It is a music-lover's dream come true. Type into your computer the name of an artist and the title of a song, and within a few seconds a list of matches pops up. Each entry in the list provides a link to the song in question, stored somewhere on the internet. A couple more clicks, and a few minutes later the song has been downloaded and is blaring from your computer's loudspeakers. Hundreds of thousands of tracks by thousands of artists can be accessed from this global jukebox, called Napster, which is available around the clock, and free. No wonder 20m people are using it-and no wonder the record industry is trying to shut it down.
The case came to court in San Francisco in July. Napster lost. But the service is still operating, pending an appeal, which will be heard in early October. However, in many cases, the verdict is already irrelevant. Whatever happens, Napster, or at least the principle it represents, has won. Even if the plug is pulled on Napster, there are dozens of similar services waiting to replace it, most of which have been deliberately designed to be far more difficult to shut down. The genie is out of the bottle: file-swapping is here to stay. This is bad news for all those trying to protect intellectual property, be it music, software, movies or books. Napster and its like will increase the pressure on record companies, software firms and publishers to switch to new ways of doing business. They will also make it harder than ever to control or censor internet content.
Napster is significant for other reasons, too. It is the best illustration to date of the way in which the internet can overturn long-established business practices practically overnight (it was launched just over a year ago). It is also part of a broad trend towards a less centralised internet, as users and programmers discover what can be done by linking together thousands of ordinary PCs. Pooling the storage, communication and processing capacity of machines distributed around the network makes it possible to assemble virtual supercomputers capable of extraordinary things. To some extent, the result is a return to the internet's original communitarian spirit. As recently as ten years ago, most machines on the internet talked to each other as equals; scientists and academics used the system to exchange information and download files from each other. But since the commercialisation of the internet started in earnest, in about 1994, the trend has been towards growing domination by an ever-smaller number of important sites. According to a 1999 Media Metrix study, one fifth of all online time is spent visiting the ten most popular sites. In 1998, the figure was one sixth. Such is the dominance of the most popular sites that, by one estimate, Yahoo! accounts for five minutes of every hour spent by a typical user on the web.
Share and share alike
Napster's technical cleverness and legal doubtfulness are closely linked. If you take a song off a commercial CD, translate it into the MP3 format which has become the standard for encoding music on the internet, and then post it on your personal website for anyone to download, it is pretty clear that you are doing something wrong. The threatening letters from the record company will start rolling in; the company on whose web servers you have posted the file will probably remove it at the first sign of legal trouble.
But Napster is different. Each song in the system has no particular location online; instead, there are multiple copies of it, sprinkled all over the internet. The files are not stored on servers, but on the PCs of individual users, who connect to the Napster network as and when they feel like it. While they are connected, they can access other users' song libraries, and other users can access theirs.
In order for all this to work, however, users must be able to find what they are looking for. So Napster, which was developed by a 19-year-old American called Shawn Fanning, provides a centralised directory of who has what. The company's own computers keep track of which users are connected, and which files they have on their PCs. If you are looking for a song, you can then find out which other members of the collective who are currently connected have a copy of it. You choose one of them, a connection is established between your PC and theirs, and the file is transferred.
Because the file-swapping goes on between individuals, and the music files themselves never pass through Napster's own computers, the company argues that it is merely acting as a matchmaker, and cannot be held responsible for the fact that the vast majority of the files in the system have been pirated from CDs. And because multiple copies of each song exist all over the internet, it would be impractical to try to find and erase them all-the record companies can hardly take 20m people to court. So instead, the Recording Industry Association of America (RIAA), acting on behalf of the big record companies, took Napster itself to court in July in the hope of shutting down its central directory service. Without the ability to locate files on other users' computers, the system cannot function.
Neither side presented a convincing case. The RIAA argued that Napster encouraged the violation of copyright law and that CD sales had fallen as a direct result. It brandished dubious statistics purporting to show that sales in record shops near university campuses, where Napster use is rampant, had fallen. In fact, the evidence on how Napster affects its users' CD-buying habits is far from clear. One survey found that Napster users are 45 per cent more likely to have increased their overall music purchasing than non-users. On average, users were found to have only ten songs in their file libraries, which suggests that they are using the service as a temporary jukebox, rather than a substitute for buying CDs. And sales of CDs in the US reached an all-time high in the first six months of 2000, just as Napster took off.
