Intellectual Property's Canary in the Digital Coal Mine
Of all the content industries affected by the digital environment, the music industry has, for a variety of reasons, been thrown first into the maelstrom. Events have proceeded at the dizzying pace that has been called "Internet time," with technical, legal, social, and industrial developments occurring in rapid-fire succession.1 Yet the problems facing the music industry will likely soon be found on the doorstep of other content industries. This chapter presents the developments in the music industry and reviews its early phases of coming to grips with digital information. These developments offer an intriguing case study illustrating the problems, opportunities, possible solutions, and cast of characters involved in dealing with digital intellectual property (IP). The focus is not on the day-to-day specifics, as these sometimes change more rapidly than daily newspapers can track. Instead, the perspective is on the underlying phenomena, as a way of understanding the issues more generally. Not all these issues will play out identically in different industries, of course. But some
NOTE: Underground coal deposits are invariably accompanied by methane gas, which is highly explosive, but colorless, odorless, and tasteless. Before more advanced detectors became available, miners would take canaries into the mines with them because the birds were far more susceptible to methane and thus offered advance warning. The concept has since become a metaphor for anything that serves in that role.
1For example, according to net measurement firm Media Metrix, an estimated 4 million people in the United States listened to digital music in the month of June 1999, up from a few hundred thousand less than a year ago (Wired News, 1999).
of the problems will be widespread, because they are intrinsic to digital information, no matter what content it carries. The problems include distributing digital information without losing control of it, struggles over standards and formats, and evolving the shape of industries as the new technology changes the previous balance of power.
The problem, or opportunity, has hit music first for a variety of reasons. First, files containing high-fidelity music can be made small enough that both storage and downloading are reasonable tasks. Digitized music on a standard CD requires about 10 megabytes per minute of music; with a format called MP3, that same information can be compressed so that it occupies about one-tenth as much space.2 As a result, music files currently require about 1 megabyte for each minute of music (or about 45 megabytes for a typical album) yet offer generally acceptable (though not quite CD quality) sound. With multigigabyte disk drives common, dozens of albums are easily stored directly on a hard drive or inexpensively written to writable CDs.3 Video, by contrast, contains a great deal more information: A digitized 2-hour movie (e.g., on a DVD) contains about 5 gigabytes of information.
Second, access to digitized music is abundant, and demand for it is growing rapidly. Numerous MP3 sites offer free MP3 playback software, songs, and albums. With a 56K modem (which provides a sustained transfer rate of about 5K bytes/second), a 5-minute song takes about 17 minutes to download, an album about 3 hours. With access to high-speed network connections growing more commonplace (e.g., at work and on campuses), sustained download speeds of 50K bytes/second and higher are widely available, making possible the transfer of a song in under 2 minutes and of an entire album in about 18 minutes.
2MP3 is shorthand for Moving Picture Experts Group (MPEG) 1 Layer 3. Other compression techniques are available that offer even higher compression ratios, making the files smaller still. Although the MP3 format has been the most recent concern of the recording industry, problems involving digital music and IP arose some years ago, with the proliferation on the Web of unauthorized versions of songs written in MIDI format, and recordings in WAV and AU format (see Box 1.2 in Chapter 1 for an explanation of formats and file compression).
3A gigabyte of hard drive space can hold 1,000 minutes (roughly 20 albums' worth) of music; desktop machines routinely ship with multigigabyte hard drives and can thus easily hold 50 to 100 albums. The hardware to write one's own CDs was a few hundred dollars in 1999 and is sure to become cheaper, and a blank recordable CD costs about a dollar. Using MP3 encoding, a recordable CD will hold 10 albums' worth of songs. Hence to someone with the equipment, the marginal cost of copying an album is approximately ten cents and is sure to decline further as technology improves.
But you need not go to the Web to find digitized music. A very large percentage of the music industry's current content is already available in an unprotected digital form: CDs. Widely available software programs known as ''rippers" (or "digital audio extractors," in more polite circles) can read the digital data from CD tracks and rewrite it in a variety of formats, notably as MP3 files.4 These files are easily shared among friends or posted around the Web.
The third reason that the problem has surfaced first in the music world is that music is popular with a demographic group (students in particular, young people generally), many of whom have easy access to the required technology, the sophistication to use it, and an apparently less than rigorous respect for the protections of copyright law.5 Students also constitute a well-defined and geographically proximate community, which facilitates the sharing of digital music files.6
Fourth, music can be enjoyed with the existing technology: Good speakers are easily attached to a computer, producing near-CD quality sound, and a variety of portable players (e.g., the Rio from Diamond Multimedia) are available that hold 30 minutes to an hour of music. By contrast, even if it were available on the Web, downloading a best-selling novel is not enough; it would still have to be printed before you could enjoy the work.
W(h)ither the Market?
