New Internet-based technologies appear tothreaten the ability of copyright owners to collectrevenues for their intellectual creations, as epito-mized by the recent public trials and tribulationsexperienced by Napster. That has resulted in newlegislation against pirating and has given rise tonew technologies to protect intellectual products.Both the new technologies and the counter-tech-nologies that have followed them have attractedattention and analysis, sometimes bordering on theapocalyptic, from competing camps. The basicissue, whether technologies that enhance the abili-ty to create unauthorized copying are destructive tothe principles of copyright, is not a new one, how-ever. Technologies that make it easier to piratecopyrighted materials have undergone economicexamination for over two decades. Prior analysis,and prior experience, has indicated that the previ-ous generations of copying technologies have nothad dire consequences for copyright owners.

This paper examines whether new Internetcopying technologies are likely to be differentfrom prior technologies in their ability to destroythe value of intellectual property rights and con-cludes that they are. It then examines the evi-dence that has been put forward to support aclaim that Napster had a negative impact on the

compact disk industry and concludes that theevidence does not support such a finding. I thenexplain why it is that the negative impacts ofNapster were unlikely to have been felt at the timethese examinations were undertaken.

Finally, the analysis examines the impact of apossible market-based solution to this potentialproblem, based on new anti-piracy technologiesknown as digital rights management. This tech-nology not only promises to make copying hard-er, but also allows the copyright owner to chargetiny micropayments for various degrees of use ofthe product. This extra control by copyright own-ers over the use of the copyrighted material has setoff a firestorm of controversy by individuals con-cerned that traditional “fair use” of a product willdisappear and further claiming that digital rightsmanagement will lead to economic inefficiency.Fair use has historically allowed scholars and oth-ers to use small amounts of copyrighted materialsfor research or study without being obligated tomake copyright payments. I show that digitalrights management, contrary to these claims, doesnot eliminate fair use and is likely to enhance eco-nomic efficiency. Nevertheless, attempts by gov-ernment to force the adoption of anti-copyingtechnology appear misguided.

Policing Pirates in the Networked Ageby Stan Liebowitz

_____________________________________________________________________________________________________

Stan Liebowitz is a professor of economics at the University of Texas at Dallas. This study is based on portionsof his forthcoming book, Rethinking the Networked Economy, to be published by Amacom Press in 2002.

Executive Summary

No. 438 May 15, 2002

Introduction

As the storage and transmission of artisticcreations become increasingly digital, the pri-mary form of property right in such creations,that is, copyright, appears to face unprecedent-ed enforcement challenges. The recent focus ofattention has been on the digitization of audio,the circumvention of copyright in the tradingof digital audio files, and particularly theNapster case. Yet the digitization of any artisticcreation—whether audio, video, or the writtenword—threatens, or appears to threaten, cur-rent copyright regimes, as Napster successorssuch as Bearshare, Limewire, AudioGalaxy, andMorpheus make clear.

It is unclear whether copyright protectioncan continue to provide ample incentive forartistic creation. Will authors be able to appro-priate more or less of their works’ value thanthey have in the past? How does digital storagechange the balance between authorized andunauthorized use? What pricing schemes arelikely to arise? What legal rules strike the bestbalance between consumptive efficiency andproductive efficiency? That is, how do we maxi-mize use and creation at the same time?

For perspective, we should remember thatcopying technologies have been in existencefor several generations. Doom-and-gloomscenarios have been raised before, and copy-right owners have not often suffered greatharm even when they were not given extraprotection from copying technologies.Videocassette recorders (VCRs) are one exam-ple. Audiotaping is another.

This paper argues that the greatest threat tocopyright owners has always come from orga-nized, large-scale unauthorized copying. Thedigitizing of works and the ubiquity of theInternet have brought with them an increasingpotential to organize what otherwise would beunorganized, making pirating cheaper, easier,and more widespread than ever before. This iswhat makes the current copying crisis moresignificant than the earlier “crises” involvingvideotaping and audiotaping and justifies aserious examination of the issues.

Current technologies for distributingpirated material appear capable of destroyingthe value of copyright, for reasons explainedin detail below (although the recordingindustry failed to present evidence of such animpact in the Napster case). There are twoslightly different technologies now in use totransfer files among users. Both systemsessentially allow individuals to access anddownload music (usually in MP3 format) orother files from other users of the system.These are referred to as peer-to-peer systemssince the files that are transferred are allstored on the users’ PCs rather than commer-cial sites. One type of system, such as Napster,uses a central server to act as an intermediaryin searches for particular songs or files. Theother type, pioneered by Gnutella, forgoes thecentral server, allowing users to search forfiles (often audio, video, or software programfiles) on other computers and to downloadfiles at will. Although the latter seems like apirate’s dream, it is possible, in theory at least,for copyright owners to weather this newform of piracy in one of two ways.

First, copyright harms could be ameliorat-ed through the mechanism of “indirectappropriability” if copyright holders were toset up their file-trading systems according-ly.1(Indirect appropriability is when copyrightowners collect for unauthorized copying bycharging higher prices for originals.)Unfortunately for copyright holders, it wouldhave been easier to convert a centrally orga-nized system such as Napster into a morecopyright friendly system than it will be forthe peer-to-peer systems that appear to bereplacing Napster. Thus, the recent attemptsby the record industry to shut Napster downmay backfire by moving users to the harder-to-control peer-to-peer systems, therebyharming the chance for appropriation ofvalue by copyright holders.

Second, copy protection schemes known asdigital rights management (DRM) or auto-mated rights management (ARM), can ame-liorate copyright harms by making it muchmore difficult to make unauthorized copies inthe first place. DRM often works by inserting

2

It is unclearwhether copy-

right protectioncan continue to provide ample

incentive for artistic creation.

code in the digitized product (e-book, song,movie, or software) that prevents copyingsome or all of the work without acquiring therights through legitimate means. One option(feared by opponents of DRM technologies) isto have pay-as-you go pricing built into theproduct. These payments, sometimes knownas micropayments, can, in principle, allocatemuch more limited rights of access than theall-or-nothing purchases that are typical ofmost physical manifestations of intellectualproperties, such as compact disks (CDs) andbooks. The specter of such self-enforcingcopyright has caused alarm among groups ofscholars and users who argue that it wouldgive too much power to copyright owners anddamage the current balance between produc-ers and users. However, as will be demonstrat-ed below, there is little economic support forthis concern. A review of the basic economicsinvolved is worthwhile.

The Winners andLosers of Copying

The issue at the heart of copyright, indeedof all intellectual property law, is the degreeto which copyright owners can appropriatethe value produced by the consumption, orappreciation, of their works by others.2

At the center of many disputes, both currentand historical, is the question, How muchappropriation is the right amount?3 Is it possi-ble to have too much appropriation? Whatimpact do technologies have on appropriation?

Economists have tended to focus on thetradeoff between consumption efficiency(maximizing the amount consumers get ofany intellectual product) and productionefficiency (preserving incentives to createthese products). On the one hand, if thecopyright holder could not appropriate anyrevenues, the creators of intellectual proper-ties would probably produce far fewer intel-lectual products than would be optimal.4 Onthe other hand, restricting others’ ability tomake copies would put consumption ofthese products at a less than “ideal” level.

This restriction in use is sometimes carelesslyreferred to as a loss due to the “monopoly” ofthe copyright owner. As University ofVirginia law professor Edmund Kitch cor-rectly points out, providing property rightsdoes not confer economic monopoly—whichwould imply that consumers have only asmall number of alternative products thatare not very good substitutes.5 In reality, the“monopoly” conferred by copyright is nogreater than the monopoly that each workerhas on his or her efforts, or that each firm hason products bearing its name. Still, monop-oly power or not, the ideal number of repro-ductions of a public good—a public goodbeing defined as a good that does not getused up when consumed—would requiremaking a quantity of reproductions abovethe level that copyright owners would find intheir best interest to produce.6

Even if a technology were to increase therevenues of copyright owners, say by increas-ing the pool of users, the relative level ofappropriability might still be diminished. Tomake a simple analogy, if we increase the sizeof a pie, even a smaller share of that pie mightresult in a larger piece. In such a case, thecopyright owner would be worse off than hewould be if appropriability were to be con-stant (unless it were impossible to increasethe size of the pie without also decreasing theshare going to the copyright holder).7 Thisdistinction is relevant to the impact of tech-nologies on the financial remunerationachieved by copyright holders.

Surviving Piracy: Exposure and NetworkEffects

The pirating of copyrighted materials isnormally thought to be harmful to the inter-ests of copyright owners.8 This is becausepiracy is often expected to prevent the copy-right owner from appropriating any of thevalue created by his work. The mechanism bywhich unauthorized copying may harm theowners of intellectual products is straightfor-ward enough that no detailed explanationseems necessary. Potential consumers are nolonger compelled to purchase the product

3

The issue at theheart of copyrightis the degree towhich copyrightowners can appropriatethe value produced by theconsumption, orappreciation, oftheir works byothers.

from the copyright owner when the option ofusing unauthorized copies is available tothem. Defections from the legitimate marketare normally expected to reduce the revenuesthat can be earned in the market.

In some instances, however, the piracyimpacts on the copyright holder’s ability toappropriate will be negligible. One obviousinstance is the case where an individualengaged in pirating would not have pur-chased an original even if pirating were notan option.9 In this case, the prevention ofpiracy would provide no pecuniary rewardfor the copyright owner and would onlydiminish the gratification of the individualengaged in piracy.

Another phenomenon that may makecopying helpful, or at least not harmful, tocopyright owners is the “exposure effect”—aform of advertising or sampling that couldultimately lead to larger sales of legitimateversions.10 Users of pirated software, forexample, might find themselves wanting themanuals and technical support that wouldbe available only to authorized users. Or, asclaimed in the Napster defense, Napsterusers may just sample songs to get a betteridea of which CDs to buy.

A more recent claim is that copying maybenefit copyright holders when networkeffects are strong.11 Network effects exist whenthe value consumers place on a good is a func-tion of the number of individuals who use thegood, with the archetypal case being the faxmachine.12 An example of network effects forproducts prone to piracy might be word pro-cessing software that becomes more valuablethe more individuals are using the same soft-ware. In such an instance, it is conceivable thatthe extra value that paying customers receivefrom the larger user base might outweigh anyrevenues lost by the copyright holder frombeing unable to prevent piracy. The preventionof unauthorized copying might actually provefinancially harmful to the interests of the copy-right owner if it resulted in former pirates notusing the product. Of course, if all or enoughof the pirates were to become purchasers ofauthorized versions, the prevention of piracy

would be remunerative for the copyrightowner, even in the presence of network effects.

Even though, in a world characterized byexposure effects or network effects, piracymight work to actually enhance the revenuesof the copyright owner, appropriabilitywould not necessarily be enhanced. Both net-work and exposure effects increase theamount of the value received by consumers,allowing the copyright owner to generatemore revenues with constant or even some-what reduced levels of appropriability. In ourpie analogy, the size of the piece might beincreasing even if the piece’s share of thewhole were decreasing, as long as the pieitself were growing rapidly enough. So even ifpirating were beneficial to copyright owners,meaning that the copyright owner receivedlarger revenues (a larger absolute size piece ofthe pie) that doesn’t mean that appropriabil-ity was increasing (the copyright owner’sshare of the now larger value generated by theintellectual product might be smaller).

Of course, these are the exceptions to themore general rule, which holds that allowingpotential consumers to pirate copies of awork is likely to reduce the revenues availableto the copyright owner.

Surviving Piracy: Indirect AppropriationAs noted earlier, copyright owners are some-

times able to collect revenue from unautho-rized copiers by charging higher prices for theoriginals from which the unauthorized copiesare made. The mechanism is simple: If the copy-right owner knows which originals will be usedto make copies, a higher price can be chargedfor them, allowing the copyright holder to cap-ture part, all, or more of the revenue than mighthave been appropriated through ordinary salesif unauthorized copying could be prevented.

Assume that every purchaser of a compactdisc makes a single audiocassette copy to playin his automobile. No one makes copies fromborrowed CDs. Assume further that this copy-ing, although illegal, is unstoppable. Whatwould be the impact on the copyright holderswho, in addition to selling compact discs, hadalso planned to sell prerecorded tapes?

4

The “monopoly” conferred by

copyright is nogreater than the monopoly that

each worker hason his or her

efforts, or thateach firm has onproducts bearing

its name.

