Forward to the Past

Copyright, Performance, and the Net


It is the contention of this note that the most pressing legal problem of the Information Age, namely just compensation for intellectual property transmitted through the Internet, is in fact not new at all and has already been solved: the solution is to be found in performing rights organizations. If the solution outlined here were adopted by the music industry, music piracy on the Internet could be completely eliminated almost instantaneously.

The Fundamental Problem of the Networked World that once information is made available to the public on any machine on the network, it can no longer be controlled. Thanks to remailers, proxy servers, mirror sites, privately-owned computers, networked file sharing, informal store-and-forward systems, public-key crpytography, and the like, there is no effective way to determine who has access to anything, let alone limit access to any particular individual or group.

Even cryptography is, ultimately, ineffective, since once a user has decrypted a copy of some document, nothing prevents them from re-distributing it (after re-encrypting it to prevent detection). Similarly, unique document identifiers and cryptographic signatures are easily removed or replaced.

The current instantiation of this problem is distributed filesharing schemes such as Napster, Gnutella (which lacks a centralized database), and Freenet (which adds encryption-enforced annonymity). But in fact, and despite the recent spate of news stories, this problem is not new: the Usenet store-and-forward news system has been raising the same problem for over a decade, and users could set up filesharing schemes with NFS or Microsoft's SMB long before Gnutella made its appearance.

This fundamental problem raises two more specific questions: how do authors receive just compensation for the intellectual property they create (``who pays whom for what?''), and which locality's laws should be applied in any particular case (``where is cyberspace?''). It is my intention to answer the first of these questions in this paper. (I suspect that the answer to the second can be found in admiralty law, but that's a topic for a later essay.)

In brief, the key to solving the electronic distribution problem is to treat the operator of an HTTP server, not as a publisher, but as a broadcaster; to treat moving a file across a network not as ``making a copy'' but as a performance. This view of the situation leads to a solution which is not only fair to both users and authors, but which completely eliminates the need for piracy. The imagined threat of Napster and so on could be eliminated overnight. As we will see, even the recording companies (presently pursuing Napster with lawsuits) will ultimately benefit.

Who Pays Whom, and for What?

The problem of electronic distribution of copyrighted material is not new, and did not originate with the advent of the Internet. The problem first appeared when the radio was invented, and its solution was invented shortly thereafter. This solution actually takes advantage of the fact that it is impossible to identify, or even to count accurately, all of the individuals who may listen to a song when it is performed.

Electronic Publishing = Broadcasting

Musical works were the first copyrighted works to be distributed electronically, via radio, about 90 years ago. It is only reasonable, then, to look to the music industry for the solution to some of today's problems.

The solution the music industry came up with is the performing rights society. ASCAP (the American Society of Composers, Authors and Publishers) was the first in the US, and was founded in 1914. The solution worked then, and (as we will see below) would work even better in today's world of digital networking.

Who Pays Whom?

In all cases, royalties are paid to the creators (authors, composers, lyricists) of a work by the publisher or the performance venue (i.e., the broadcaster), and they are paid in turn by the public at large. Publishers are paid directly in exchange for works in ``tangible form'' on media such as CD's, tapes, or books. The publisher then pays the creators a ``mechanical license fee'' in exchange for the right to make and sell copies of the work.

In the case of performances, however, the payment is indirect: the venue collects money either directly from the public (in the form of ticket sales or subscriptions) or indirectly from advertising, and pays the authors and their publishers through performing-rights societies such as ASCAP and BMI.

Composers and songwriters are compensated for their works by means of a statutory license fee which amounts to a tax on the performance venues. This tax is split between the publisher, which owns the distribution rights, and the creator, who owns the copyright. (The BMI contract specifically forbids its members from producing work for hire under the name they use for their own work.)

An advantage of this indirect scheme is that the performance venue is free to subsidize their listeners by writing the license fees off as a business expense. An additional advantage is that, because the performing rights societies take care of collecting and redistributing the fees, venues are not required to keep track of individual audience members or even, in many cases, of individual performances. Hotels, for example, are charged for background music according to the number of loudspeakers in use, and the fees collected from them allocated according to a sampling of broadcasters' playlists.