What are the consequences for the recording industry? It is facing an age-old question that lurks in the background of most innovations that affect intellectual property: Something is about to happen, but will it be a disaster or an opportunity? New technology and new business models for delivering content are almost always greeted with the belief that they will destroy the existing market. In 17th century England, the emergence of lending libraries was seen as the death knell of book stores; in the 20th century, photocopying was seen as the end of the publishing business, and videotape the end of the movie business (Shapiro and Varian, 1998). Yet in each case, the new development produced a new market far larger than the impact it had on the existing market. Lending libraries gave
4Rippers copy the digital data from the CD to the computer's hard drive, reading the CD as if it were a (very large) floppy disk.
5On college campuses, the use of digital music files has created sufficient levels of network congestion to cause some network administrators to put MP3 servers on their own subnet to minimize the disruption to the main campus network (as reported by a briefer at the committee's meeting on July 9, 1998).
6See Gomes (1999).
inexpensive access to books that were too expensive to purchase, thereby helping to make literacy widespread and vastly increasing the sale of books. Similarly, the ability to photocopy makes the printed material in a library more valuable to consumers, while videotapes have significantly increased viewing of movies (Shapiro and Varian, 1998). But the original market in each case was also transformed, in some cases bringing a new cast of players and a new power structure.
Will digital information do the same for music? Some suggest that the ability to download music will increase sales by providing easy purchase and delivery 24 hours a day, opening up new marketing opportunities and new niches. For example, the low overhead of electronic distribution may allow artists themselves to distribute free promotional recordings of individual live performances, while record companies continue to focus on more polished works for mass release. Digital information may also help create a new form of product, as consumers' music collections become enormously more personalizable (e.g., the ability to create personalized albums that combine individual tracks from multiple performers). Others see a radical reduction in aggregate royalties and, eventually, in new production, as pirated music files become widely distributed and music purchases plummet.
The outcome for digital music is still uncertain; there is of course no guarantee that the digital music story will play out in the same way as it has in these other industries. But past experience is worth considering, reminding us as it does that at times innovation has contributed to a resurgence in the market rather than a reduction.
What Can be Done?
There are two major lines of response to the challenges outlined above: find an appropriate business model, and develop and deploy technical protection mechanisms. Each is considered below.
The Business Model Response
"The first line of defense against pirates is a sensible business model that combines pricing, ease of use, and legal prohibition in a way that minimizes the incentives for consumers to deal with pirates" (Lacy et al., 1997).7 This view nicely characterizes the business model response, suggesting that one way to cope with piracy is to provide a more attractive
7The paper cited describes a new technical protection mechanism for music, yet (appropriately) begins by acknowledging the power of a good business model.
product and service. The difficult part of this approach is that it may require rethinking the existing business model, industry structure, and more.
Make the Content Easier and Cheaper to Buy than to Steal
One example of that rethinking is the suggestion by Gene Hoffman, CEO of EMusic, Inc. (formerly GoodNoise, a digital music provider): "We think the best way to stop piracy is to make music so cheap it isn't worth copying" (Abate, 1998).8 EMusic sells singles in digital form for 99 cents.
But if those singles are in the form of MP3 files, they are unprotectednothing prevents the purchaser from passing a file on to others or posting it on a Web site (illegal though these actions may be). So how could there be a market in such things, when it appears that all the value in the item could be extinguished by the very first sale?
One answer becomes evident to anyone who has actually tried to download MP3 music from any noncommercial site: The service is terrible and the experience can be extraordinarily frustrating.9 Search engines can assist in finding songs by title, performer, and so on, but you have to know how to look: Can't find what you're looking for when you type in "Neil Young"? Try "Niel Young." In any collection, quality control is a problem; when the data are entered by thousands of individual amateurs, the problem is worse.10
When the links are found, the next question is, How long are you
8A similar comment came from Steve Grady, vice president of marketing for EMusic, who concluded that "the best way to combat piracy is to make it easier to buy rather than to steal" (Patrizio, 1999a).
9Some commentary from the Lycos Web page on downloading MP3 files illustrates these problems:
In some situations there can be problems downloading the MP3 files. . . . This is due to the unstable nature of the MP3 serversmost of these are run on an amateur basis, and the reliability is not very good.
On many servers only a very limited number of users can be logged in at a time. The only way around this problem is to repeatedly try to download the files until there are available connections on the server.
Many servers are up for only a limited time. On others, the MP3 files may have been removed. In both these cases we will have dead links until these changes have been updated in our database ... it is not usually known whether a server is permanently down, or whether it will be available again. We have therefore chosen to keep the files in the database until we are sure that they have actually been removed, with the cost of displaying a few dead links.
10A particularly insidious example of the sites distributing music illegally are the so-called "ratio" sites, which permit downloading of MP3 files only in exchange for the up-loading by the user of at least one "ripped" MP3 file.
willing to keep trying, when receiving responses such as "Host not responding," "Could not login to FTP server; too many usersplease try again later," and ''Unable to find the directory or file; check the name and try again"? The computers containing the files are often personal machines that are both unreliable and overloaded.