Since each original CD will have a copymade from it, and since it is reasonable toinfer that the consumers of originals placesome value on the ability to make a copy,each consumer’s willingness to pay for theoriginal CD is higher than it would otherwisebe. The copyright owner can capture some ofthis additional value by charging a higherprice for the CD.13 This is the basic ideabehind indirect appropriability. The logichere is the same as would be true for anydurable good that can be resold into anothermarket. If automobiles could not be resold,for example, the price that consumers wouldbe willing to pay for new autos wouldundoubtedly fall.

Whether the copyright owner is better offor worse off in a regime of unfettered copyingdepends on the particular circumstances.Assume, for example, that all consumerswould be willing to pay $9 for a particularCD and would also be willing to pay $4 for acassette tape of the same music that theycould play in their automobile cassette play-ers (assume they do not have CD players intheir cars). If home taping were allowed andconsumers made cassettes, the sellers of CDswould discover that they could raise the priceof CDs to $13 without any loss of sales(assuming no cost to the consumer for thecassette and the time to make the tape). Ifhome taping were disallowed, under thesame assumed circumstances, the seller ofprerecorded tapes could charge a price of $4and capture this group’s value. Either way,the seller would be unharmed by the copyingand thus presumably indifferent to whethercopying was allowed or not.

If there are costs involved in making copies,whether profits would be enhanced or hinderedwould depend on the cost of individuals mak-ing and delivering copies relative to the cost offirms doing so. If it turns out to be much lessexpensive to make prerecorded cassettes com-mercially than to have them made at home, oneat a time, it would be inefficient to have person-al copying replace commercial production. Thecopyright owner will not be able to net as muchfrom the home-taping consumer (who deducts

the cost of the blank cassette and time from hiswillingness to pay) as he would from the sale ofcassettes. Note, however, that commercial costsinclude shipping, inventorying, and delivery tothe consumer, not just manufacturing, whichmakes the cost advantages of prerecorded tapesat least questionable.

Another complication will arise if there is asubgroup of music listeners who purchaseprerecorded tapes for the home instead ofpurchasing CDs. If the price established forthis group is also $9, say, then the seller of pre-recorded tapes is in something of a bind interms of capturing revenues from bothgroups of cassette listeners. If the price of pre-recorded cassettes were lowered to $4 to cap-ture the value from the automobile cassetteusers, the seller would lose $5 from thoseindividuals who would be willing to pay $9 topurchase prerecorded tapes for home listen-ing. If the seller keeps the price at $9, sales tothose wishing to listen to cassettes in auto-mobiles will be greatly diminished.

In this case allowing copying would benefitthe copyright owner. Indirect appropriabilitywould allow the seller to capture the $4 from CDpurchasers by raising the price of CDs to $13,and the seller could still collect the full $9 fromthose who buy prerecorded tapes for the home(the assumption that no copies are made fromborrowed CDs is still in place). In this instance, itwould be more profitable for the seller of tapesand CDs to allow copying. It is possiblethat allow-ing unfettered copying would improve the rev-enue position of the copyright owner.

Note that indirect appropriability impliesthat the purchasers of CDs in the exampleabove actually pay copyright owners, albeit indi-rectly. Fair use, a defense to copyright infringe-ment that allows copying in certain cases (dis-cussed in more detail below), might protect thecopiers from legal liability, but it does not pre-vent the “fair users” from indirectly paying thecopyright owners.

Of course, just because indirect appropri-ability might help create profits doesn’t meanthat it will succeed in any particular case. Animportant influence on the likelihood thatindirect appropriability will work is the vari-

5

Network effectsexist when thevalue consumersplace on a good isa function of thenumber of individuals whouse the good.

ability in the number of copies made of eachoriginal. In the automobile cassette examplediscussed earlier, each CD was used to makeone tape. But if some CDs were used to makeno copies and others were used to make 1,000copies, indirect appropriability would becomedifficult or impossible.

In an atmosphere of rampant copying andvariability in the number of copies madefrom each original, the seller will generallyfind it impossible to identify which originalsshould have the higher price. That is whyinstances of illicit organized copying, where asingle original might be used by a copier tomake thousands of copies, are so much moredangerous to copyright holders than unorga-nized copying where individuals make one ortwo copies for themselves.

Note also that there is less variation in thenumber of copies made from originals whencopying is ubiquitous and similar. Thus, if somecopying is difficult to stop, it might be prof-itable for copyright owners to encourageeveryone to engage in the same degree ofcopying because that can afford the copy-right holder some degree of appropriability.This has interesting implications for Napsterand other digital distribution techniques, asdiscussed below.

There is one other form of indirect appro-priation worth noting. In some instances,legislation may allow copyright owners tocollect revenue in a manner other than charg-ing for use. For example, a tax could beimposed on blank audiotapes or recorders.14

This would not directly charge users for theright to copy, since audiotapes can be used totape works for which copyright clearance wasgiven or for taping uncopyrighted works. Forexample, people buying digital audiotapespay an additional amount that goes to thecopyright owners, but is only indirectly relat-ed to copying. We can refer to this as explicitindirect appropriability as opposed to theimplicit indirect appropriability described ear-lier. On the other hand, organizations suchas the Copyright Clearance Center try todirectly appropriate revenues for the copy-right owners. The CCC gathers rights from

publishers and licenses libraries to makecopies upon payment to CCC—with the pay-ment being a function of how much andwhat is copied based on CCC surveys of copy-ing in the libraries.

The Impacts of NewTechnologies on

Copyright OwnersEach new copying technology may appear

to require fresh analysis as each generationargues that the new technologies created dur-ing its watch require total upheavals of the sta-tus quo. This has been the case with the adventof sound recordings, television, photocopiers,or, most recently, the Internet. But history tellsus that when it comes to copyright, the morethings change, the more they remain the same.Copyright law seems generally to have beensuccessful in balancing the costs and benefitsto both users and producers as technologieshave changed. The trick, of course, is in strik-ing that balance. Some of the major techno-logical challenges are listed below.

PhotocopyingThe ability to photocopy all books and

magazines with ease might have been thoughtto jeopardize the livelihood of authors andpublishers. After all, anyone could take a copy-righted work and make copies without com-pensating the copyright owner. Yet the photo-copier proved a boon to those whose workswere most frequently copied.15

This occurred for two reasons. First, pub-lishers were able to appropriate a portion ofthis additional value by raising their prices.Second, the convenience of being able tomake copies was so great that the nature ofscholarship changed within academic com-munities, and the market for journals grewrelative to the market for books.

The mechanism underlying this growthin journals was indirect appropriability.16

Publishers were able to identify those locationswhere photocopying copyrighted materialsmost frequently occurred—libraries and other

6

Copyright ownersare sometimesable to collect revenue fromunauthorized

copiers by charging higher

prices for theoriginals from

which the unauthorized

copies are made.

similar institutions—and which materials weremost frequently photocopied—academic jour-nals. Publishers then began charging a muchhigher price for library subscriptions than forpersonal subscriptions—often two, three, orfour times as much. The price differentials thatare practically ubiquitous now among publish-ers of academic journals did not exist before thephotocopier arrived on the scene.

Moreover, before the advent of the photo-copier, researchers needed either to have a per-sonal subscription to a journal or to take notesin a library. Books tended to be on single topics,unlike journals, which contained articles on avariety of topic that would appeal to differentscholars. The inconvenience and cost of photo-copying entire books made the practice prohib-itive. Articles in journals, on the other hand,were well suited to the photocopier and becamethe major target of copying activities.Photocopying articles was fast and cheap.Subscriptions were no longer necessary formost users. Having a photocopy of an articlewas such an improvement over handwrittennotes taken in the library, in terms of conve-nience and accuracy, that articles and journalsbecame a far more important means of trans-mitting information than had previously beenthe case. Books, on the other hand, were signif-icantly diminished in importance.17

The price discrimination that the advent ofthe photocopier engendered may or may nothave increased overall appropriability. Clearly,however, photocopying did not harm the copy-right owners, as has been made clear by thegrowth in the number of academic journalsand the financial health of the publishers. Theclaims to the contrary by journal publishers,and there were many, were false.

Videocassete Recording: The Betamax CaseThe Betamax case18 (so called because at

the time the case was brought, the VHS for-mat had not yet begun its obliteration ofBeta) played a central role in the Napsterdefense. The Supreme Court ruling allowedindividuals to make private recordings oftelevision shows on their VCRs. The Betamaxcase represented a situation in which copying

was unlikely to harm copyright owners.19

In the early 1980s, almost all televisionviewing (particularly of the big three net-works) consisted of over-the-air broadcastssupported by advertising. The originalBetamax could accommodate only one hourof recording. The major use of VCRs wasexpected to be “time shifting” programs formore convenient viewing. Although VCRcontrols made it possible for viewers to fast-forward through commercials, close atten-tion had to be paid to avoid fast-forwardingthrough the programming.20 Thus, timeshifting was unlikely to significantly lowerthe revenues that would be derived by televi-sion broadcasters.21 The Court concludedthat time shifting was unlikely to harm copy-right owners.

Although the Court did not rely on thisargument, it was also fairly clear that theamount of time shifting taking place wouldbe small. For one thing, a single VCR couldeither make a recording or play one back, butit could not do both simultaneously.Combine this with the fact that the averagehousehold viewed six or seven hours of TV aday, including virtually uninterrupted view-ing during prime time, and very quickly aconstraint on behavior takes hold. If a familywas going to watch three hours of prime timetelevision on Monday, say, they could notalso watch a tape. If they watched a tape ofthe previous night’s programming, theycould not record the programming that wason while they watched the tape (unless theyhad a second VCR, which was quite rare atthat time). Therefore, when it came to tapingbroadcast television, it was apparent that notthat much taping was going to occur.

Of course, we now know that VCRs areused primarily to play back prerecordedtapes. Ridiculing the difficulty of program-ming VCRs to tape programs unattended hasbecome a staple of stand-up comedians, andtime shifting has not played the damagingrole that copyright owners expected it to play.Nor is there much evidence that individualshave been copying prerecorded video tapesexcessively, although many prerecorded tapes

7

Whether thecopyright owneris better off orworse off in aregime of unfettered copying dependson the particularcircumstances.

do have a fairly primitive anti-backup tech-nology built in.

The difficulty of avoiding commercials,combined with the time limitations involved,made it apparent that videorecording was notgoing to harm copyright owners substantially.Fortunately, the Court managed to get it right,albeit by a narrow, five-to-four vote. Severalyears later, Hollywood learned that by lower-ing the price of popular prerecorded moviesfrom $100 to $20 they could sell far more ofthem. Today, the sale of videotaped moviesgenerates more revenue than theatrical show-ings.22 Hollywood’s claim of impending doomwas, once again, false.

AudiotapingAlthough it lacked the high profile of the

Betamax case, audiotaping was a significantissue in the early 1980s, as illustrated by thisquote from Alan Greenspan in 1983:

Severe economic damage [is beingdone] to the property rights of own-ers of copyrights in sound recordingsand musical compositions. . . . Unlesssomething meaningful is done . . .the industry itself is at risk.23

In a world with no copying, record pro-ducers might find that consumers would beunwilling to pay as much for CDs, whichcould lower revenues and profits.

In response to the dire warnings from therecording industry, Congress considered leg-islation, but it wasn’t until a decade later thatthe Audio Home Recording Act of 1992,which was concerned primarily with digitaltape recording, was passed. The act, whichwas considered a compromise between cre-ators and users, allows personal copying butrequires that recording devices include sys-tems to prevent “serial copying,” that is, mak-ing second-generation copies or copies of acopy. In addition, the law has provisions torequire producers of these recording devicesand recording media to pay a tariff for eachunit produced or imported. The original tar-get of this law, digital audiotapes (DAT),

never achieved any serious market penetra-tion. The devices that have achieved muchgreater penetration, CD writers on comput-ers, are not considered recording devices, donot have copy protection features, and arenot subject to duties. Since the anti-copyingtechnology built into DAT players did notenvision the advent of MP3 files (which arevery compressed versions of the digital fileformat found on CDs rather than strictcopies), the entire MP3 phenomenon wouldhave bypassed those controls anyway.

Digitized NetworkedCopying: Lessons from

NapsterAs mentioned earlier, the entertainment

industry has often expressed the damage thateach new copying technology would bring—from cassette tapes and videorecorders to MP3sand Napster. Crying wolf too many timesshould not, however, negate claims that a newtechnology will harm copyright owners. Napsterand its descendants (as well as movie and elec-tronic book-copying technologies) appear to beinstances where real harm is a possibility.