For its part, Napster argued-somewhat implausibly-that use of its software counts as "fair use" under copyright law, according to which it is legal to make a copy of an album on cassette and then lend it to a friend. The judge hearing the case rejected this argument, because fair use covers exchanges between friends, not millions of anonymous users. In any case, it is clear from internal e-mails between Napster executives that they were fully aware of the system's main use-to facilitate the exchange of pirated music. A recent string of rulings against internet firms and in favour of record companies suggests that things are not looking good for Napster.
Caught napping
But if Napster is ordered to close, its users can be expected to migrate to various sons-of-Napster within days. This could actually make things worse for the record companies, because these followers will avoid Napster's technical and legal Achilles's heel: the central directory service. Instead, they are totally decentralised. They are not operated by companies who can be sued; they do not rely on central computers that can be unplugged.
The first of the pretenders to Napster's throne is Gnutella, which was created by a 21-year-old programmer called Justin Frankel and was posted on to a website for just a few hours before being withdrawn. But that was long enough for thousands of people to download it and spread it around the internet. The software is now being maintained by an international group of volunteers. There is no organisation behind Gnutella, so anyone hoping to shut it down has nobody to sue.
To join the Gnutella network, you need to find a few people who are already members of the club. Your computer, with the Gnutella software installed, then connects to their computers, and talks to the wider Gnutella network through them. If you are looking for a particular file (a song perhaps, although Gnutella, unlike Napster, can distribute any kind of file) your computer starts by asking its immediate neighbours on the network if they have the relevant file. Those neighbours, in turn, pass the request on to their neighbours, and so on. Thus the search request propagates across Gnutella space. Searching takes longer than it does on Napster, because there is no central directory; but this also means that the service cannot be shut down.
A variation on Gnutella is Aimster, which uses America Online's messaging service to do the matchmaking between users who wish to exchange files, and then uses Gnutella to perform the actual transfers. From a legal point of view, Aimster is particularly cunning because it has all the advantages of Gnutella yet is designed for use by small groups of people who already communicate via instant messaging. This means that exchanging music files using Aimster may fall within the "fair use" provision of music-swapping between friends.
Another approach is Freenet's, which was conceived as a secure information storage system by a 23-year-old Irish programmer called Ian Clarke. He has created a system in which information can be posted and retrieved anonymously, and in such a way that it cannot be censored or modified.
One problem that these file-distribution systems have is that not all users are prepared to muck in. Many users download files from others without making their own files available in return. A study by the Xerox Palo Alto Research Center in California, which looked at the behaviour of Gnutella users, found that 70 per cent of them offered no files to download. Just 20 per cent of Gnutella's users provided 98 per cent of the files in the system, and the most generous 1 per cent provided 40 per cent of the total.
Hence the approach adopted by Mojo Nation, another file-swapping system that tries to prevent free-riding. Users of the system earn credits-called Mojo-by contributing and distributing files; these credits can then be spent on downloads. This opens the way to real-world payments for downloads, by establishing a link between Mojo and real money. Mojo Nation's creators say they are keen to work with copyright holders; their system, unlike Gnutella or Napster, has the basis of a payment mechanism built-in. But such arguments are unlikely to cut much ice with the RIAA or Hollywood.
Wake-up call to the music industry
It is obvious why the RIAA wants to stop file-swapping; but in some ways it is missing the point. The music industry has been slow to embrace the internet. It has spent years squabbling over its own secure music-distribution format, thus leaving the way clear for Napster. Napster's success should be a wake-up call to the music industry, for it has clearly identified a gap in the market. Although users enjoy being able to listen to free music, it is popular for other reasons, too-reasons which the record companies, and other companies which rely on protecting intellectual property, would do well to acknowledge.
