” [slide], provided by Matt Klein-
`schmidt, senior research manager, Ipsos). If the growth in penetration for the population of
`downloaders was similar, this would have led to a penetration rate of 75 percent among those
`engaged in file sharing. Nielsen NetRatings reported that Broadband users represented 64
`percent of Internet users in October 2005, up from less than 10 percent at the time of Napster,
`although the growth in broadband appeared to be slowing (data were provided by Kaizad Gotla,
`senior Internet analyst, NeilsenlNetRatings, September 2004).
`8 Nevertheless, they are not quite perfect substitutes. The compression involved with MP3
`files reduces sonic quality, according to audio purists. Also, it takes some time and effort to
`download the files, so they are not quite free.
`
`This content downloaded from 198.65.204.101 on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00006
`
`Page 00006
`
`
`
`
`
`
`
`
`
`
`NapsterDecision
`
`
`
`
`mm
`mm
`Io-osa
`lO-AON
`10-100
`
`10-das
`Io-finv
`Io-Inr
`
`' Io-Imr
`Io-Kuw
`10-Jdv
`10-mw
`10-an
`IO‘WI'
`oo-osa
`
`I 00-A0N
`00-100
`
`oo-des
`oo—Snv
`oo-Inr
`oo-unr
`
`oo-an
`oo-Idv
`OO‘WW
`004195
`
`
`
`FIGURE1.—Americanfilesharersathome(1,000s)
`
`
`
`
`
`0 O O O O
`O O O O O
`O. 0,
`ca 0.
`on
`00
`\O
`‘1' N O
`v—1
`P1 H v—I H
`
`O
`
`mam
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00007
`
`Page 00007
`
`
`
`FILE SHARING
`
`7
`
`file sharing, these reports generally cite the same few sources. At the time
`of Napster, the press quoted Webnoize almost exclusively regarding the num-
`ber of files that were being downloaded (Evangelista 2001). It is not clear
`what methodology the (now defunct) Webnoize used, but it reported that
`2.79 billion files had been transferred in Napster’s peak month (February
`2001) and that by August 2001 the number of files transferred on the four
`leading Napster replacements (FastTrack, Audiogalaxy, iMesh, and Gnutella)
`had reached above 3.05 billion per month (Geralds 2001). To put this in
`perspective, worldwide sales of music amounted to about 3 billion songs per
`month in 2000, so one might conclude from these figures that the number
`of songs being downloaded on file-sharing networks was equivalent to the
`number of songs purchased in the authorized retail market (IFPI 2001).9
`IDATE (2003) claims that worldwide file sharing was four times as large as
`worldwide sales in 2003.
`
`The statistics reported for the American market, the focus of our analysis,
`also come from just a few key sources. At the high end, there are claims
`that up to 60 million Americans have used peer-to-peer networks,10 that
`perhaps as many as 5 billion music files are downloaded by Americans in
`a typical month (18 files for every man, woman, and child!),11 and that perhaps
`60 percent or more of all Internet bandwidth is taken up by file sharing
`(Reuters 2003).12 Although these are among the more striking numbers that
`have been put forward, even the more modest estimates appear less reliable
`than we would like. Before discussing the actual measurements, however, a
`brief discussion of the methodologies is in order.
`There are various methodologies for measuring file sharing. Most count
`the number of participants in file-sharing activities (for example, comScore
`Media Metrix, Nielsen NetRatings, BigChampagne, the Pew Internet, and
`American Life Project). Nielsen and comScore examine the number of users
`of particular file-sharing programs, such as Kazaa or BitTorrent, on the basis
`
`9IFPI (2001) reports 3.5 billion albums per year. If we assume 10 songs per album, this
`works out to 2.91 billion songs per month.
`‘0 The Electronic Frontier Foundation reports that 60 million Americans use file-sharing
`software (File Sharing: It’s Music to Our Ears, at http://www.eff.org/sharel), but it is not clear
`where that estimate comes from.
`
`" According to IDATE (2003), there were either 65 billion audio files downloaded in the
`United States in 2003 or 12 billion in the world, depending on which of two seemingly
`inconsistent statements you wish to believe. The breathless prose goes on to predict that by
`2007 broadband users will download an average of 4,300 audio files per year, which seems
`somewhat fanciful.