How Would This Work in Cyberspace?

There are really two ways to do it.
  1. Pay royalties out of a ``tax'' on bandwidth. This would be collected by ISP's, and the tax collected from a user would be proportional to their ability to download copyrighted works. It essentially puts ISP's in the position of ``broadcasters,'' and is essentially the same as the current taxes on DAT and CD-R media for consumer use.

    (Note that this is still possible in spite of the fact that there is currently a moratorium on taxing the Internet, because the ``tax'' is really a license fee imposed by the performing rights societies, rather than a true tax imposed by the government.)

    This fits well with a world in which network access is ``too cheap to meter'' and most users pay, as they do now, a fee based on the bandwidth of their connection and the amount of time they spend on-line. In other words, people pay not for the specific items they actually use, but only for what they could have used. It has the huge disadvantage, however, of either increasing the cost to users or decreasing the profits of the ISP's, many of whom are already operating on razor-thin margins.

  2. Royalties are paid by the companies that serve the data; typically they will make their profits as they do now, by selling advertising and physical goods such as books and records, or by performing other services. They could even sell subscriptions, as public broadcasting stations do. This puts the servers (web sites) in the position of ``broadcasters'' and is much closer, not only to the way things are done now on the web, but to the situation in radio and TV broadcasting in which the broadcasting companies are also in the best position to collect advertising revenue.

    One of the advantages of this approach is that, because HTTP requests are logged, it is easy to allocate royalties to the authors. A second feature of this approach is that the publishers could split their share of the royalties with the operators of caches and mirror sites, reducing the load on both the primary servers, and the network connections of the major private networks (such as universities and AOL, most of which already operate caching servers).

    What appears at first sight to be a disadvantage is that content on non-profit or privately-owned sites (for example, a musician's personal web site in a home studio, connected by a DSL line) does not bring in any revenue. The typical solution will be links to ``publishers'' like that host the content in exchange for the publishers' share of the royalties (50% in the music world).

In either case, revenue and server logs would be collected by performing rights organizations on behalf of the copyright owners. And in either case, users do not pay for each download. They pay indirectly, either through a uniform bandwidth assessment on their ISP's, or even more indirectly through their purchases from advertisers.

The Payoff

The greatest benefit of a system in which users do not pay for the specific items accessed is that there is no incentive for piracy. It costs no more to request something from its rightful owner than to get an illegal copy from a friend. In fact, it costs less in time and effort to get files directly from a publisher's well-connected server than to get them from another user through a file-sharing program like Napster or Gnutella.

Similarly, if the network bandwidth is sufficient there is little need to retain copies, and so the creator of a work will be compensated according to how often it is actually used. Nevertheless, many people will want to buy their own copies on actual CD's, and both the player software and the websites that serve the content will make this very easy indeed. In spite of initial fears, radio airplay increased record sales, and TV similarly increased the sale of movies on videotape.

But even in the unlikely event that the recording industry loses some of its profitability because of the Internet, the music industry will prosper. Users will download free music and look at the enticing ads, and occasionally buy physical recordings (conveniently, at the click of a button). Singer-songwriters will upload their creations to servers, and the more popular ones will be well compensated, just as they are now for radio airplay. The rest of us will at least earn enough to keep us in guitar strings and hard disk space. In effect, music broadcasting will simply move, over time, from the ether to the Ethernet, widening its audience and increasing revenues in the process.

Fair Use

While it's true that sharing files using a system like Gnutella can be considered a form of copying, this copying is taking place between individuals, and so falls into the area traditionally considered ``fair use.'' Like the sharing of tape recordings, this use has a cost to the individuals doing the sharing. But unlike a tape, the cost in time and bandwidth for sharing a file is twice as high as the cost of simply downloading it: it takes exactly as much time and bandwidth to send the file as it does to receive it.