Even once connected, the speedy download times cited earlier are ideals that assume that both the computer on the other end and its connection to the Internet are up to the task. The real-world experience is often not so good: Creating a Web site with a few music files is easy; providing good service on a site with hundreds or thousands of songs is not: The hardware and software requirements are considerably more complex.
Where is the business then, if files are unprotected by technical mechanisms (even if legally still protected)? It may be in the service, as much as the content. Why experience 30 minutes or an hour of frustration, if for a dollar or so you can have what you want easily, reliably, and quickly? This is one example of how, in the digital age, content industries may mutate, at least in part, into service companies. The key product is not only the song; it is also the speed, reliability, and convenience of access to it.11
There is also a more general point here about the relative power of law and business models. Although legal prohibitions against copying are useful against large-scale pirates (e.g., those who would post MP3 files for sale and hence have to be visible enough to advertise), they are unlikely to be either effective or necessary against individual infringers, where detection and enforcement are problematic. Where such private behavior is concerned, business models may offer a far more effective means of dealing with IP issues.
Use Digital Content to Promote the Traditional Product
Another business model approach sees free online distribution of music as a way to build the market for the traditional product. In March 1999 Tom Petty put a song from his new album online; it was down-loaded more than 150,000 times in 2 days. Other groups have made similar efforts, releasing digital versions before the albums were available in stores (Cleary, 1999), all in the belief that distribution of a sample track will increase sales of the traditional product.
11 This is an increasingly common observation. For example, the success of Web sites such as Amazon.com, Barnes and Noble, and others are attributable in part to the speed, reliability, and convenience of the service they offer, as well as the product available for sale.
In October 1999 the rock group Creed made the most popular song from their CD release "Human Clay" available for free download at more than 100 sites. The Creed CD jumped to the number one spot on the Billboard top album sales when it debuted, notwithstanding the free giveaway of what was believed to be its first hit.12 This phenomenon calls into question the conventional wisdom that, although one may choose to give away free songs as a promotional tool, to give away the most popular song will eliminate some motivation to buy the entire album. The utilization of the digital download of free songs in their entirety in order to drive sales of traditional "packaged goods," in the form of CDs and audio cassettes, is perhaps yet another example of the so-called "clicks and mortar'' business strategy, under which the Internet does not replace brick and mortar sales, but can be used as an adjunct to traditional brick and mortar retailing methodologies.
Give Away (Some) Digital Content and Focus on Auxiliary Markets
A more unconventional approach takes the position that most digital content is so difficult to protect that a more sensible business model would treat it as if it were free (Dyson, 1995):
Chief among the new rules is that "content is free." While not all content will be free, the new economic dynamic will operate as if it were. In the world of the Net, content . . . will serve as advertising for services such as support, aggregation, filtering, assembly, and integration of content modules. . . . Intellectual property that can be copied easily likely will be copied. It will be copied so easily and efficiently that much of it will be distributed free in order to attract attention or create desire for follow-up services that can be charged for.
The value instead is in the auxiliary markets. The classic example is the Grateful Dead, who have long permitted taping of their live shows (and have taken the additional step of permitting fans to trade digitized versions of those recordings on the Web) (Buel, 1999). They gave away their live performances (which were generally not released as records) and profited from the increased draw at concerts, and the income from related merchandise and traditionally produced studio recordings.13
12See Philips (1999).
13They were, however, rigorous at pursuing those who sought to profit from selling recordings of the live shows. Trading among fans was fine; selling the material was not. This is also an example of the suggestion above that distribution of live performance recordings might be handled by the artists rather than the record companies, although of course in this case the artists let the fans do it themselves.
The breadth of applicability of this model is of course not immediately clear. The model may be idiosyncratic to a particular band, audience, and tradition. But the thought is at least worth entertaining that the model might be used in other musical genres and perhaps in other publishing businesses, as well.
This variety of approaches illustrates the challenge and the opportunity of finding an appropriate business model in the world of digital IP. One challenge is in determining whether the existing models and existing industry structure can be made to work in the face of the new technology. In some cases it can; for example, movie theaters remain viable in the presence of VCRs and rental tapes. But in other cases the existing business model and industry structure cannot be maintained, no matter how vigorous the legal or technical efforts. In such cases some form of adaptation and creative rethinking can be particularly effective; the business model responses noted above offer a few examples of that type of thinking.
The Technical Protection Response
Many technical protection mechanisms are motivated by the key issue noted earlier, lying at the heart of the difficulties with digital IP: the liberation of content from medium. When content is bound to some physical object, the difficulty of duplicating the physical object provides an impediment to reproduction. However, when digitally encoded, text need no longer be carried by a physical book, paintings by a canvas, or music by a record or CD, so reproduction becomes easier. What options does technology offer for controlling reproduction?
Mark the Bits
One response is to "mark" the bits, that is, add to the content the digital equivalent of a watermark that identifies the rights holder. While it does not prevent the content from being copied and redistributed, this technique can at least make evident who owns the material and possibly aid in tracking the source of the redistribution.