`
`‘2 IDATE (2003) reports that “According to virtually all the experts in this field, P2P rep-
`resents on average between 50% and 60% of all broadband traffic during the daytime, and as
`much as 80% to 90% of all night time traffic."
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00008
`
`Page 00008
`
`
`
`8
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`of panels of users.13 Since the number of programs available to users is large
`and growing, data based on the number of users of a particular program have
`become less reliable over time as a measure of aggregate downloads.‘4
`BigChampagne measures the number of users of file-sharing networks.”
`BigChampagne’s method of measuring the activity on file-sharing networks
`is proprietary and therefore difficult to judge.16 The NPD Group is the only
`organization, to my knowledge, that attempts to monitor the number of files
`actually transferred by a panel of users. The Pew Internet and Society Project
`conducts numerous surveys on Internet usage, as do Ipsos/Tempo and For-
`rester. The hearings that led to the preliminary injunction against Napster
`featured dueling surveys offering opposing opinions about the impact of file
`sharing (on the Napster surveys, see Liebowitz 2002, chap. 7).
`Unfortunately, each data source, regardless of data type or methodology,
`suffers from one or more imperfections. All panel-based data sources are
`open to the criticism that the panel might not reflect the user population. For
`example, it is plausible that voluntary Internet panels might underrepresent
`the population of people actively engaged in file sharing, as those users may
`be particularly reluctant to have their computers monitored by third-party
`software.17
`
`‘3 Panel members agree to allow a program to monitor and report their computer usage in
`return for some nominal compensation. These panels range from a few thousand to the hundreds
`of thousands. Alternatively, it is possible to create a panel whose members are unaware of
`being monitored, as Bhattacharjee et al. (2006) have done.
`'4 If file sharers were migrating away from monitored programs toward programs that were
`not monitored, such measurements will undercount users. Companies such as comScore tried
`to update their list of programs to keep up to date, but this always lagged somewhat behind
`the behavior of users. In August 2005 I was told that comScore was no longer reporting these
`measurements because they did not feel they could keep up with all the new programs (Graham
`Mudd, industry analysis manager, comScore Networks, telephone conversation with the author,
`August 1, 2005). The number of users followed by Bhattacharjee et al. (2006) would also
`suffer from this same problem.
`'5 There are often several different software programs that use the same network. These
`networks are not identical to the programs that use these networks. For example, Kazaa,
`Grokster, Kazaa Lite (a competitor to Kazaa), and iMesh all use the FastTrack network, but
`there are also other networks, such as Gnutella and DirectConnect, that are used by other
`programs.
`‘6 For a critique of BigChampagne’s methods, see Lawrence (2003). I have a concern with
`the claimed peak and average values, which seem too close to one another. The peak monthly
`values (measured every few minutes) are only about 60 percent above the average values in
`the last half of 2003 and only about 30 percent higher in the first half of 2004. Eric Garland,
`chief executive officer of BigChampagne, suggests that this small difference between peak and
`average monthly values is due to the fact that many Internet users keep their computers and
`file-sharing software running 24 hours a day (e-mailed correspondence with the author, October
`2004). I find it unlikely that dial-up users (who were in the majority) keep their phone lines
`occupied 24 hours a day, but further analysis would be useful.
`'7 This criticism loses some force from the fact that most file-sharing software (for example,
`Kazaa) includes spyware and adware, which monitor the usage of the computer for vendors
`of various products. This means that many file sharers have already allowed third—party software
`on their computer and in many cases they were aware of it.
`
`This content downloaded from 198.65.204.101 on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00009
`
`Page 00009
`
`
`
`FILE SHARING
`
`TABLE 1
`
`9
`
`PERCENTAGE OF ADULT POPULATION ANSWERING YES To THE QUESTION,
`“Do YOU EVER DOWNLOAD MUSIC FILES ONTO YOUR COMPUTER So
`YOU CAN PLAv THEM AT ANY TIME YOU WANT?”