The Napster program, created in 1999 bythen-teenager Shawn Fanning, offered theability to search for songs encoded in MP3(which is near-CD quality) and identify othercomputer owners willing and able to transferthose songs. Programs that allow computerusers to exchange files from each others’ com-puters (which are not full-time servers) createa type of network that is known as peer-to-peer. Napster was a peer-to-peer-based pro-gram, albeit with a central server that allowedusers to find one another. Napster grew at anexplosive pace and soon had millions ofusers.24 As Napster grew in popularity, so didpotential investors, interested in the brandname and the millions of people reached bythe Napster program and its website.

Some Napster supporters claim that theonline sharing of songs is a latter-dayBetamax scenario. They argue that Napsterusers actually purchase more CDs as a result

8

Each new copyingtechnology may

appear to requirefresh analysis aseach generationargues that the

new technologiescreated during its

watch requiretotal upheavals of

the status quo.

of sampling music with which they mightotherwise be unfamiliar. But given the factthat files downloaded from the Internet are,or at least soon will be, very good substitutesfor the originals, and since they can be“burned” onto CDs and copied to increasing-ly popular MP3 players, it seems likely thatthese files will substitute for the actual pur-chase of authorized CDs. Until most users ofMP3 files are able to transfer them onto CDs,however, MP3 files are unlikely to greatlydamage CD sales. Thus, empirical testing atthis early stage in the development of theMP3 market is likely to find much smallernegative impacts than are likely to occurlater. Also, although some samplingundoubtedly occurs, it seems unlikely thatbeneficial sampling could reverse the nega-tive impacts on copyright holders that wouldbe expected from the substitution of com-puter files for purchased CDs.

Napster-style copying is unlikely to allowrecord companies to indirectly capture thevalue of the copies being made from legaloriginals since some originals will havedozens or hundreds of copies made, and oth-ers none. Nor does it seem likely that theamount of copying will be small—there areno time constraints or confusing instruc-tions preventing widespread copying. Finally,copies are likely to serve as substitutes for thepurchase of originals in this case. The peoplemaking the copies are the very group thatwas expected to purchase originals (which iswhy surveys that indicate that Napster usersare among the heaviest purchasers of CDs areso worrisome to copyright holders).

The Impact of Peer-to-Peer Networks onRevenues: Theory

Napster (now largely dismantled by courtorder) appeared to be a clear threat to the rev-enues of copyright owners of recorded music.The number of individuals using Napster atits peak reached approximately 70 million.25

Many of those users were in the groups mostresponsible for the purchase of recordedmusic—teenagers and young adults. It wasrational for the record companies to fear

Napster: The market is political to someextent, and if enough users wanted to trademusic files for free, they could get Congress tomake that desire the law. That phenomenonhelps to explain the urgency demonstrated bythe courts and the copyright owners.

Although theory alone does not tell us whatimpact peer-to-peer networks such as Napsterhave on copyright owners, theory can provide agood deal of guidance about the likely out-come. What follows are two possible means bywhich Napster-style copying would not harmrevenues of copyright holders—indirect appro-priability and exposure effects.

Indirect Appropriability. First is the possibil-ity of indirect appropriability. One can imag-ine a scenario wherein each individual buyshalf the quantity of CDs that he would nor-mally buy and downloads the other half fromthe network (assuming individuals continueto listen to the same number of songs). Thismay not be unrealistic, since many individu-als would need to have duplicate CDs, or atleast duplicate files. If it were otherwise,everyone trying to download a song from asingle original CD stored on one machinewould run up against limited bandwidthfrom that individual’s machine. Even thoughthe number of copies grows exponentially(since copies can be made from copies), itwould still be much more effective to startwith a large number of originals. Fewer salesof original CDs would make it more difficultfor a large number of users to successfullydownload a song during the period of itspeak popularity.26

Given that scenario, one can still imaginerecord companies selling half the number ofCDs that they would have sold in the absenceof copying. The question is whether the com-panies could then charge approximatelytwice the price (setting aside for argument’ssake that part of the “rationale” given forunauthorized downloading is the high costof today’s CDs), allowing them to appropri-ate roughly the same value as they did before.This could occur if Napster enforced a rulestating that the number of files downloadedhad to match the number of files uploaded.

9

The price differentials thatare practicallyubiquitous nowamong publishersof academic journals did notexist before thephotocopierarrived on thescene.

In such a case, having a file that was in highdemand would be valuable to users wishingto generate credits with which to later down-load files. Under such a system, we wouldexpect users to be willing to pay a higherprice to purchase the original CDs early inthe process, when it would be difficult to geta downloaded copy to make available to oth-ers. Moreover, if it were possible to changethe code in recordings (as envisioned by the1992 Home Recording Act) so that copiescould be made only from an original, but notfrom a copy of an original, users of the sys-tem would be required to purchase CDs ifthey wanted to download songs. If the systemwere “balanced” sufficiently, one can imagineindirect appropriability working to keepboth revenues and appropriability intact.

Of course, in its original format Napsterhad no rule requiring uploads before down-loads could proceed. Users did not have toearn credits in order to download files. Underthis scenario, Napster users who purchasedlegitimate copies had no incentive to pay ahigher price to allow unnamed and unknownusers to download their files.

Finally, some observers (such as Stanfordlaw professor Lawrence Lessig, author of Codeand Other Laws of Cyberspace) have argued thata mandatory indirect means of appropriabili-ty could be put in place, as described earlier inthe case of taxing digital audiotapes.27Lessigused the example of cable retransmission ofbroadcast signals, although similar analogiescould be made to cable networks’ use ofmusic. In each of these cases, cable operatorspay for the right to transmit programs ormusic, but the price is set by a governmentalbody, not by the copyright owner. Lessigargues that this is preferable to a negotiatedprice. These analogies, however, are inapt.

In the instances mentioned by Lessig, theusers of the copyrighted material (e.g., cablenetworks) pay some portion of their revenuesto copyright owners in the form of perfor-mance or retransmission rights.28 Such indi-rect payments work because these uses areadditional or incremental uses and they do notremove the revenues copyright owners

received in their original markets. Thus, thisnew source of revenue would be a net gain tothe copyright owner–– an instance of makingthe pie bigger but also increasing theabsolute size of the piece going to the copy-right owner, whether or not the share of thepie goes up. One can argue about whetherthe payments are too high or too low, butthese extra uses cannot make the copyrightowner worse off than before.

In the case of Napster and other peer-to-peer systems, on the other hand, this logicdoes not work. Getting music from peer-to-peer systems is likely in many cases to substi-tute for the purchase of a CD.29 For this rea-son, it is not just a case of Napster creatingadditional value without payment but one ofNapster reducing payments in the originalmarkets. Napster is not a new use of copy-righted music, with the only problem beingone of allocating revenues. If Napster hadcontinued in its original guise, it is unlikelythat the revenues it generated, say from adver-tising, would have been as large as the losses itimposed on the CD market. Thus, even a taxof 100 percent would not have made copy-right owners whole, to say nothing of givingthem a piece of a new market.30 Napster mayvery well have made the pie smaller. Lessig’ssuggestion is not a useful antidote for thedamage done to copyright owners by the ram-pant copying engendered by Napster.

Alternatively, a tax could be devised on theactivity of making copies. Using digitizedmusic on one’s computer requires the com-puter, a hard drive, a sound card and possiblyrewriteable CDs, and drives. Unlike in the caseof audiocassettes, however, the vast majorityof computer use is not for the purpose ofduplicating copyrighted material; therefore itdoesn’t make sense to tax computers, hard dri-ves, or sound cards. That leaves the rewriteableCDs and blank recordable CDs.31 It is likelythat these devices are sufficiently noninfring-ing to prevent their being classified as infring-ing. So there might not even be an appropriatetarget on which to place a tax as a means ofcompensating copyright holders.

Exposure Effects. Then, of course, there is

10

The data aremore

consistent withthe theory thatonline sales arereplacing brick-

and-mortar salesthan with the

claim thatNapster is

hurting sales.

the possibility that peer-to-peer systemsmight help copyright owners by making iteasier for users to sample songs. If Napsterwere used to merely “try out” songs oralbums (as might be done in a record store orby listening to the radio), Napster use wouldbe a complement to a CD purchase, not asubstitute for it. In fact, Napster’s experts inits court hearings have made this claim, theevidence for which will be examined in moredetail in the next section. Of course, the dif-ference between listening to a song in a storeor on the radio and listening to the songusing Napster is that, with Napster, an actu-al physical representation of the song is inthe possession of the user, whereas in the caseof sampling music in a store or on the radio,only the memory of the tune remains in thelistener’s possession.

Even if Napster were being used for sam-pling only, its impact on the CD market neednot be the benevolent one espoused by itssupporters. The usual assumption is that ifNapster merely helps people decide whichCDs to purchase, it cannot be harmful andwould most likely be beneficial to the copy-right owner. Since Napster is only providinginformation to consumers, or so the argu-ment goes, this activity must benefit societyand copyright owners. From this perspective,because of Napster consumers are better ableto select songs that provide the greatestenjoyment for the time and money. It seemsnatural that they should then be willing topay more for the CDs they purchase.

As appealing as this story is, it is not cor-rect and can be quite misleading.32 The factthat Napster makes the consumer better ableto satiate his desire for music with the CDsthat he purchases implies that the number ofCDs purchased quite possibly would fall.

With better sampling, CDs purchasedprovide greater utility because they better fitthe desires of consumers; therefore con-sumers have a higher willingness to pay.Assuming that all CDs meet the same needfor music consumption, the CDs purchasedprovide greater value and do a better job ofsatiating the desires of consumers. Therefore,

consumers may discover that they do notneed to purchase as many CDs since theirthirst for music can be quenched with fewerof them.33 Depending on supply conditions,it can be shown that the total quantity ofCDs, their price, and the total revenue in themarket may go either up or down.34

The Impact of Napster on Revenues:Evidence

In the Napster case (A&M Records v.Napster) a group of record companies broughtsuit in the Northern District of Californiaagainst the leading online server-based peer-to-peer system, Napster. The hearing occurredin the fall of 2000 and a preliminary injunc-tion barring Napster from allowing users todownload copyrighted files was granted inMarch 2001. The evidence put forward in thehearings on the preliminary injunctionagainst Napster consisted of a set of expertreports by economists, marketers, and others,which were mainly focused on two issues.35

The first was whether or not Napster was like-ly to increase or decrease sales of CDs in themarket. The second was whether or notNapster’s existence as then configured wouldhandicap the nascent market for the legiti-mate selling of music online. Unfortunately,the evidence from these reports provides verylittle in the way of useful guidance on theseissues. Nevertheless, a great deal of misinfor-mation has been put forward about thesereports, so examining them is worthwhile.

These expert reports were submitted forthe purposes of the hearing on the prelimi-nary injunction to stop Napster from trans-mitting copyrighted materials and not for acomplete trial.36 Therefore, it is to be expectedthat these reports might not have the level ofsophistication and completeness that mightcome about in a full trial. Nevertheless, thehearing on the injunction had very high visi-bility and several of the experts were quite wellknown. Additionally, most of those reports(or their authors) were challenged accordingto court rules on the qualifications of experts.The lower court’s rulings on these challengeswas illuminating for the hostility they showed

11

Although thecourtroom decision was technically avictory for therecord industry, itisn’t clear that therecord industry’sstrategy was a wiseone.

toward the Napster experts, a hostility thatseemingly indicated a predisposition of thejudge against Napster. Additional evidence onJudge Patel’s impartiality, or lack of it, camein her nonsensical and later overruled “zero-tolerance” ruling threatening to shut downNapster for any trivial illegal copying that itwas unable to prevent.37

A majority of the reports focused onwhether or not Napster was decreasing thesale of CDs. As already explained, one mightexpect Napster’s impact to be quite small dueto the difficulty of using Napster downloadson the home stereo. In an attempt to demon-strate harm, the plaintiffs had as their center-piece two reports—one that examined thepattern of CD sales in stores near collegecampuses (the Fine report) and the other asurvey of college students asking them abouttheir views on Napster and its impact onsome of their musical habits (the Jayreport).38 The defense had its own survey ofNapster users (the Fader report) and severalcritiques of the Fine report.