Music can be watermarked by very small changes to some of the digital samples. As explained in Chapter 1, music is typically digitized by measuring the sound intensity 44,100 times a second, using a 16-bit number (0 to 65,535) to indicate intensity level. If the intensity of a sample is actually 34,459, it can be changed slightly and the human ear will never hear the difference. This permits encoding information in the music, by deciding, say, that the last two bits in every 150th sample will not encode the music, but will instead encode information about the music (e.g., the
identity of the rights holder). This change to the music will be imperceptible to a person but will be easily read by a computer program.14
A variety of such watermarking techniques is commercially available, including those that are '"robust" (i.e., difficult to remove without affecting the music) and those that are "fragile" (i.e., distorted by most modifications to a file, as for example compression using MP3). One proposed plan is to embed both a robust and a fragile watermark in newly released CDs and then require that licensed portable players refuse to play digital music that has a robust watermark but lacks the fragile watermark.15 The presence of the robust watermark indicates that the music is newly released, while the absence of the fragile watermark suggests that the music file is no longer in its original form (and hence may have been copied).
Reattach the Bits
A second, more ambitious approach is to find a way to "reattach" the bits to something physical that is not easily duplicated. A number of technical protection mechanisms are motivated by this basic observation; the description that follows draws on features of several of them as a way of characterizing this overall approach.
One relatively straightforward technique for reattaching the bits is to employ special hardware that enforces copy protection. Digital audio tape players, for example, have a serial copy management system (SCMS) in which the hardware itself enforces the prohibition against making digital copies of copies. The first-generation copy contains an indication that it is a copy rather than the original, and any SCMS-compliant device will not copy this copy.
This technique works well but is limited to single-purpose devices. It will not work on a general-purpose computer, because the user would be able to gain access to the original information and make copies by means other than the SCMS. As a result, the challenge of reattaching the bits becomes more difficult for a general-purpose computer. A succession of increasingly more sophisticated and complex mechanisms can be used to approach this goal:
• Encrypt the content. This mechanism provides, at a minimum, that the consumer will have to pay to get a decryption key; without it, a copy of the encrypted content is useless. Buy a song, and you get both an encrypted file and a password for decrypting (and playing) the song.
14See Chapter 5 and Appendix E for further discussion of watermarking and other technical protection mechanisms.
15See Robinson (1999).
• Anchor the content to a single machine or user. Simply encrypting the content is not enough, as the purchaser can pass along (or sell) both the encrypted content and the key, or simply decrypt the content, save it, and pass that along.16 There are a variety of ways to anchor the content; one conceptually simple technique encodes in the decryption key (or the song file) information about the computer receiving the encrypted file, such as the serial number of its primary disk. The decryption/ playback software then checks for these attributes before it will decrypt or play the song.17
• Implement persistent encryption. The scheme above is still not sufficient, because a consumer might legally purchase content and legally decrypt it, then pass that on (or sell it) to others who can modify the decrypted file to play on their machine. Some technical protection mechanisms attempt to provide additional security by narrowing as much as possible the window of opportunity during which the decrypted information is available. To do this the information must be decrypted just in time (i.e., just before it is used), no temporary copies are ever stored, and the information is decrypted as physically close as possible to the site where it will be used. Just-in-time decryption means that decrypted information is available as briefly as possible and then perhaps only in very small chunks at a time. Decrypting close to the usage site reduces the number of places inside (or outside) the machine at which the decrypted information might be "siphoned o." Persistent encryption is complex to implement in its most ambitious and effective form, because it requires the IP protection software to take control of some of the routine input and output capabilities of the computer. If this is not done, there are a large number of places (e.g., in the operating system) from which decrypted information might be obtained.
A variety of systems in use in 1999 employ one or more of these mechanisms, including a2b from AT&T, the Liquid Audio player from Liquid Audio, the Electronic Music Management System from IBM, ContentGuard from Xerox, the InterTrust system, and others.18
16Encryption by itself does convey to the user that he or she is not supposed to pass along the content casuallythus, it could have some deterrence effect in helping to keep honest people honest, at least for the present, when encryption is still a novel technology for most people. More about this is presented in Chapter 5.
17Hard drives are of course sometimes changed (e.g., to upgrade to a larger drive), so this technique is not foolproof. But the claim is that there are enough such stable attributes about a particular computer that this scheme will work most of the time.
18Information on a2b (both the company and the technology) is available online at <http://www.a2bmusic.com>; Liquid Audio, at <http://www.liquidaudio.com>; IBM's system, at <http://www.ibm.com/security>; ContentGuard, at <http://www.contentguard.com>; and InterTrust, at <http://www.intertrust.com>.
Another element of technology provides a useful additional capability. Differences among consumers make it useful to offer content on a variety of different terms. For music, for example, one might want to sell the right to a time-limited use, a finite number of uses, or an unlimited time and usage count. A variety of systems have been developed to provide an easy way to specify a wide variety of such conditions. When a music file is downloaded, then, in addition to the music it will contain information indicating the license conditions under which the music may be used. The playback software checks these conditions and enforces them appropriately.