`
`
`June November May—June February
`July—August August—September October
`2003
`2003
`2004
`2005
`2000
`2001
`2002
`19
`9
`13
`13
`11
`15
`19
`43
`23
`31
`32
`25
`36
`41
`20
`9
`1 1
`13
`1 1
`16
`21
`8
`4
`6
`7
`3
`6
`8
`1
`2
`2
`1
`2
`2
`3
`23
`12
`17
`14
`12
`19
`22
`
`
`10 12 13 16 15 7 9
`
`
`
`
`
`Overall
`18—29
`30—49
`50—64
`65+
`Men
`Women
`
`SOURCE—Pew Internet Project, “Usage over Time" (spreadsheet) (http://www.pewinternet.org/trends/
`UsageOver’l‘imexls).
`
`Metrics based on the number of users alone might fail to capture increases
`or decreases in the number of files exchanged per user. Such changes can
`be large even when the number of individuals using file-sharing software is
`stable, as, for example, when users are shifting to broadband. Finally, to the
`extent that movies, video games, and computer files take up changing shares
`of the file-sharing universe, measures of overall file sharing might not prop-
`erly reflect the downloading of music files.
`Surveys suffer from potential problems where consumers might not know
`the answers to the questions they are being asked or where they might be
`reluctant to tell the truth if that means possibly incriminating themselves or
`(in their minds) inviting a lawsuit. These fears should tend to generate answers
`that understate the extent of file sharing. The survey used by Zentner (2006)
`is based on consumers in Europe, where this controversy has been less
`political and at a time when such lawsuits had not yet come into existence.
`The surveys used by Rob and Waldfogel (2006) were conducted in conditions
`where the respondents knew the identity of those giving the surveys and the
`purpose of the survey. In each of these cases the likelihood of intentionally
`incorrect answers was lessened. Nevertheless, the expectation in each case
`would be that the answers might be somewhat biased toward lowering the
`impact of file sharing. The Pew surveys suffer more seriously from this
`problem, particularly after the lawsuits began in mid-2003.
`With these caveats in place, what do the data tell us about file Sharing?
`Results from the Pew Internet and American Life project indicate that music
`file sharers tend to be young and male and more likely to be poor and less
`educated (see Rainie et a1. 2004, p. 4). Table 1 provides some of their
`statistics.18 These numbers are consistent with a view that file sharing is a
`
`“‘ To construct this table, I multiplied the percentage of respondents answering this question
`in the affirmative, which was asked only to those with Internet access, by the number of
`respondents claiming to have Internet access.
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use Subject to JSTOR Terms and Conditions
`
`Page 00010
`
`Page 00010
`
`
`
`10
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`very popular activity.19 The sharp decline from June 2003 to November 2003
`is also fully consistent with the Bhattacharjee et a1. (2006) finding that the
`initiation of Recording Industry Association of America (RIAA) lawsuits
`reduced file sharing. These numbers appear to be considerably higher than
`the European value of 9 percent reported by Zentner (2006) based on the
`October 2001 European Forrester Research survey, but that survey asked if
`users “regularly” downloaded music, so the two are not strictly comparable.
`A slightly later Forrester survey (Stagia 2002) had numbers closer to, but
`still lower than, the Pew values.20
`What do the other data sources have to say about the number of file sharers?
`ComScore claims that there were 40 million unduplicated users of file-sharing
`software during January 2003, which is in general agreement with the Pew
`numbers. BigChampagne measured the average number of “simultaneous”
`users in January 2003 to be just shy of 4 million users. It is difficult to
`compare BigChampagne numbers with comScore since comScore essentially
`measures, by way of analogy to television and radio, the reach per month,
`which is always a higher number than average audience.
`Do these data allow us to draw any conclusions about the historical trend
`in organized file sharing? Clearly, organized file sharing stood at zero in
`1998. We know from Figure 1 that the important growth did not start until
`2000. The Pew surveys in Table 1 indicate an increase in the popularity of
`file sharing on the order of 70 percent from the summer of 2000 to October
`of 2002, which is smaller than what comScore reports in Figure 1, even
`though the comScore figures only go through February 2002.