In principle, a statistical analysis of howactual CD sales were impacted by the populari-ty of Napster is much preferable to a methodol-ogy based on surveys. For one thing, surveys areself-reported and most Napster users were like-ly aware that Napster was in legal difficulty. Forthis reason, survey respondents might find it intheir self-interest to minimize any evidence thatNapster actually decreased their purchases ofCDs. Further, even if respondents were to tellthe truth, it is unclear they would actually knowwhat impact Napster had had on their behav-ior. Respondents very well might not knowwith any precision how much they spend onCDs per month. Unless they track their expens-es very carefully, their impressions of the impactof Napster on their behavior may very well beincorrect. They may in fact buy some CDsbecause of the songs they downloaded fromNapster and this may color their impressionseven if they purchase fewer CDs overall.

Statistical analysis of actual sales is reallythe only way to determine what Napster’simpact was. Unfortunately, the statisticalanalysis of CD sales reported by Fine pro-

vides ambiguous answers. The Fine reportexamines sales at CD retailers near collegecampuses and compares the sales trends inthose stores to those of other CD retailers.39

The theory is that students at college cam-puses use the Internet and Napster morethan do typical consumers. Therefore, anydifferences in the behavior of sales betweenthe two groups of stores can be attributed toNapster. Although Fader, one of the Napsterexperts, criticized the focus on college stu-dents, this focus is a very practical way of iso-lating the overall impact of Napster, even ifimperfectly. Despite claims to the contrary, itseems unlikely that the behavior of Napsterusers who are not college students (many ofwhom are high school students) is so differ-ent as to counteract the measured impact oncollege students.

This particular design, although poten-tially useful, has some serious problems withits implementation. One major problem withthe Fine study, as both Fader and Hall pointout at length, is that it neglects to control forthe impact that online purchases of CDs, asopposed to downloads of songs, might havehad on brick-and-mortar record stores nearcollege campuses. Internet merchants suchas Amazon and CDNOW increased their CDsales during the period of Napster’s growth.Since Internet access is a requirement forboth using Napster and ordering online, it ispossible that college students have merelyswitched their patronage from brick-and-mortar retail outlets to online retailers.40

Napster experts Hall and Fader claim theFine report is fatally flawed by this oversight.In principle, however, it could be corrected ifdata were available for online retailers andthe two groups of brick-and-mortar retailerson, say, a monthly basis, along with the num-ber of songs being downloaded on Napster.41

Instead, however, the Fine study presentsdata that are very coarse; once-a-year quarter-ly data from the first quarter of 1997 to thefirst quarter of 2000. Given Napster’s briefexistence (it became publicly available inAugust 1999) the data that are supposed toreveal Napster’s influence amount to no

12

Copyright ownersmay have wound

up with a muchmore vibrantdecentralized

system that willprove far harderto stop or from

which to wrangleindirect revenues.

more than a single before and after snapshotof the impact of Napster.

What does the Fine study find? Fine focus-es on the fact that from the first quarter of1999 to the first quarter of 2000, a 12-monthperiod during which Napster came into exis-tence at about the midpoint, sales at brick-and-mortar CD stores near colleges fell byfrom 2 to 3 percent but rose at other brick-and-mortar CD stores by approximately 7percent. From this Fine concludes thatNapster led to a decrease in sales of CDs.Fine’s conclusions are undermined, however,by his data for earlier years. From 1998 to1999, a year preceding Napster’s existence,sales near colleges fell by about 5 percentwhile rising elsewhere by approximately 3 per-cent. Since year-to-year sales at brick-and-mortar CD retailers near colleges were laggingbehind those of other brick-and-mortar CDretailers prior to Napster’s introduction, thefact that they continued to do relatively poor-ly after Napster’s introduction can hardly betaken as evidence that Napster is responsiblefor the difference.

The data, in fact, are more consistent withthe theory that online sales are replacing brick-and-mortar sales than with the claim thatNapster is hurting sales.42 The fact that Napstercould at most have influenced these figures forabout six months, a period during which itwould have had the smallest number of users,makes their data less useful than data based onlater time periods. It is unfortunate that Finechose only four time periods to examine, but itis fortunate, although not for his client, that heprovides more than two years’ worth of data.

Napster’s experts preferred to focus onthe continued robust growth of CD salesoverall, even after Napster’s birth. Certainly,this growth, the 7 percent figure at brick andmortar stores reported above, is inconsistentwith the idea of “irreparable” harm claimedin the preliminary injunction, particularlyconsidering that Napster downloads werereported to be four times as large as the num-ber of legitimately purchased songs. But ithardly demonstrates that Napster had abenign impact on CD sales since there might

well have been other factors at work and theincrease in CD sales might have been evenlarger without Napster’s impact.

A major impediment to measuringNapster’s long-term impact during this peri-od is that MP3 files, the format of music filesused on Napster, were not initially very goodsubstitutes for CDs (though they have sincebecome better substitutes and continue toimprove). At first, MP3 files could generallybe listened to on computers only and couldnot be played on a home audio system (unlessthe computer was hooked up to the home sys-tem). That may help explain why Napster’snegative impact on sales was not apparent. AsMP3 use has increased, however, more andmore audio components have been convertedto play MP3 files. Also, most users were down-loading files over slow telephone modems sothe impact was probably not as large as itwould have been had higher bandwidth con-nections been more readily available.43 Asmore computer users adopt CD writing hard-ware, MP3 files are being converted back intoCD formats that are playable through normalaudio systems. Over time, it is likely that MP3files would have become better substitutes forCDs and would have been played more fre-quently on primary audio systems. If this hadhappened, then the true negative impact ofNapster would have been felt had it not beenshut down.

The other category of report presented inthe Napster case is based on surveys. As notedearlier, the results from surveys should beregarded with great skepticism. The RecordingIndustry Association of America presented asurvey of college students by Deborah Jay, amarketing consultant, that attempted to inferwhether Napster increased or decreased thepurchase of CDs from answers given to ques-tions that do not directly address the point. Jayconcluded that 41 percent of Napster’s sub-scribers used it in ways that displaced the pur-chase of CDs. But the fact that Jay does not askthis question directly of her subjects makes herconclusions suspect. Furthermore, her inter-pretations seem biased against Napster’simpacts. On the other hand, the bias of respon-

13

The record industry has wonthe battle againstNapster, but itmay be losing thewar againstdecentralizedcopying.

dents trying to support Napster might work inthe other direction.

Consider her classification of responses toan open-ended question asking respondentswhy they use Napster. Two of the categoriesof answers, “buy fewer CDs” and “make myown CDs,” are classified as indicating a sub-stitution of Napster files for CD sales. Thefirst answer obviously fits this characteriza-tion. The second, however, is not at all clear.If someone wished to sample music for laterpurchase, and created a CD to sample themusic on a stereo system, Jay would classifythat answer as indicating that the respon-dent uses Napster to decrease CD purchases.The fact that Jay doesn’t provide separatenumbers for each of these two answersmakes it impossible to determine whetherthis is a potentially serious problem or not.Further complicating this issue is the factthat 22 percent of respondents say they eitherbuy fewer CDs or make their own CDs,whereas 8.4 percent say they purchase moreCDs.44 This classification seems capricious.Why not just tell us how many say they pur-chase fewer CDs instead of lumping them inwith those who make their own CDs?

Peter Fader, a marketing professor fromWharton hired as an expert by Napster, criti-cized Jay for using only college students andalso for her interpretation of open-endedquestions. Although I do not believe his crit-icism on the use of college students is valid,the concern with Jay’s interpretation of open-ended questions seems quite legitimate.

Fader conducted his own survey forNapster that tries to answer the same ques-tion. He concludes that Napster decreases CDsales for 8.1 percent of the respondents butincreases sales for 28.3 percent of the respon-dents, virtually the opposite of the conclusionreached by Jay. Fader, however, was harshlycriticized by Judge Patel who questioned hiscredentials and the degree to which he partici-pated in the conduct of the study bearing hisname.45 It appears that he did not supervisethe execution of the study as closely as hemight have. Nevertheless, the harshness of thecourt’s criticism of Fader contrasts with its

generally benign view of the problems in theJay report.

There were other reports, particularly thoseby economists David Teece on behalf of recordcompanies and Robert Hall on behalf ofNapster. Teece’s report is unavailable due to itsuse of confidential information. Teece appearsto have examined the impact of Napster on thecurrent CD market as well as nascent or plannedonline sales of music by copyright owners. Heconcluded that damage to copyright owners isclear. Hall, on the other hand, concluded thatNapster increased CD sales because, in his view,it largely allows users to sample music beforepurchase.46 Once again, the court was muchharsher in its discussion of the Napster expertHall than in its discussion of Teece.47 Withoutseeing Teece’s report, however, it is hard to gaugethe validity of the court’s relative rankings.48

All in all, my reading of the reports in the caseindicates that the plaintiffs in the case failed tomake as persuasive a case for harm as thedefense did for the lack of harm.49 Nevertheless,I believe Napster was likely dangerous to theindustry. The inability of the experts to demon-strate harm may be due to the fact that MP3 fileswere not yet good substitutes for CDs—mostordinary stereo systems cannot yet play MP3files and CD writers are not yet prevalentenough to allow most users to convert MP3s toaudio CDs. Even with the proper equipment,the conversion takes some effort.

Nevertheless, although the courtroomdecision was technically a victory for therecord industry, it isn’t clear that the recordindustry’s strategy was a wise one.

Peer-to-Peer Technologies:The Devil You Don’t Know

Napster is not the only game in town. Anew generation of programs plays by a some-what different set of rules. These are programsbased on the Gnutella protocols or some vari-ation thereof—programs such as BearShare,Aimster, Limewire, Morpheus (FastTrack),and others. The important difference betweenthese programs and Napster is that these sys-

14

Copy protectiondoesn’t have to be

perfect. To be successful it

merely has tolimit the numberof pirated copies

that actuallyreplace sales.

tems are more decentralized—there are oftenno central servers keeping track of the down-loads and uploads.50

Obviously, songs downloaded using theseprograms are likely to have the same type ofdirect impacts on revenues as does Napster.One question is whether they could be as pop-ular as Napster. Current evidence, if it is to bebelieved, indicates that they already are moreheavily used than Napster was at its peak.51

This poses an extremely serious problemfor copyright enforcement. Because there is nocentralized location, firm, individual, or serverthat can be monitored and controlled by legalauthorities, copyright enforcement is going tobe very messy at best, and impossible at worst.There is a provision in the Digital MillenniumCopyright Act that requires Internet serviceproviders (ISPs) to block access when notifiedthat users are serving up copyright materialusing the ISPs’ facilities. The Motion PictureAssociation of America has brought actionasking ISPs to crack down on users providingmovies on pure peer-to-peer-based systemswhen the MPAA is able to monitor those sys-tems and determine the Internet Protocoladdresses of those allowing movies to becopied.52 This is not likely to be a successfullong-term tactic since it would seem to requirestatic IP addresses. Users have a choice of ISPsand it would be possible for many users tomove to systems that use dynamic IP address-es on high-speed lines, such as EarthLink’sDSL (Digital Subscriber Line) service. Also, thevast majority of users still use slow dial-upaccess, and these also have dynamicallyassigned IP addresses.

It is painful to imagine the authorities try-ing to monitor individual computer users andthen prosecuting copyright infringers, oftenteenagers, for downloading music and otherfiles. Yet that is exactly the specter that facesthe copyright owners, who do seem willing toundertake it. They might do well to consider,however, the public relations fiasco that theAmerican Society of Composers, Authors, andPublishers created when it decided to enforcecopyright against summer camps, includingthe Girl Scouts, who believed they were no

longer allowed to sing copyrighted songsaround the campfire.53 The blizzard of nega-tive publicity engendered by that actionrequired ASCAP to backpedal at full speed.Teenagers trading copyrighted songs may notcreate the same degree of empathy as younggirls singing around the campfire, but lots ofparents have such teenagers and the recordindustry will have to be very careful not toalienate the public while punishing otherwiselaw-abiding infringers.