Note that the technology picture outlined here is optimistic, in a sense discussed further in Chapter 5, where the significant difficulties involved in providing secure content handling in the real world are considered in detail. Note, too, that as with any security mechanism, the key question in the real world is not the purely technical issue of whether it can be defeated. All mechanisms can be, eventually. Instead the key questions go back to the three fundamental factors: technology, business models, and the law. The legal system sets the basic rules on what may be controlled; technology and business models then work in tandem: Is the technology strong enough to provide a meaningful disincentive for theft, yet not so expensive (for either the distributor or the consumer) that the added costs drastically curtail demand? At this point in the development of digital content delivery mechanisms, companies have relatively little real-world experience on which to make such judgments.
A system using some of the technology described above is easy to imagine; the details used here are intended to convey the general idea rather than describe any particular system. The user downloads and installs software that provides for music playback and assists with online purchasing. He or she then connects to a Web site offering music in that form, selects a song or album, and provides a credit card number to pay for it. As part of the transaction, the vending site is provided with information specific to the computer requesting the file (e.g., the serial number of its disk); this information is embedded in the decryption key supplied with the file. The vending site may, in addition, mark the music file with a unique ID that enables linking back to this transaction, in the event that the decrypted audio file is later found to have been distributed.
are maintained: The user can store large amounts of music compactly, have random access to any track on any album, and create personalized albums containing selected tracks from selected albums in a particular order.
The same basic model applies to portable players like the Rio: To anchor the bits (i.e., make them playable on only one portable player), a hardware identifier may be built into the file, such as the serial number of the player. Just as on the PC, the player's software can check for this information and refuse to play the song unless the identifier in the file matches the device.
Constraints on Technological Solutions
The scenario above sounds simple, but there are inherent difficulties. First, no protection scheme lasts forever. Any time content is valuable, some people will be motivated to find ways to break the protection mechanism, and some of them will be more than willing to share their techniques. For example, by the end of 1998 a program called a2b2wav that was available on the Web purportedly cracked the protection scheme then in use by both a2b and Liquid Audio, producing an unprotected file of the music in another audio format called WAV. Another example illustrates both the interest in digital music and the speed with which protection mechanisms are subject to defeat. On August 17, 1999, Microsoft released Windows Media 4.0, intended to be a secure format for music and other media files. On August 18, 1999, various Web sites offered a program that reportedly defeated the security features of Windows Media, stripping out the license information and making the files shareable. The new program had apparently been in development for only a month, based on the beta releases of Windows Media (Livingston, 1999).
One plausible countermeasure is to design protection to be renewable (i.e., easily changed so that the new protection scheme can quickly replace the old). This solution protects content from that point forward, limits the profits from piracy, and keeps the protection a moving target for those trying to break it.
A second difficulty in developing technical protection solutions is that consumer devices must be easy to use. Cumbersome content protection schemes may discourage use, particularly as consumers are likely to be impatient with mechanisms that they perceive are intended to protect someone else's interests. This requirement puts stringent performance demands on a system (e.g., decryption must be fast enough to become imperceptible) and requires careful design, to ensure that the system is conceptually simple.
A third difficulty arises with any system that "anchors" content to a specific device: the potential loss of all of that content if the device in question fails or is replaced. Does the consumer have to repurchase every piece of music that he or she owns if the portable player fails, is lost, or is replaced?
A fourth difficulty is the diversity of interests at work here, including the computer owner (i.e., music consumer), computer manufacturers (of both hardware and software), music publishers, and performers. Consumers have expectations about the ability to share and the ongoing use of content, publishers are concerned about the overall market, and performers are concerned about their audience and royalties. Getting significant content protection machinery in place and widely distributed would require a concerted and coordinated effort, yet each of the players has its own goals and aims that may not necessarily align (see, for example, Hellweg, 1999).
A fifth difficulty is the inherent complexity of providing end-to-end protection within a general-purpose computer. PCs have been successful to a significant degree because they have open architectures; that is, components of the machine can be replaced by the user (e.g., replacing a hard drive with a larger one or buying a new sound card19). As long as the machine is designed this wayto be accessible to usersdecrypted information can be captured in numerous ways as it passes from one place to another inside the machine. One could modify the software used by the sound card, for example, so that it not only generates the signal for the speakers but also stores away the decrypted music samples. Clearly, hardware and software designers could make such steps progressively more difficult, but the effort they must expend and the consequential costs would be substantial.
A sixth difficulty arises from the installed base of PCs. With more than 100 million computers in use, any scheme that requires new hardware faces a significant barrier to acceptance. Clearly the benefits have to outweigh the cost and inconvenience of changing machines. One benefit that may encourage the adoption of new hardware is the interest in electronic commerce. Efforts have been mounted to create new hardware and software with security built in at all levels.20 If this succeeds, PCs may routinely come with technology that makes possible secure electronic
19The sound card is the hardware that turns digital samples into the signals necessary to drive the speakers.