Much of Napster's appeal stems from its convenience. Its "universal jukebox" facility is very attractive; in one survey, 60 per cent of college Napster users said they would be quite happy to pay a monthly subscription of $15 for it. But anyone who wants a universal jukebox now has no choice but to break the law and use Napster, because it is a service that the world's record companies do not provide.
Offering a huge library of recorded music online, on a subscription or pay-per-song basis, in a fast, reliable, searchable database (Napster et al are still not exactly user-friendly), would ultimately be the best way to defeat piracy, because it would make the legal option more attractive than the illegal option.
The same is true of other forms of intellectual property. Publishers of electronic books, which are expected to account for 10 per cent of US consumer book sales by 2005, will soon be grappling with the same problem that the record companies face today. Similarly, as high-speed internet access becomes common, it will become feasible to pirate entire movies, too.
Fancy protection systems with encryption and digital envelopes are not the answer. To prevent piracy in the digital age, whether of music, books, movies or software, publishers will have to switch to new business models which make the legal alternative more attractive. In the case of the music industry, the model is clear: a universal jukebox with an all-you-can-eat monthly subscription. Similarly, the software industry is moving towards a model where software is given away, and firms make money by charging for services.
The power of collaboration on the net
The fuss that Napster and its offspring have caused shows what can be done by harnessing the collective power of millions of individual machines. Napster-like systems (known as distributed storage networks) are aggregating storage space on users' hard disks to produce an enormous communal storage system. But disk space is not the only abundant resource which can be aggregated via the internet. There is also an enormous amount of processing power going to waste, because most computers spend much of their time doing nothing. And, with the right kind of software, this resource can also be tapped, through a technique called distributed processing.
By far the best known example of distributed processing is SETI@home, a US-based project which looks for evidence of extraterrestrial life by analysing radio signals from space using home computers. The software takes the form of a screen-saver which activates when the computer on which it is installed is not doing anything. Rather than wasting processor cycles drawing flying toasters on the screen, this software scrutinises a chunk of data from the Arecibo radio telescope downloaded from a central clearing-house, and reports back if it finds anything unusual.
So far, SETI@home has yet to find any aliens. But with 2m computers around the world running the software, the system constitutes the largest computer ever built. Unlike most computers, it is only capable of doing one thing. But its collective power is staggering-about 10 times greater than any existing conventional supercomputer. As with Napster, individual machines join and leave the network continually. But the combined system keeps on functioning-and keeps on ploughing through the Arecibo data. Its success has inspired several firms to look into the possibility of exploiting distributed processing for commercial gain. The idea is to build a virtual supercomputer, like the SETI@Home system, and then rent it out to companies which need large amounts of computing power only occasionally. Members of such collectives would be paid for contributing their spare processing power; in other words, your PC could earn its keep by working for someone else during the night.
This vision is being pursued by start-ups including Parabon of Fairfax, Virginia; Popular Power in San Francisco; and Distributed Science of Toronto. Each company has developed its own distributed processing software, has recruited thousands of users, and is now sniffing around for paying clients. Proof-of-concept calculations being carried out already include modelling of influenza-vaccination strategies and an analysis of designs for nuclear-waste storage canisters.
Another application for which distributed computing would be well-suited is the improvement of internet search engines. At present, a large search engine such as Google or Alta Vista must revisit all of the billion-or-so pages in its index every so often to see if anything has changed, and then update its index accordingly. Because downloading a billion pages takes a long time, most pages only get revisited every few weeks. Distributing the job among thousands of machines sprinkled around the internet would make much more sense. Each machine would only have to visit a tiny fraction of the total number of pages, and would then report back to the central search engine, letting it know which pages had changed.
Strength in numbers
The potential uses of the millions of individual machines plugged into the internet are enormous; today's systems merely scratch the surface of what might be possible. But the result is that the internet is changing from a system in which individual machines sit on the fringes and consume content from the centre to one in which collaboration and communication between machines is becoming common.
The emergence of distributed systems such as Napster is a welcome reversal of the trend towards centralisation. In biological terms, it is akin to the evolution of the first multi-cellular organisms, which were far more versatile than their single-celled predecessors. Distributed systems have already turned the music industry upside down. Who knows what they will evolve into next?