If enforcement against the pure peer-to-peersystems does prove more difficult, how does thepossibility of indirectly appropriating revenuesstack up? At first blush, one might think thatthese pure peer-to-peer systems are rather likethe more traditional exchanges of music orCDs that occur between friends, since there isno central server. But that is not the case. Peer-to-peer networks are able to find an enormousnumber of computers, far more than the num-ber of people in anyone’s circle of personalfriends. On the other hand, there is some indi-cation that users downloading but not upload-ing files may be ostracized by other copyrightviolators since several of these programs have“anti-freeloading” tools (a delicious irony) toprevent users who are not sharing large num-bers of files from being able to download files.54

If an “anti-freeloading” rule became preva-lent, it is possible that the type of balance dis-cussed earlier might arise, so that the value ofnew and popular CDs to users who need toupload songs to “qualify” would increase andsome level of indirect appropriability might bepossible. Nevertheless, it seems unlikely thatdecentralized systems could provide sufficientconstraints to reduce “freeloading” and there-by facilitate indirect appropriability on behalfof copyright holders, as would a centralizedsystem like Napster.55 Thus it might have beensomewhat shortsighted of the copyright own-ers to have brought aggressive action againstNapster, a relatively easy target, when theymay have wound up with a much morevibrant decentralized system that will provefar harder to stop or from which to wrangleindirect revenues. Napster should have beenmuch easier for copyright holders to deal with.

15

For DRM toachieve the typeof power its critics are so concerned with,the technologywould have to be nearlyinvulnerable tocracking.

Although the record industry appears to bemaking peace with Napster, Napster’s usershave largely gone on the more decentralizedsystems. The record industry has won the bat-tle against Napster, but it may be losing thewar against decentralized copying.

Digital Rights Management

In the face of seemingly unstoppable pira-cy, the solution tantalizingly held out tocopyright owners, whether the product ismusic or the written word, is digital rightsmanagement (DRM), also known as auto-mated rights management (ARM). DRMrefers to technologies that seek to preventunauthorized copies of copyrighted materi-als from being made. These mechanisms,buried deep within the digital code of themusic or other copyrighted material, alsohave the ability to allow copies to be madeupon payment, or to charge “micropay-ments” for each small use of the product.56 Inprinciple, such technology could restrictcopying, or even using a copyrighted prod-uct, unless there were a payment.

Of course, copy protection has existed in avariety of forms (computer software, video-tapes, digital audiotapes, scrambled cable TVsignals, etc.) for some time. But none hasproven to be invulnerable. Even the new tech-nologies are prone to being cracked.57 Criticsof DRM as an encroachment on fair useassign far too much power to this technolo-gy. The reason that pirated versions of soft-ware, music, and videos have not dominatedusage is as much the law-abiding nature ofusers as it is the difficulty of copying. Thefact that digital copies of music are techni-cally better than analog copies is a trivial dif-ference to most listeners not using anextremely high-end audio system.58

Copy protection doesn’t have to be perfectto do the job. To be successful it merely hasto limit the number of pirated copies thatactually replace sales. Nevertheless, this mod-ern anti-copying technology has arousedconcerns that it might abridge our freedoms

and tilt the historical balance that existsbetween users and creators too far in favor ofthe creator. I now turn to the fulcrum of thisbalance, known as fair use.

Fair use is a legal defense against claims ofcopyright infringement. Four factors are con-sidered in determining whether a use is fair;and if a use is deemed fair, no copyright pay-ment is required.59 Certain activities are listedin the copyright statute as exemplars of fairuse, such as criticism, comment, news report-ing, teaching, scholarship, or research. There isa debate currently raging about what theimpact of DRM will be on fair use and whatthe consequences will be for society if fair useis largely eliminated.

This newest change in technology has ledto an outpouring of commentary and analy-sis, with two very distinct schools of thoughtemerging. On one hand, some academicsresponded to perceived dangers to copyrightowners by lobbying for passage of the DigitalMillennium Copyright Act. On the otherhand, scholars such as Pamela Samuelson ofBerkeley suggest that these dangers are dra-matically overstated and that the attempt tostrengthen protection has been overdone.60

Following in this vein, Lawrence Lessig hassuggested that the digitizing of artistic workscoded with DRM systems will lead to a farhigher level of appropriability than has his-torically been the case and that technologymay have shifted the balance of economicpower too far in favor of copyright owners ifthe government doesn’t step in to limit it.61

Tom Bell of Chapman University, on theother hand, has argued that systems basedon DRM, what he calls “fared-use,” are logi-cally and legally sound.62

For DRM to achieve the type of power itscritics are so concerned with, the technologywould have to be nearly invulnerable tocracking, and analog versions of the materialwould have to be considered poor substitutesfor the original. The latter condition does notseem to exist in the real world, making manyof the claims of the anti-DRM camp seemunreasonable. For example, one can recorddigital music played on a computer using a

16

Because lowprices can be

charged to consumers whowould not have

purchased theitem at the

high monopolyprice, price

discriminationwill generally

increase the totaloutput sold.

microphone and tape recorder, if one is soinclined. And even if DRM made it impossi-ble to cut and paste electronically from an e-book, one could always type the material intoanother document.

History tells us that the zero-crackingrequirement cannot be achieved through tech-nical sophistication alone. Instead, the powersof the state have to be brought into play, andthe Digital Millennium Copyright Act doesjust that. The DMCA has some draconian pro-visions to prevent copying, in particular a pro-vision that makes it illegal not only to makecopies but to circumvent, or to create toolsthat allow the circumvention of, copyrightprotection technologies. To violate these pro-visions of the DMCA, it is not even necessaryto make copies of the copyrighted materials.These aspects of the DMCA have raised trou-bling free speech and civil liberties concernsthat go beyond the scope of this paper.

Can DRM shift the balance too far infavor of copyright owners? Since Lessigcouches his argument against DRM in termsof economics, it is fair to analyze this claimon that basis.63 The concern that DRM willsomehow cause economic inefficiency—defined as a reduction in the amount of intel-lectual property produced and used—is basedon two largely false premises.

DRM and Fair Use First, the specter of DRM that seems to

haunt its critics most is the version in whichmicropayments must be made for every singleuse, no matter how small. This would appearto eliminate fair use in its traditional role as amechanism for allowing copying when thecosts of collecting payment are greater thanthe payment itself, a view of fair use mostoften associated with Boston University pro-fessor Wendy Gordon.64 She has argued thatfair use provides a mechanism whereby copy-ing may occur when the transactions costs ofgetting permission might have been too greatto allow even worthwhile copying to occur.That argument is similar to that put forwardin my 1981 study, which views fair use as aform of cost/benefit analysis. The critics of

DRM correctly note that, in general, thestrengthening of copyright, and the concomi-tant reduction in consumption of the copy-righted good (since consumption of unau-thorized copies would no longer be allowed),might decrease the value to consumers bymore than the value of any increased produc-tion that might be brought about by thestronger copyright. Fair use, they claim, is amechanism that must be retained to keep anefficient balance.

What the critics of DRM fail to notice inthis instance is that the micropayment thatthey so fear would not reduce the consump-tion of copyrighted goods. By charging foreach minor instance of use, each paragraphread or minute of listening to music, say, oreach page printed, consumers can be chargedamounts that are closely related to their usageof the copyrighted product. This extremeform of DRM becomes, virtually, an instanceof what economists refer to as “perfect pricediscrimination,” wherein each user is chargeda price strongly related to his willingness topay.65 There would be no more “missed oppor-tunities” (where people would forgo copyingbecause the transaction costs of getting per-mission were too high relative to the value).

This is in contrast to the normal textbookrepresentation of a market that has but a sin-gle price charged to all consumers for identicalunits of a product. Although a textbookmonopolist charges a higher price than isfound in a competitive market, it is still a sin-gle price. This monopoly market results in asmaller quantity being sold and consumed. Itis the decrease in quantity consumed that isthe harm engendered by monopoly, whateconomists refer to as economic ineffici-ency.66 On the other hand, when a seller is ableto charge several different prices—such as air-lines that charge higher ticket prices for busi-ness travelers whose trips usually do notinclude a Saturday night stay over—it isknown as price discrimination. Because lowprices can be charged to consumers whowould not have purchased the item at the highmonopoly price, price discrimination will gen-erally increase the total output sold. The more

17

The concern thatDRM will somehow causeeconomic inefficiency—defined as areduction in theamount of intellectual property produced andused—is based onlargely falsepremises.

successfully and completely a seller can matchprices to the maximum prices consumers arewilling to pay, the closer the total output willbe to the ideal (competitive) level.

Unlike simple monopoly, which restrictsoutput from optimal levels, perfect price dis-criminators do not restrict output at all.Therefore, they are every bit as efficient as acompetitive market, a perfectly standardresult that can be found in any textbook.

DRM is not perfect price discrimination.DRM does not charge each consumer anamount exactly equal to his maximum will-ingness to pay. That ideal can never beachieved. Nevertheless, by tying the usageclosely to the payment, DRM will move agood way toward perfect price discrimina-tion. Much of the difference in willingnessto pay is a function of how much and howoften an item is going to be used. In thoseinstances where there is only a small use ofthe copyright material, DRM can charge avery low price and usage (copying) will notbe deterred. Consequently, fair use no longerhas an important role to play in thecost/benefit analysis.

Admittedly, some of the differences in val-uation of the copyrighted good are due toincome or taste differences, which DRMdoesn’t address. The aspect of DRM thatmost concerns its critics, however, is its abili-ty to charge for each small transaction and tokeep copies from being made if payments arenot forthcoming.

This is, of course, a purely economicargument, and it turns on the concept of effi-ciency. Perfect price discriminators, whileefficient, remove the “surplus” received byconsumers. In layman’s terms, the sellersucks up all the difference between the valuethe consumer puts on the product and itscost of production (what is known as sur-plus) and turns it into profits. Critics ofDRM might argue that this is “unfair” tousers. But such an argument is not based oneconomic efficiency and is not the one that ismade by the critics of DRM.

The harm that DRM critics envision—thereduction in use, fair or otherwise, of a copy-

righted good—is not in fact a likely outcome.

DRM and New WorksDRM critics, after incorrectly asserting

that DRM will reduce consumption of copy-righted materials, then argue that there willbe no countervailing benefit, which wouldusually be the additional production of copy-righted materials brought forth by addition-al revenues. This notion that additional rev-enues will not bring forth additional outputarises from an influential paper by Landesand Posner.67 Prior to their paper it had beengenerally accepted that increasing copyrightprotection increases appropriability and,thus, incentives to produce. Greater copy-right protection led to more payments to cre-ators, which led to more creative works beingproduced. It seemed clear enough.

But Landes and Posner broaden that basicmodel by assuming that new works are oftenderived, at least in part, from old works, sothat making it more difficult for the authorsof new works to build upon old works mightactually reduce the number of new works.

In spite of the originality of this claim,there are several reasons to believe that thetraditional expectation that increasing copy-right protection increases the body of copy-righted works still holds true. The Landesand Posner article was examining fundamen-tal questions about copyright, such aswhether ideas should be copyrighted, andmany of their results make the most sense ina legal setting very different from the actualone. For example, if someone were able toclaim ownership over the phrase “goodmorning” and to receive payment each timeit was used, the cost to users of this phrasecould be great, with no concomitant bene-fits. Similar problems might arise if someonecould copyright ideas, such as the idea of twoyoung people falling in love even thoughtheir families disliked one another.

Actual copyright law limits its protectionto the “expression” of ideas. Individuals maycreate a particular expression, even if it hasalready been created and copyrighted bysomeone else, as long as it was not copied

18

Unlike in the caseof patent law, the

first person tocopyright a work

cannot preventothers from

creating the sameexpression of an

idea as long as the later

expressions areindependently

created.

from someone else’s work. Unlike in the caseof patent law, the first person to copyright awork cannot prevent others from creatingthe same expression of an idea as long as thelater expressions are independently created.For example, if you can prove that a sonnetyou wrote, which happens to be identical toone written by Shakespeare, was createdentirely on your own, it will not be consid-ered a copyright infringement. (In patent law,by contrast, even if later inventors wereentirely independent in creating their inven-tion, if it is turns out to be similar to the firstone patented it will not be allowed to standas an independent patent. This can be con-trasted with trademarked characters, whichcan not be copied. Trademark law, for exam-ple, may result in fewer stories that haveMickey Mouse in them. But the profits gen-erated by Disney’s rodent provide incentivesfor others to create their own animal icons.)

Thus, the major problem for creators of newworks comes when they wish to copy direct pas-sages from older works. Such copying is onlylegitimate with attribution, however, and suchwritings tend to be in the nature of reviews oracademic works. In these cases, there is little rea-son to believe that an accommodation wouldnot be made by the creator of the original,unless the derivative work were hostile to theoriginal. But it is unusual to have more than aparagraph or two directly quoted at a time, andthis can be done the old-fashioned “analog”way (i.e., typing by hand) if necessary. Then weare back in the realm of old-fashioned fairuse, even if it now serves a somewhat differ-ent purpose.68

Another concern about DRM is that itappears capable of extending protection indef-initely. In reality, all DRM schemes can be bro-ken, just as all copy protection on software hasbeen broken during the last two decades.Traditional copyright enforcement, and thedesire of users to remain within the law,should limit the distribution of these“cracked” items while the term of copyright isstill in force. Once the statutory copyright pro-tection period is over, however, everyone can ingood conscience distribute “cracked” items.