20The Trusted Computing Platform Alliance, a collaborative effort founded by Compaq, HP, IBM, Intel, and Microsoft, is aimed at providing security standards for computers, trying to create "an enhanced hardware- and operating system-based trusted computing platform." For more information, see <http://www.trustedpc.org>.
commerce, and that may also be usable for enforcing intellectual property rights.
A second mitigating factor here is the relatively short lifetime of computers, at least in the corporate environment, where 3 years is a common figure for turnover. A related opportunity arises with new technology: With portable players, a relatively recent development, there is not a major installed base requiring compatibility, offering the chance to set security measures in place near the outset of the new technology (as indeed the Rio player and others have agreed to do).21 The installed base problem here is more likely to arise from the need to be compatible with the existing MP3 format, widely and legally used by consumers to make personal copies of CDs they own and who are likely to want players capable of playing those files.
There is also a traditional chicken and egg problem involving technology development and content owners. Investing in new content delivery technology is risky without content to deliver, yet content owners are reluctant to release their content for digital distribution until they feel the delivery system has been tested in real use, is secure, and will be accepted by consumers. For example, in June 1999, Digital Video Express, the company that invented and was marketing Divx technology, ceased operations. Its announcement mentioned as one of the contributing factors its inability to obtain adequate support from studios.22
Finally, there is the problem of the digital infrastructure, one element of which is transaction support: Buying content online requires systems for secure transactions, in high volume, possibly involving rather small amounts of money (e.g., $1 for a song). A second element of infrastructure, public-key cryptography, is effective at protecting content but requires a substantial infrastructure to make it easy to use on a wide scale (see Appendix E for a description of public-key certificate authorities). Although some progress has been made on creating the infrastructure to support electronic commerce, these and other elements of this digital infrastructure are not yet in place for routine use and will require both a sizable investment for their creation and a major effort at agreeing on standards.
Industry Consequences of the New Technology
The digitization of music in general and the availability of an easily used format like MP3 in particular have wide-ranging consequences for
21See Strauss and Richtel (1999).
22See Ramstad (1999).
the industry, consequences that are being played out in a number of struggles.
One consequence is the possibility of a radical shift in power. As noted earlier, one of the fundamental changes brought about by the Web is the availability of an inexpensive publishing and distribution medium with worldwide reach. If composers and performers choose to take advantage of that medium, what is to be the role of traditional music publishers and distributors?
This phenomenon has been called disintermediation, referring to the elimination of middlemen in transactions. In the view of some, traditional publishers are becoming unnecessary, because authors, composers, and performers will be able to publish and distribute their product online themselves. Some performers have already done so, though generally offering their products as free samples. A variety of MP3 Web sites have also emerged, modeled on the notion of artists' cooperatives: Composers and performers can post their work and receive royalties, with no effort on their part other than the original posting.
Even as ease of publication may provide alternatives to and hence reduce the demand for one kind of intermediarytraditional publishers and distributorsit may simultaneously increase the demand for another kind. If anyone can be a creator and publisher, content will proliferate, producing a world of information overload. The consumer's problem will not be obtaining content, but rather wading through it all. This difficulty has long been recognized: Nearly 30 years ago Herbert Simon suggested that "a wealth of information creates a poverty of attention" (Simon, 1971). In the content-rich world, then, information intermediaries may become even more important because, although content may proliferate, attention is on an immutable budget.
But these intermediaries would not be publishers in the traditional sense; hence the phenomenon may not be so much disintermediation in general, as the diminishing need for one variety of intermediary and an increasing role for another. The new role of publishers in an information-rich world may require a different kind of company with a different focus. This scenario presents the possibility of a significant shift in the power structure of the music industry and a significant economic impact. Little wonder, then, that battles are emerging over the future character of the industry.23
23One example of a new kind of publisher is Garageband.com, an Internet start-up company that is attempting to bridge the competing worlds of online digital music and the traditional recording industry. Garageband's focus is on the thousands of small and struggling musicians and groups that have not been able to sign with a major recording studio and have discovered that they cannot make a profit by giving their music away over the
(footnote continued on next page)
Those battles are sometimes indirect, as for example struggles over music delivery hardware. In October 1998, the Recording Industry Association of America (RIAA) sued Diamond Multimedia, maker of the Rio. The suit sought to block the sale of the device on the grounds that the Rio violated the Audio Home Recording Act (AHRA) because the device did not contain a serial copy management mechanism. The request was denied on the grounds that the Rio was strictly a playback device; there was no way to copy from it to another device. This ruling was subsequently upheld by an Appeals Court in June 1999.24 The Appeals Court noted that the main purpose of the AHRA was "the facilitation of personal use," pointing out that "[a]s the Senate Report explains, '[t]he purpose of [the Act] is to ensure the right of consumers to make analog or digital audio recordings of copyrighted music for their private, noncommercial use.' S. Rep. 102–294, at § 86." It also mentioned in passing the notion of "space-shifting" as analogous to the "time-shifting" permitted under Sony Corp. of America v. Universal City Studios: "The Rio merely makes copies in order to render portable, or ''space-shift," those files that already reside on a user's hard drive. . . . Such copying is paradigmatic non-commercial personal use entirely consistent with the purposes of the Act."25
A similar sort of lawsuit occurred in March 1999, when the International Federation of the Phonographic Industry (IFPI) sued the Norwegian company FAST Search and Transfer over the use of its MP3 search engine and database, licensed to Lycos. IFPI claimed that the database consists almost entirely of MP3 files that violate copyright. The suit is
(footnote continued from previous page)
Internet. Musicians will be able to upload recordings that will be rated by music enthusiasts on the Web. Beginning in the year 2000, Garageband.com plans to begin awarding a recording contract for the most popular music each month. The company plans to attract listeners who are willing to sample unknown recording artists by offering prizes such as backstage passes to concerts and visits to recording studios (Markoff, 1999).