This means that filesharing systems like Napster and Gnutella will be roundly condemned by network administrators as a waste of bandwidth, and by users as a waste of time; soon they will mostly disappear. The few that are left will be rethought as distributed storage systems and used primarily as replacements for backup on magnetic tape (which will also disappear, because the price per megabyte is decreasing faster for hard drives than it is for any removable medium).

... And Back to the Future

In time, authors, playwrights, screenwriters, and even programmers may start to wonder why they can't use the same system.Indeed, there is no reason why a similar scheme can't be applied in other fields. The details are likely to differ by genre, for a variety of reasons.


Some distributors of open source software, including Linux, already pay royalties for the software they sell on CD-ROM. This royalty comes out of the profits on the CD-ROM, so that nobody (not even the distributor!) makes any money on giveaways. One big difference between this and the music industry is that the royalty rate for software is not set by legislation, so nobody is making distributors pay. In particular, nobody pays royalties for free downloads.

Another difference, though, is that the typical software package is much smaller than the typical song, which takes 10MB/minute. Instead of a dozen or so songs with a handful of songwriters, a Linux distribution may contain thousands of programs, each with a different author. Furthermore, all are under open-source licenses, which specifically permit free distribution.

On the other hand, there are additional benefits for software authors. For example, RedHat and VA Linux allocated IPO stock to software authors, and even to people who submitted bug reports.

For these and many other reasons it is likely that the software business, and especially the free software business, will continue to be essentially chaotic. It's unlikely that any more formal solution could evolve without doing violence to the gift culture prevalent in the open source community.

Graphics and Multimedia

Although they originated several thousand years ago, multimedia works have enjoyed an unbroken four-hundred-year tradition in the Western hemisphere under their Italian name of opera. The ``video'' tradition goes back even farther, arguably at least as far as Greek drama. Although images are older still, they remained tied to physical media (whether cave wall or canvas) until the invention of photography in the nineteenth century.

All of these media share with audio the property that they are fairly large, with sizes ranging from a few hundred kilobytes for a scanned image to several gigabytes for an MPEG-compressed movie. At least on the high end of this range, such works are currently impractical to download.

They are no longer impractical to store, however. At mid-2000 prices, the hard disk space taken by a full movie is somewhat cheaper than the cost of a recordable DVD. The cost of an image on disk is about a tenth the cost of printing or copying the same image on paper.

But bandwidth to the home is increasing, and it is not too soon to consider what to do when high-bandwidth connections bring streaming video to the home into everyday use. The important thing to recognize is that fundamentally, nothing will change! The situation will be no different, really, from the present-day combination of broadcast TV and videotape, and there is no reason why the present-day combination of subscriptions (cable) and advertising can't continue to provide revenue for the movie studios and TV networks.


At first sight, the hardest problem appears to be text. While it may be true that a picture is worth a thousand words, it is certainly true that a full page image is worth at least a hundred kilobytes. A page of pure text contains two kilobytes at most. The time required to transmit it, and the space required to store it, are negligible. Even a full-length novel requires only a megabyte or so: a few minutes to transmit, and less than a cent to store.

The closest thing to performing rights organizations in the print world is the Copyright Clearance Center, which attempts to collect copyright license fees for photocopies of journal articles. Unfortunately these fees amount to several dollars per page, which means that text is overpriced by several orders of magnitude relative to music. Original print media are significantly cheaper: the price of a printed page is usually less than the cost of photocopying it, and comparable to the statutory license fee of a song. This suggests that an ``ad-per-page'' model might be successful.

Most commercial providers of text ``content'' on the Web have already embraced the performance model, supporting themselves by advertisements and occasionally subscription fees or contributions. Unfortunately this leaves the non-commercial provider and the few remaining individuals who don't want their personal websites marred by banner ads, out in the cold. We may have to wait for the establishment of the text equivalent of, a place where any author can upload a few essays or books and have the licensing revenue automatically collected.

Stephen R. Savitzky <>