DRM, therefore, will not lead to permanentcopyright protection. The fact that copyrightprotection doesn’t last forever might be onereason to oppose DMCA provisions thatattempt to prevent all circumvention of pro-tection schemes, since such provisions mightextend protection indefinitely.

It is fair to conclude, therefore, that theincreased appropriability brought about byDRM will enhance the production of newcopyrighted works. If DRM allows a verystrong degree of price discrimination, thenthere would be very little loss from the possi-bility of users being disenfranchised frompurchase by the extra appropriability given tocopyright owners. Even if it eliminated fairuse (and Bell claims that it wouldn’t), DRMwould be economically efficient.69 Of course,one could always argue that DRM wouldmake authors too rich and readers too poor.Or perhaps that it might lead to some formof censorship.70 But these arguments are notbased on economic efficiency, nor do theyseem reasonable.

Finally, although DRM, if it works, wouldbe a salutary development, proposed legisla-tion that appears to require DRM (theSecurity Systems Standards and Certifica-tion Act) is misguided.71 There is no need tohave the government certify which securitytechnologies are to be used or to force a deci-sion on market participants regarding thechoice of which, if any, of these technologiesshould be used. This is just another case ofthe government meddling in an area where itserves no useful purpose.

Other Alternatives for Copy ProtectionWhether embodied in a CD or a stream of

bits over the Internet, copyrighted items (here-inafter to be called “titles,” a shorthand to dis-tinguish them from reproductions of goods)are public goods, which, as defined above,means they don’t get used up when con-sumed. Efficient consumption of the titlerequires that any consumer with a value forthe title (e.g., a collection of songs) that isabove the cost of producing its physicalembodiment or reproduction (the CD) be

19

Once the statutory copy-right protectionperiod is over,however, every-one can in goodconscience dis-tribute “cracked”items. DRM,therefore, willnot lead to permanent copy-right protection.

allowed to consume it, since allowing suchconsumption deprives no other consumer ofpossible consumption. Assume for themoment that the cost of making reproduc-tions is zero. If the price of reproductions werelow enough to allow all potential users to pur-chase them (in other words, zero), there wouldbe no money to pay the creator of the title.

This tradeoff is known to exist in the cre-ation and distribution of public goods.Efficient consumption of reproductionsrequires a price close to the cost of makingreproductions, whereas efficient productionrequires that producers receive sufficient pay-ment to compensate them for the act of cre-ating titles. Traditional market mechanismsare not expected to produce the theoreticallyideal quantity of titles. In reality, too few titlesare likely to be produced and too little con-sumption of any single title is likely to occur,compared to the theoretical ideal.

Ironically, the sole market mechanismthat can theoretically produce the ideal levelof public goods is perfect price discrimina-tion. That is because the perfect price dis-criminator, by definition, charges each con-sumer exactly the amount that consumer iswilling to pay, thus deterring no consumerfrom consuming. The ideal consumptionlevel for any title is thus a by-product of per-fect discrimination. Furthermore, since theperfect discriminator generates revenuesequal to the value the product provides to theconsumer, all titles with total values greaterthan the cost of production are likely to beproduced, leading to the optimal number oftitles produced.

That is the direction in which DRMpromises to take us. DRM promises to makepayment a function of use. Those who listento a song more frequently are likely to alsohave the higher values. Thus it approachesperfect discrimination. So even if critics ofDRM were to change their argument to onebased on the equity or inequity of havingproducers taking so much and consumers solittle, they would then be arguing againsteconomic efficiency, which would be a turn-about of one hundred and eighty degrees.

Although DRM may prove central to resolv-ing the issue of unauthorized copying, thereare other potential solutions that might incor-porate some of DRM’s anti-copying character-istics, although not the payment mechanismsthat are normally associated with DRM.

Some copyright markets, for example tele-vision and radio broadcasts, use a deviceknown as a “blanket license.” A blanketlicense allows the purchaser of the license touse any amount of the copyrighted materialcontained in the repertoire covered by thelicense for a single fee that is not tied tousage. The best-known instances of blanketlicenses have been sold by performing rightsthose organizations such as ASCAP and BMI.

These licenses are sold to television andradio stations, with the price of the licensebeing tied to the revenues earned by the broad-caster. Blanket licenses have some very usefuleconomic characteristics. First, since the cost ofusing another copyrighted item in the reper-toire is zero, consumers who purchase thelicense use the optimal amount of these publicgoods. From an economic efficiency vantage,this is much better than selling the individualitems in the repertoire one at a time (unless theseller is a perfect price discriminator).

It is, of course, possible that some userswould not purchase the blanket license, cre-ating an inefficiency on the consumptionside, although in the case of television andradio, all stations have purchased such licens-es, and the artificial restriction on the num-ber of stations tends to ensure that the blan-ket license fee doesn’t reduce the number ofstations. Also, since a price based on revenuesis likely to be related to willingness to pay(i.e., approaches perfect price discrimina-tion), the system would appear to have excel-lent efficiency characteristics, given that theproducts are public goods. There is no reasonthat a record company could not just as easi-ly employ a similar device in selling music toindividual consumers, what we might call ablanket subscription.

Sellers of music, among them the newNapster, are publicly discussing subscriptionsystems whereby users are charged some

20

The fact that copyright

protection doesn’t last

forever might beone reason to

oppose DMCAprovisions that

attempt to prevent all

circumvention ofprotection

schemes, sincesuch provisions

might extend protection

indefinitely.

monthly fee for access to some amount ofmusic. A monthly subscription fee would belike a blanket license as long as it allowedunrestricted usage of the material covered bythe license and didn’t limit the number ofdownloads each month.

It doesn’t appear that this is the initialroute that is going to be taken by the recordcompanies. News reports indicate thatmonthly subscription fees are going to applyto only a fixed number of downloads permonth.72 These reports also indicate thatrecord companies envision a pricing structureof at least $10 a month for a limited numberof downloads. The downloaded songs, accord-ing to initial plans, will not be playable on anydevices other than PCs and will be limited interms of how many times they can be playedor copied. At higher prices, a small number ofsongs can be “burned” onto CDs.73

This is quite different from the blanketlicense and doesn’t have the same efficiencyconsequences. One might argue that it wouldbe foolish for the record companies to allowunlimited access to their repertoire, since con-sumers would seem to have an incentive todownload everything they want all at once andthen stop paying. This concern, however,ignores the fact that the vast majority of salesare of new music and that it is the new addi-tions to the repertoire that consumers now pur-chase. These additions would provide contin-ued incentive to subscribe to a monthly service.

The original pricing plans appear some-what misguided, but perhaps the recordindustry knows something others do not. Onthe other hand, it is possible that they willfollow the lead of the movie industry, whichinitially overpriced prerecorded videotapesbecause they thought that video rental storeswere the primary market. Only when it wasdiscovered that individual movie viewersactually wanted to purchase videotapes didthe price come down to levels that made itaffordable for individuals. It isn’t really aquestion of whether the sellers will get theprice right—but of when they will get it right.

Further complicating the issue is the factthat web sales compete with brick-and-mor-

tar retail outlets. The brick-and-mortar retailoutlets that sell CDs should pressure recordcompanies to keep online prices at a levelhigh enough not to discourage retail sales.The initial restriction on the number ofburns from a single artist is also consistentwith an attempt to keep web sales from can-nibalizing CD sales. Large retail chains havesignificant leverage on the recording indus-try. To the extent that a large fraction of salesis done through brick-and-mortar retailers,web sales policies may be forced to remainpoor substitutes for CD sales. This would bein spite of the fact that web sales are inher-ently more efficient since they do not incurthe costs of packaging, shipping, distribu-tion, or physically producing the CD. Thismay help explain the seeming unreasonable-ness of current prices and restrictions.

Finally, a bill, the Music OnlineCompetition Act, recently proposed by Rep.Rick Boucher (D-Va.), contains a mandatoryform of nondiscriminatory licensing betweenrecord companies and distributors in themidst of some useful provisions.74 The con-cern is a perennially misguided one that saysthat having a monopoly on both the whole-sale and retail segments of a single market issomehow worse than just having a monopolyon the wholesale side. Economic analysisdoes not support this concern.75 It is quitepossible that the music industry, for exam-ple, will no longer need the services of a sepa-rate distribution channel. If there is littlecompetition in the wholesale segment, theretail segment can do little to change that,and it serves no useful purpose for the gov-ernment to try to tinker with the contractualrelationships that are likely to arise. It is pos-sible that joint ventures between the differentrecord labels might allow a degree of collu-sive behavior with regard to setting prices.Yet, the nondiscriminatory pricing clause ofBoucher’s act is of no value if the nondis-criminatory price is a monopoly price.

The government should not try to preventthe withering of the current separate retailsegment since its existence, competitive orotherwise, serves no particular competition

21

It isn’t really aquestion ofwhether the sellerswill get the priceright—but ofwhen they will getit right.

enhancing function. If there is no longer auseful economic function to be served by aseparate retail segment, and there might nolonger be one, consumers would be better offwithout it.

Conclusion

The impact of pirating has often beenmisunderstood, and copyright owners havefrequently claimed harm when little or nonewas being done. This was true for many copy-ing technologies. Nevertheless, record com-panies and copyright owners are right to fearInternet-based copying of digitized products.It is a potentially serious threat to their well-being. Still, the arguments against Napsterand its relatives remain basically theoretical.As strong as they appear to be, it is somewhatpremature to say we know what will happenone way or another, since there is as yet nocompelling empirical support. The evidencethat has been put forward to this point doesnot clearly point to the direction of theimpact, to say nothing of the magnitude.

Even if Napster had been as serious a threatto record companies’ and copyright holders’revenues as is implied by theory, it is not themost formidable threat facing copyright own-ers. Pure peer-to-peer pirating would seem tobe far more dangerous—Napster could havebeen tamed by changing its rules, but peer-to-peer networks cannot. By crippling Napster,copyright holders may have strengthened a farmore fearsome foe, diminishing their chancesto appropriate revenues from activities thatthey might not be able to stop.

In the not too distant future, DRM tech-nology should allow copyright owners toreduce large-scale unauthorized copying andany harm to copyright owners brought aboutby unauthorized copying. At that point, web-sites and technologies that allow unautho-rized copies should be of far less importanceas long as the DRM technology proves diffi-cult enough to break.

Many scholars and commentators fearthat DRM is dangerous because its practical

elimination of fair use would seem to upsetthe delicate balance between creators andusers. These fears are largely unfounded, atleast in terms of economic efficiency. Becauseprice discrimination is enhanced, there isevery reason to believe that efficiency isenhanced as well. These critics imagine DRMproviding more power to authors than DRMwill in reality be able to deliver.

Inexpensive copying technologies, whichhave been with us for at least 40 years, have notas yet caused great damage, notwithstandingthe claims of the recording and film indus-tries. Although it is possible that the currentgeneration of copying technologies will in factlive up to the dire predictions of doom comingfrom the copyright owners, there should havebeen some powerful evidence to support thisclaim before we considered the type of legisla-tion that has already been enacted, not tomention any new proposals that further tram-ple on individual freedom.

As has been the case for many other tech-nologies, the Internet should prove a boon,not a blight, to record companies and copy-right owners once they learn how to use iteffectively. It provides a wonderful improve-ment in distribution. As was true in thevideocassette industry, however, record com-panies will need to experiment to find theappropriate pricing levels for their products.It is possible that startups with better busi-ness plans will replace the incumbents, butthat is largely irrelevant to copyright issues.Internet distribution should make brick-and-mortar record stores almost obsolete, andwhen it does, the old distribution methodol-ogy will seem as primitive as horses and bug-gies seem today. DRM is likely to be a usefultool in this process.