24The issue before the court concerned whether the Rio was a digital audio recording device subject to the terms of the AHRA (devices under the purview of the AHRA must prohibit unauthorized serial copying). The court indicated clearly that the Rio was not such a device, affirming in passing that computers are not such devices either, even though:
The district court concluded . . . the exemption of computers generally from the Act's ambit, "would effectively eviscerate the [Act]" because "[a]ny recording device could evade regulation simply by passing the music through a computer and ensuring that the MP3 file resided momentarily on the hard drive." RIAA I, 29 F. Supp. 2d at 630. While this may be true, the Act seems to have been expressly designed to create this loophole.
Diamond Multimedia later formed an alliance with Liquid Audio, a vendor of audio files in a protected format that makes songs easily playable only on the PC to which they were downloaded.
25RIAA v. Diamond Multimedia, F.3d (9th Cir. 1999).
novel in part in claiming a search engine firm as a party to copyright violation. The suit illustrates as well the mixed collection of interests involved: One of the backers of the suit is the Swiss company Audiosoft, which offers its own version of a secure container for digital music distribution (Robertson, 1999).
The attempt to ban the sale of the Rio illustrates several interesting aspects of the struggle over digital music. First, attacking the Rio on serial copying grounds is curious, because the device can't copy. Virtually all of the copying of MP3 files is done on computers, but (as noted) computers are not covered by the Audio Home Recording Act (because their ability to do digital recording is not "designed or marketed for the primary purpose of . . . making a digital audio copied recording"26). Hence the RIAA aimed at the available target, on the available basis, even if it was a tenuous match.
Second, the battle illustrates the difficulty of drafting law and policy that are technology specific in the age of general-purpose computers. The AHRA goes to some effort to define the particular class of device it attempts to regulate. But general-purpose computers can, by their nature, do anything with digital information, including the very things outlawed by the AHRA. Short of insisting on hardware that limits the behavior of the computer, little can be done to prevent this.
The difficulty is further highlighted by the announcement in October 1998 of a company called Empeg, in Somerset, England, that claimed to be bringing out a portable device that not only played MP3 files but also ran the Linux operating system. In other words, it was a portable MP3 player that was also a computer and hence outside the purview of the AHRA (Patrizio and Maclachlan, 1998).
A third interesting aspect of the battle is the indirection involved in going after the playback device, when the real effort is to discourage the proliferation of illicit MP3 files. The rationale is clear: "The RIAA claims that it is unable to stamp out the proliferation of pirated music download sites. . . . [Cary] Sherman [senior executive vice president of the RIAA] stated that the only viable solution to prevent free downloads is to attack the problem on the receiving end . . ." (Anderson, 1998).
Fourth, the battle over the Rio illustrates the difficulty of large-scale policing of private behavior. Trying to take the devices off the market might make relatively little difference even if it succeeds, as MP3 files were downloaded to personal computers and played in large numbers before the Rio, and would continue to be without it. Removing the Rio (and all the other similar players) might prevent the appeal of MP3 files
26Title 17, sec. 1001.
from increasing, but would do little to change the personal behavior of individuals using their own personal computers.
The final and perhaps most important point in the struggle over the Rio is that the real battle is over standards, because the future character and structure of the industry will be determined to a significant degree by what becomes the popularly accepted format for digital music. This fact is the major motivation behind the various industry consortia that have been formed to develop formats and standards (e.g., the Secure Digital Music Initiative (SDMI) formed by BMG Entertainment, EMI Recorded Music, Sony Music Entertainment, Universal Music Group, and Warner Music Group; the pairing of Sony and Microsoft; and others). If MP3 wins out, putting the genie back in the bottle will be difficult, and we will all be engaged in an experiment to see what effect unrestricted digital distribution has on the music industry.27 If one of the more secure formats wins out, the industry character and structure will likely suffer far less upheaval.