`In Figure 2, which allows examination of the more recent trends, each of
`three data sets is normalized so that its starting measurement is 1 in order
`to make the numbers somewhat comparable. Both of the data sets with full-
`year 2003 data indicate a substantial drop during 2003, with comScore (and
`Pew in Table 1) matching these declines directly to the RIAA lawsuits (as
`do Bhattacharjee et a1. 2006), although BigChampagne shows the declines
`beginning before the lawsuits were announced.21 File sharing appears to have
`
`. ” were interpreted in the past tense, such as “Have you
`.
`‘9 If the question “Do you ever .
`ever .
`.
`. ” we would not expect the numbers to ever fall as long as respondents are being
`truthful. The decline that does occur can be taken either as evidence that the question is not
`interpreted as “Have you ever .
`.
`. ” or that some respondents might have begun to lie because
`of fear of prosecution. It is also possible that some respondents might have stopped downloading
`but still answer the question in the affirmative because they interpret the question to mean
`“Have you ever .
`.
`. ?”
`2° A Forrester survey in the second quarter of 2002 indicated that 29 percent of European
`Internet users had ever downloaded music, and since slightly more than half of the population
`used the Internet, this would imply that on the order of 15 percent of the population had
`downloaded music, which is fairly close but still slightly below the American Pew figure.
`2' Pew and comScore produced a joint report directly crediting the lawsuits with decline in
`file sharers (Madden and Lenhart 2004). This result is not without contention. Some have
`argued that the lawsuits have had virtually no impact (Karagiannis et a1. 2004). The Karagiannis
`et al. paper bases its results on a single l-hour observation prior to the lawsuits and a single
`l-hour observation after the lawsuits, with very few controls.
`
`This content downloaded from 198.65.204.10] on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00011
`
`Page 00011
`
`
`
`sun: WW
`9002 'HOJEW
`
`goo: 'ruemqad
`500a ‘mmmr
`1002 'JeqLueoea
`
`7003 'JaqUJBMN
`#002 ”9510100
`
`mug ’Jeqwaldes
`
`#002 WWW
`
`$003 'I‘F‘l‘
`V002 'QU'T
`
`#002 "EN
`
`V00? 11"“?
`1'00? 'HOJEW
`
`#002 “190109;!
`
`moz "WW“?
`5003 'Jaqweaea
`
`€002 'JE'QUJBWN
`
`E002 'JanPCI
`
`€002 'JBCIWBIUSS
`
`£002 'IW‘W
`E002 "1 PP
`£002 'aunr
`
`E002 “PW
`
`8003 'IPdV
`
`8003 WWW
`
`E002 ‘f‘JE’anad
`
`can: - Murmur
`:00: - Jeq-ueneu
`
`zonz - JaqueMN
`zoo: - Jaqono
`
`zone - Jaqwezdes
`
`
`
`
`
`FIGURE2.—Variousmeasurementsoffilesharing
`
`3 E
`
`3%
`
`a?
`
`:3;
`EZ
`
`E §3 E s
`
`, g.
`
`99
`
`$
`
`O
`In
`N
`
`In
`N
`N
`
`0
`o
`N
`
`In
`b
`'-‘
`
`0
`In.
`v-‘
`
`In
`N
`H
`
`C
`o_
`H
`
`In
`If
`O
`
`O
`In
`O
`
`zoo: - lsnfinv
`
`This content downloaded from 198.65.204.101 on Wed, 19 Mar 2014 11:47:21 AM
`All use subject to JSTOR Terms and Conditions
`
`Page 00012
`
`Page 00012
`
`
`
`12
`
`THE JOURNAL OF LAW AND ECONOMICS
`
`increased again in 2004, although BigChampagne shows file sharing returning
`to (and then surpassing) its 2003 peak, whereas the Pew surveys indicate
`that it did not return to its prior peak through 2004.22
`This brings us to measurements of the absolute size of file-sharing activ-
`ities. The dispersion of estimates is nothing short of remarkable. If the Web-
`noize statistics on Napster use had been accurate, and if the United States
`contributed 30 percent of the world usage of Napster, this would have implied
`about
`1 billion files per month downloaded by Americans.23 By way of
`comparison, there were about 800 million albums sold yearly in 2003 and
`2004, which works out to 800 mil