Notes1. The concept of indirect appropriability, discussedin more detail later, was first propounded in StanLiebowitz, “The Impact of Reprography on theCopyright System,” Copyright Revision Studies,Bureau of Corporate Affairs, Ottawa, 1981, but theactual term “indirect appropriability” was coined inStan Liebowitz, “Copying and Indirect Appropria-

22

As has been thecase for many

other technologies,the Internet

should prove aboon, not a

blight, to record companies and

copyright ownersonce they learn

how to use it effectively.

bility: Photocopying of Journals,” Journal of PoliticalEconomy (October 1985): 945–57. The 1981 mono-graph is available at http://papers.ssrn.com/sol3/papers.cfm?cfid=565423&cftoken=13632430&abstract_id=250082.

2. This focus leaves aside the moral rights to that valuethat are so important under Napoleonic legal sys-tems, but is in keeping with the practical purpose ofintellectual property laws in countries such as theUnited States.

3. One school of thought at the extremity of thesedebates is populated by those who believe that nocopyright at all is required for an efficient function-ing market for artistic and creative goods. The mem-bers of this group believe that either (1) being first inthe market provides sufficient appropriability thatno additional legal protection is required or (2) otherforms of remuneration (perhaps of a nonpecuniarynature) provide sufficient incentive to produce theseproducts, making legislation that restricts the con-trol of these products to their creators unnecessary.The former school of thought is represented byArnold Plant in “The Economic Aspects ofCopyright in Books,” Economica(May 1934): 167–95,and R. Hurt and R. Schuchman in “The EconomicRationale of Copyright,” American Economic Review(May 1966). The latter school of thought is repre-sented by organizations such as the Free SoftwareFoundation, www.gnu.org/fsf/fsf.html.

4. In truth, there is virtually no empirical evidenceon the extent to which copyright owners requireremuneration to create their artistic works.However, the claim that production requires, to atleast some extent, remuneration of the producers, isfully consistent with the usual market principle.Adam Smith’s famous quote about how productiondoes not come from the “benevolence” of butchers,bakers, or candlestick makers, but instead derivesfrom their self-interested behavior, certainly has aplethora of empirical evidence to support it.

5. See Edmund W. Kitch, “Elementary andPersistent Errors in the Economic Analysis ofIntellectual Property,” Vanderbilt Law Review 53(November 2000): 1727.

6. There are actually two definitions of publicgoods in the economics literature. The first definesthem as goods that do not get used up when con -sumed (known as nonrivalrous consumption). Theother, more prevalent, definition is attributed toPaul Samuelson. It adds to the nonrivalrous con -sumption definition the inability to exclude indi-viduals from consuming the good, as would be thecase for national defense or any good withoutdefined property rights. I believe this latter defini-tion to be far less useful since it conflates two inde-pendent ideas that need not have anything to do

with one another. Any good for which nonexclud-ability is a property will not be efficiently producedin markets. And nonexcludability usually has moreto do with the laws and technology than with thegood itself.

7. This might seem to complicate the policyissues, but it actually simplifies them. If a tech-nology decreased appropriability but increasedpayments to copyright holders, it would provideboth greater incentives to create the copyrightedmaterial and greater value to consumers who getto keep the nonappropriated value. Removingthis technology would decrease value regardingboth the number of titles and the value receivedfor each produced title, and could not be eco-nomically beneficial.

8. For a review of the economic impacts of copy-ing, see Richard Watt, Copyright and EconomicTheory: Friends or Foes? (Cheltenham, England:Edward Elgar, 2000). This is the most thoroughreview of this material that I have found. My onlyquibble is that he attributes most of the modelingthat was originated in Liebowitz, “The Impact ofReprography,” to Stanley Besen and Sheila Kirby(1989), “Private Copying, Appropriability, andOptimal Copying Royalties,” Journal of Law andEconomics 32 (1989): 255–80.

9. One point that has been neglected is the pricethat is proffered to the pirate that would lead tohis decision to forgo the product as opposed tomaking a legitimate purchase. There is presum-ably some price above zero at which the piratewould make a purchase when confronted withthis choice.

10. See Liebowitz, “The Impact of Reprography,”for a discussion of the exposure effect.

11. See Lisa N. Takeyama, “The WelfareImplications of Unauthorized Reproduction ofIntellectual Property in the Presence of DemandNetwork Externalities,” Journal of IndustrialEconomics 42 (1994): 155–66; K. R. Conner and R.P. Rumelt, “Software Piracy—An Analysis ofProtection Strategies,” Management Science 37(February 1991): 125–39; and Oz Shy andJacques-Francois Thisse, “A Strategic Approachto Software Protection,” Journal of Economics andManagement Strategy 8 (1999): 163–90.

12. See, for example, Stan J. Liebowitz and StephenE. Margolis, “Network Effects and Externalities,” TheNew Palgrave’s Dictionary of Economics and the Law, vol.2 (New York: Macmillan, 1998), pp. 671–75.

13. Unless, that is, the extra value that the mar-ginal purchaser of originals receives is zero. Thiswould seem unlikely, however.

23

14. Such payments are quite common and can befound in many countries including Canada andmuch of Europe. These payments would normal-ly go to an organization or collective representingcopyright owners.

15. The claims in this section are documented inLiebowitz, “The Impact of Reprography.”

16. It is also true that the Copyright ClearanceCenter came into existence to allow copiers to makedirect payments to copyright holders. But theimprovement in the economic well-being of journalpublishers occurred quite independent of the CCC,since the CCC was not organized until well after themarket for journals had experienced enormousgrowth. See Liebowitz, “The Impact of Reprogra-phy,” pp. 64–68.

17. This is documented in Liebowitz, “The Impactof Reprography,” and “Copying and Indirect Ap-propriability.” Expenditures for books were morethan three times as much as those for periodicalsfrom the 1940s to the 1960s when the ratio beganto fall dramatically. The ratio fell to about 1:1 inthe early 1980s. In 1996, expenditures for serialsoutpaced those for books and bound periodicalsby 8 to 5. See Table 11 in “The Status of AcademicLibraries in the United States,” U.S Department ofEducation, Office of Educational Research andImprovement, NCES 2001-301, May 2001.

18. Universal Studios Inc. v. Sony Corporation ofAmerica, 1984. The original district court rulingwas in 1979.

19. This material is based on Stan J. Liebowitz,“The Betamax Case,” 1986, unpublished manu-script soon to be available on the Social ScienceResearch Network, www.SSRN.com.

20. It was also the case that remote controls at thetime were tethered by wires to the VCR, renderingthem not very convenient to use.

21. Defendants in the Napster and MP3.comcases argued that their products “space shifted”music from a CD to a computer, a putative anal-ogy to the time shifting that occurred in theBetamax case. A problem with this analogy is thatwithout indirect appropriability, space shiftingwould decrease revenues to copyright owners, aresult not analogous to that of time shifting sincethe VCR users still were exposed to commercials.A more important defect with this analogy in thecase of Napster is the fact that what Napster doesis not actually space shifting. Since Napster usersdo not download their own files into their com-puter, but instead download files from others, it isbetter described as “user shifting” than spaceshifting. User shifting could, in other circum-

stances, be considered a euphemism for “theft”except that the theft is from the copyright ownerin the form of a lost potential sale, rather than theuser who provides the original to be copied.

22. According to the U.S. Statistical Abstract, Table909, theatrical movie revenues were $32 per per-son per year in 1998, whereas revenues from pre-recorded movies were $92.

23. Alan Greenspan, Testimony before theSubcommittee on Patents, Copyrights and Trade-marks, of the Senate Committee on the Judiciary,98th Cong., 1st sess, October 25, 1983.

24. In February 2001, 2.8 billion files were down-loaded, the peak number in Napster’s history. ByApril, after Napster was ordered to stop allowingcopyrighted music to be transferred, the numberhad fallen to 1.6 billion. See “Napster DownloadsDrop 36 Percent,” Reuters, May 2, 2001.

25. This estimate was reported in “Napster CouldFace Shutdown,” Associated Press, April 10, 2001.

26. Given enough time, this bandwidth limitationcould be overcome, but the nature of the music busi-ness is that a small number of songs are in extremelyhigh demand for a short period of time. As long astastes gravitate around the same small number ofsongs at any one time, the argument holds.

27. See Lawrence Lessig, “Just Compensation,”The Industry Standard, April 18, 2001.

28. When music is put onto a record or CD, the cre-ator of the music receives a small payment for eachcopy, which is known as a mechanical royalty.Similarly, composers receive a percentage of televi-sion, radio, cable, or concert revenues as compensa-tion from broadcasters for the use of their music,which is known as a performing rights payment.When broadcast signals are carried on cable, that isoften known as a retransmission right.

29. However, the current evidence is unclear asreported below in the section on Napster’s impacton revenues.

30. It was never clear what Napster’s businessmodel was when it was merely allowing the freetransfer of files, but one possibility was the sale ofadvertising. The possible advertising revenues weredwarfed by the potential losses that might havebeen imposed upon record companies.

31. The Audio Home Recording Act of 1992 providesthat importers, manufacturers, or distributors of anydigital audio recording device or digital audio record-ing medium must file quarterly statements and payroyalties on each recorder or piece of medium distrib-

24

uted in the United States. The royalty is 2 percent ofthe manufacturer’s selling price for recorders and 3percent of the manufacturer’s selling price for record-able media. See Robert A. Starrett, “Copying Music toCD: The Right, the Wrong, and the Law,” EMediaProfessional, February 1998, www.cdpage.com/Audio_Compact_Disc/rightwrong.html.

32. A typical view is espoused in the expert reportsput forward by Napster in its defense. One of thosereports, by Robert Hall, states, “The exchanges ofmusic facilitated by Napster stimulate the demandfor the plaintiffs’ CDs by allowing consumers tosample CDs and develop interest in CDs that theysubsequently purchase.” http://napster.com/pressroom/legal.html. Several, but not all, of the reports fromthe Recording Industry Association of America canbe found at www.riaa.com/napster_ legal.cfm.

33. Another way of looking at this is to imaginethat some CDs that are now purchased are “mis-takes” due to insufficient information. With theadditional information provided by the Napsterexperience, fewer of these mistakes are made andfewer CDs are purchased.

34. By analogy, it is as if CDs were chocolate bars.Consumers buy chocolate bars so that they can eat thechocolate. If each bar were to contain more chocolate,holding the price of a bar constant, the number of barssold could go up or down depending on the elasticityof demand for the underlying product of interest,chocolate. If the elasticity of demand for chocolatewere greater than one, the now-lower effective price ofchocolate would lead to an increase in total revenuespent on chocolate, and with the price of bars constantthe number of bars sold would increase. (This rela-tionship between elasticity and revenue can be foundin any introductory microeconomics text.) But if thedemand for chocolate were inelastic, the number ofbars sold would decrease. Although it could be arguedthat the demand for any particular CD is elastic, sinceotherwise the seller would find it profitable to raise itsprice, it need not be the case that overall demand forCDs is elastic. CD prices are not set individually (seeSilva and Ramello), and CDs often would seem to beclose enough substitutes for one another to be classi-fied in the same market.

35. There were several other reports that were dif-ficult to classify. Lawrence Lessig submitted areport that the court rejected out of hand, statingthat “the Lessig report merely offers a combina-tion of legal opinion and editorial comment onInternet policy. Therefore, this court grants plain-tiffs’ motion to exclude it.”

36. The preliminary injunction preventing Napsterfrom allowing users to download copyrighted filesis in place, and Napster was effectively shut downin the form in which it had previously existed.

Napster is now planning to return as a legitimatesubscription service.

37. The court told Napster to end the dissemina-tion of copyrighted materials. Napster did this byblocking access to known copyrighted songs,which required many different variations for eachtitle since users tried to evade being blocked byputting up alternative titles that still indicatedwhat the song was. When Napster was only 99percent effective in this endeavor, the judgethreatened to shut it down.

38. These reports are so named in the “Memoran-dum and Order Re Admissibility of ExpertReports” issued by the trial judge. Plaintiffs alsohad a declaration by Charles Robbins, a storeowner claiming that Napster had largelydestroyed his business, but this report was thor-oughly discredited by the Fader report (men-tioned below), which pointed out that the storehad changed locations and switched from sellingnew CDs to selling used records and CDs duringthe period that its sales declined.

39. Actually, the Fine study looks at three groups ofbrick-and-mortar retailers: the overall set, a set of retail-ers near the 40 most heavily wired college campuses,and a set of retailers near college campuses that havebanned Napster. Napster’s expert Hall makes much ofthe fact that this latter group of retailers shows thesame decrease in sales as the others, claiming that forthis group sales should have improved if Napster hadbeen having a negative impact on sales. Such a claim isunwarranted since we do not know how long Napsterhad been banned at these campuses and how success-ful the ban was.