Control of the standard may also affect the disintermediation phenomenon, as control of a standard, particularly a proprietary standard, puts some degree of control over publishing into the hands of the standard owners. Control of the standard will have consequences for the consumer as well. A fundamental motivation for the SDMI effort is to produce a secure music format and corresponding devices that would refuse to play files containing unlicensed MP3 tracks. Note, too, that "winning" here means popular acceptance, much as it did in the battle of VCR formats between VHS and Beta. That in turn means that the battle will involve a wide variety of issues, such as technical adequacy, ease of use, amount of content available, and promotion by the various standard bearers.
The struggle to determine the future course of the industry naturally pits sites like MP3.com against conventional publishers. Yet signs of a convergence of interests have also appeared: In June 1999 MP3.com signed an agreement with the American Society of Composers, Authors, and Publishers (ASCAP), the largest of the performing rights societies, to license the streaming broadcast of music from ASCAP's catalog.28 This agreement allows MP3.com to act as a radio station of sorts, making songs
27There is also a successor to MP3, called MP4, that provides a framework in which IP protection information can be specified. It does not protect IP but does provide a place and a language for specifying the rights holder's desires concerning protection. Software to play MP4-encoded music would have the task of enforcing those desires, but at least there is a place in the music file itself for such information to be recorded. See Chapter 5 for additional discussion of MP4.
28See Thomason and Pegoraro (1999).
available on demand for playback from the site, but the bits are not as easily captured and stored, reducing the opportunity for piracy. This service is new for MP3.com, which previously had focused on allowing users to download MP3 files. The interesting element here is the careful dance of rapprochement engaged in by these two representatives of seemingly opposed interests, an apparent attempt to bridge the gap between the traditional world of music copyrights and royalties, and the proliferation of free music. Yet in the long term the interests are in many ways aligned, as each can benefit from the wide distribution of music and growth in the market.
One additional and possibly wide-ranging consequence of the new technology is unbundling of content. Music is typically downloaded by the track, not the album. This practice has become widespread in part because of size (downloading an album is still slow) but may also represent a challenge to the current marketing model, which emphasizes selling albums, not singles. It is common experience that not every song on a CD is equally appealing, but without a way to pick and choose, consumers had no choice but to buy the package offered. Hence the new technology may be promoting a "new" business model, in which content is more easily unbundled and as a consequence marketed and sold in smaller chunks.29
The Broader Lessons
Three general lessons can be learned from the early years of the struggle over digital music. First, what has happened there may happen in other content industries as well, as other products become digitized.
Books and movies have begun to feel the effects. Electronic books are appearing, with several Web sites selling full-length books in digital form, while others offer reloadable book-sized portable display hardware.30 This development is made possible in part by the creation of more secure forms for the content, of the sort provided, for example, by Adobe's use of ContentGuard technology from Xerox, resulting in a more secure version of its popular PDF document format.
With electronic publishing of books come many of the same issues concerning the role of the publisher: Various sites on the Web offer electronic publication of works, presenting an alternative to traditional pub-
29At one time, of course, singles were a viable part of the music market (as 45 rpm records); perhaps that time has come again.
30Electronic books are available online at sites such as <http://www.fatbrain.com and http://www.1stbooks.com>, and books and a portable reader are available at <http://www.rocket-ebook.com>. For a general discussion, see Clark (1999).
lishers, while also promising 40 to 50 percent royalty rates and author retention of copyright, practices far from common in the print publishing business.31
Movies in digital form are currently saved from widespread illegal copying because of their large size, but this barrier is likely to be overcome before too long. A number of sites have begun already to sell full-length movies in digital form,32 but at upwards of 200 megabytes for a (compressed) movie, and 5 megabytes for even a trailer, the space requirements and download times are still quite substantial. Others are exploring the possibility of Internet distribution of movies.33 Digital movie piracy has also appeared; in 1999 pirated copies of "The Blair Witch Project," "The Matrix," and ''American Pie" were all available online. These copies are relatively low-quality, still sizable to download and store, and not easy to find (they are generally traded in low-profile news groups and chat rooms). But the struggle over digital movies has clearly arrived and will grow worse as storage capacity and transmission speeds increase.
The second lesson is that struggles over protecting intellectual property take many forms and reach into a variety of areas, including battles over technology, standards, industry structure, and business models. Keeping this in mind often makes it easier to decode the disparate agendas and strategies of the many players engaged in the struggle.
The third lesson is that among the various battles, the struggle over standards is often the most intense as it typically has the most far-reaching effects, with consequences for authors, publishers, and consumers alike, as well as the shape and character of the industry.
31See, for example, Fatbrain.com's e-matter, online at <http://www.fatbrain.com/ematter/home.html>for an example of these rates, effective in September 1999. For background on the struggle over electronic rights to textual works, see Zeitchik (1999).
32For example, see Sightsound at <http://www.sightsound.com>.
33For example, Metafilmics, an established filmmaker that won an Academy Award for best visual effects in 1998, announced it is making "The Quantum Project" specifically for initial Internet distribution (Pollack, 1999).