40. It is unfortunate that online CD retailersprobably had not achieved an equilibrium marketposition prior to Napster since comparing thesales of online retailers to those of brick-and-mor-tar retailers would have provided what probablywould have been a better test of Napster’s impact,with changes in online sales being the proxy forNapster’s impact.

41. Fine reports that online sales figures were firstcollected in the first quarter of 1999.

42. The court also was aware of these problems:“The Court finds some aspects of the Fine reporttroubling—especially the fact that its shows adecline in retail sales prior to the launching ofNapster. This limitation, combined with Fine’sdecision not to track Internet music sales, reducesthe study’s probative value.”

43. Nevertheless, at its peak Napster downloads wereestimated to be in the vicinity of 2.8 billion files permonth, which would roughly be the equivalent of

25

250 million CDs per month. I suspect that thesedata include many failed downloads. According tothe Fine report, U.S. national sales ran approximate-ly 60 million CDs per month. So even with the slowbandwidths, the potential impact may have beenlarge. See “Music Downloads Soar,” Reuters,September 6, 2001, http://news.cnet.com/news/0-1005-200-7080479.html.

44. Jay includes “getting free music” and “gettingmusic that one wants” as other answers thatreveal substitution of sales. Although this mightbe true, it is not clearly so, and this interpretationproblem could have been avoided if more directquestions were asked.

45. The court stated: “He considers himself anexpert on consumer surveys. . . . However, he admit-ted in his deposition that he has never before pre-pared a consumer survey for litigation and he isunfamiliar with the standards set forth by federalcourts for the reliability of such surveys.” Lack offamiliarity with legal standards hardly disqualifiessomeone as an authority on surveys in general. Thecourt continued, “In short, his claim to havedesigned and overseen the Greenfield surveyappears exaggerated, and the generality of hisreport renders it of dubious reliability and value.”

46. Hall assumes that any music sampling byNapster users benefits the CD market, anassumption that we demonstrated to be incorrectin the above subsection on exposure effects.

47. The court stated: “Hall relied too heavily onoutside studies that favored defendant withoutperforming any analysis of the Jay report . . . .these shortcomings are not grave enough to war-rant exclusion of his expert opinion. Insofar asthe Hall report assumes the requested injunctionwould put defendant out of business, it tends tocorroborate plaintiffs’ argument that Napster hasno legitimate non-infringing uses . . . . they [theplaintiffs] would be wise not to object too strenu-ously to admission of the Hall report.”

48. One potentially questionable point, accordingto Hall, is Teece’s use of the concept of pathdependence to argue that consumers will belocked in to Napster and will not then purchasemusic online from the copyright owners. Hallnotes that Teece cites the QWERTY keyboard asan example of such lock-in. There are two prob-lems here. First, the keyboard story has no evi-dence to support it. Second, it is hard to imaginewhat the coordination problem might be thatwould have to underlie a case of lock-in since net-work effects are not sufficient for there to be lock-in without some form of coordination failure. SeeStan J. Liebowitz and Stephen E. Margolis, “The

Fable of the Keys,” Journal of Law and Economics(April 1990): 1–26. (This article also cited by Hall.)

49. The judge’s readings of the reports seem tohave been biased against Napster, even though Ithink her decision was in the end correct, even ifnot supported by the evidence at hand.

50. That distinction is somewhat artificial.Napster is also a peer-to-peer system, albeit onewith a central server. And the pure peer-to-peer-based systems can have some specialized hard-ware—LimeWire, for example, has a special routerthat is supposed to improve performance.

51. The claim is made that the four largestNapster replacements, FastTrack, Audiogalaxy,iMesh, and Gnutella, were responsible for 3.05billion downloaded files, although not all ofthose were songs. The data come from Webnoize,a company that tracks Internet usage as reportedby Reuters, as discussed above.

52. See Lee Gomes, “Recording Industry TargetsGnutella amid Signs Napster Usage Is Falling,”Wall Street Journal, May 4, 2001, p. B6.

53. For one of the more subdued articles on thisissue, see Marcus Errico, “Okay, Six Choruses of‘Kumbaya’—That’ll Be $1.50,” Eonline, August 24,1996, www.eonline.com/News/Items/0,1,109,00.html. For ASCAP’s version of this public relationsfiasco, see ASCAP’s memo “ASCAP Clarifies Posi-tion on Music in Girl Scout Camps,” August 26,1996, www.ascap.com/press/ascap-082696. html.

54. For example, LimeWire has an “Anti-FreeloaderFeature” as described in the FAQ section: “Q: Willthe number of files I share affect my LimeWireexperience? A: It could. If you’re not sharingenough files, users with certain connection prefer-ences won’t let you connect to them for download-ing. For this reason, we recommend all LimeWireusers share generously with one another.”

55. If Napster were to have a no-freeload rule, onemight think that competition between Napsterand a decentralized system would favor a decen-tralized system since freeloaders would flock thereand most everyone would want to be a freeloader.But such is not necessarily the case. The systemwith a no-freeload rule would have more heavilydemanded files available for downloading relativeto the number of downloaders and thereforeshould prove more attractive to users, particularlyusers with files to upload. What economists referto as a separating equilibrium might come to exist,with Napster having the newer and still popularsongs whereas the pure peer-to-peer networkswould have older songs.

26

56. One problem is that credit card companiescurrently charge a fixed fee for each transaction,which tends to negate the usefulness of micro-payments. This will presumably be fixed someday.

57. For example, to test the effectiveness of water-marks, a technology that in principle allows thetracking of the origin of a copy, the Secure DigitalMusic Initiative last year sponsored a hackingchallenge, offering $10,000 to anyone who couldsuccessfully remove the watermarks while meet-ing certain audio-quality standards. That chal-lenge was met and a minor brouhaha took placeover whether the codebreakers (a group fromPrinceton) would be allowed to publish theirtechniques, which they eventually did. See Lisa M.Bowman, “SDMI Hack Draws Legal Threats,”ZDNet News, April 23, 2001, www.zdnet.com/zdnn/stories/news/0,4586,5081595,00.html. Thereport of the hackers can be found at www.thereg-ister.co.uk/extra/sdmi-attack.htm.

58. Of course, digital copies can be used to makeanother generation of digital copies that remain thesame quality, whereas analog copies of copies dete-riorate in quality for each additional generation.

59. The four factors are (1) the purpose and char-acter of the use, including whether such use is of acommercial nature or is for nonprofit educationalpurposes, (2) the nature of the copyrighted work,(3) the amount and substantiality of the portionused in relation to the copyrighted work as a whole,and (4) the effect of the use upon the potentialmarket for or value of the copyrighted work.

60. See Pamela Samuelson, “The CopyrightGrab,” Wired , January 1996, www.wired.com/wired/ archive/4.01/white.paper_pr.html.

61. “Code displaces the balance in copyright lawand doctrines such as fair use.” Lawrence Lessig,Code and Other Laws of Cyberspace (New York: BasicBooks, 1999), p. 135.

62. Bell points out that fair-use imposes costs ofits own on the copier (photocopiers, time, and soforth). It also imposes uncertainty on users sinceit is often unclear whether individual acts are fairuse or not. He also moves a bit in the direction ofthe efficiency of perfect price discrimination thatI put forth when he suggests that many instancesof worthwhile copying would be missed becausegetting permission was too costly, but that ARMreduces such transactions, potentially increasingthe number of such uses. See Tom W. Bell, “FairUse vs. Fared Use: The Impact of AutomatedRights Manage- ment on Copyright’s Fair UseDoctrine,” North Carolina Law Review 76 (1998):558–618, www.tomwbell.com/writings/FullFared.html.

63. “Economists have long understood thatgranting property rights over information is dan-gerous (to say the least). This is not because ofleftist leanings among economists. It is becauseeconomists are pragmatists, and their objective ingranting any property right is simply to facilitateproduction, but there is no way to know, in prin-ciple, whether increasing or decreasing the rightsgranted under intellectual property law will leadto an increase in the production of an intellectualproperty,” Lessig, Code, p. 134.

64. Wendy Gordon, “Fair Use as Market Failure: AStructural and Economic Analysis of the BetamaxCase and Its Predecessors,” Columbia Law Review82 (1982): 1600–1657.

65. I am making the assumption that the moreone uses a product, the greater one’s willingnessto pay.

66. The harm from the reduced output is tradi-tionally referred to as a deadweight or welfare lossin economics textbooks.

67. W. M. Landes and R. A. Posner, “An EconomicAnalysis of Copyright Law,” Journal of Legal Studies18 (June 1989): 325–63. One might argue that ear-lier, Plant also thought that lengthening copy-right would not increase production since Plantargued that copyright wasn’t even necessary. Thisis, I believe, a misreading of his work. Plant does-n’t argue that production is generally unaffectedby income, although he does point out that this istrue for some authors. He also argues that theprofits from being first to market (prior to 1934,when copying was still slow and costly) were greatenough that authors received all the reward theyneeded.

68. It can be claimed that the DRM might imposea contract on the user that no copies of any sortare to be made as a condition of sale. But anybook, even with no digital technology at all, couldcome shrink wrapped with a contract that says toopen it only if you promise not to make copies.That is the nature of many software contracts. Sothis problem really has little to do with ARM.

69. See Bell, p. 591, where he discusses the legalimpacts of ARM on cutting and pasting.

70. According to Neil Weinstock Netanel,“Copyright and a Democratic Civil Society,” YaleLaw Journal 106 (1996): 283–386, fair use is valu-able to society, even if perfect metering reducedtransactions costs to zero, because in his view thepurpose of copyright is to promote a democraticsociety and not to maximize economic efficiency.He believes that fair use protects negative follow-upworks (e.g., parodies) that he suggests could be

27

eliminated by the original copyright owner withstrong enough copyright protection. This, to me,confuses censorship with the proper working ofmarkets.

71. See Declan McCullagh, “New Copyright BillHeading to DC,” Wired News, September 7, 2001, wwwwired.com/news/politics/0,1283,46655,00. html.

72. The two leading services that have beenannounced, MusicNet (supported by Warner,Bertelsmann, EMI) and Pressplay (supported bySony and Vivendi), have suggested some form ofmonthly subscription fee. The suggestions, how-ever, are for a fixed amount of downloads.Reports Newsbytes: “Acting MusicNet CEO RobGlaser said the lower end of a hypothetical tieredpricing plan might include a $9.95 per monthplan allowing customers to temporarily down-load 30 songs and stream 30 titles.” That samearticle reports on a survey of teenage users, indi-cating they would be willing to pay slightly under$3 per month for unlimited downloads. See BrianKrebs, “Plug.In: Music Ser-vices May Strugglewith Napster-Era Teens,” Newsbytes, July 26, 2001,www. newsbytes.com/news /01/168370.html. Amore recent article discusses the problems withselling music online, particularly since these sites

intend to stream music, which brings in perform-ing rights as well as other rights. See Jim Hu andJohn Borland, “Label Deal to Unclog MusicLogjam,” Cnet News, September 17, 2001, www.msnbc.com/news/630261. asp?0na=22184D2.

73. On December 19 Pressplay officially an-nounced its pricing: $10 per month for 30 down-loads plus 300 streams (listening to a song) at thelow end, and $25 per month for 100 downloads,1,000 streams, and 20 burns (songs put on a CD).There is also a monthly limit of two burns permonth per artist. The limit of two burns permonth will tend to keep the service from being asubstitute for album sales.

74. For a copy of the act, which does contain someuseful provisions, see www.house.gov/bouch-er/moca-page.htm.

75. The analysis tends to go the other way. Bothwholesale and retail channels, if they are not per-fectly competitive, tend to put their own supra-competitive markups on the product, causingmore harm than would be the case for a singleintegrated channel, a result known as “doublemarginalization.”

28

Published by the Cato Institute, Policy Analysis is a regular series evaluating government policies and offer-ing proposals for reform. Nothing in Policy Analysis should be construed as necessarily reflecting the viewsof the Cato Institute or as an attempt to aid or hinder the passage of any bill before congress. Contact theCato Institute for reprint permission. Additional copies of Policy Analysis are $6.00 each ($3.00 each for fiveor more). To order, or for a complete listing of available studies, write the Cato Institute, 1000 MassachusettsAve., N.W., Washington, D.C. 20001, call toll free 1-800-767-1241 (noon - 9 p.m. eastern time), fax (202) 842-3490, or visit our website at www.cato.org.