I'll bet plenty of people have heard of the great new thing in computing called peer-to-peer and are wondering what it's all about. It goes way back to the dawn of networking. In fact, the Windows operating system has had peer-to-peer networking capability for well over a decade. So what's the big deal now?

As you might suspect, most anything that gets sucked into the Web tends to become augmented, amplified, or otherwise transmogrified. This, I think, is what's happening with peer-to-peer networking. It starts with a very simple concept: People on a network agree to share computer resources (e.g., disk space, processing time) more or less directly. You have a file that I want, and if I have the appropriate clearance, I can go out and download it from your computer. In effect, your computer acts as a server for that file and other resources. Mine can do likewise for resources that I permit to be used in this way.

Since in larger networks it's difficult to know what resources are available, there is often a special server that is used to coordinate what's out there--sort of a master directory of resources that acts like a clearinghouse to the system.

This describes the way Napster operates. Music files are listed with Napster; you use Napster software to view the list and then make a connection with the computer offering the file and download it. You can also offer your own music files. With around 20 million users, it is the largest peer-to-peer network the world has ever seen. As I said, the Web has a way of amplifying things, and this expansion in volume suddenly makes a simple concept seem quite different.

On the other hand, Napster allows people to share music illegally. At least, at the moment it appears to be illegal, since the music industry, which brought the suit against Napster, and has been winning the court battles. The music industry and a number of recording artists object to having their music transferred willy-nilly from person to person without a tuppence of royalty in it. Or put another way, if everybody were to get their music free through Napster, would they ever go out and buy a recording? The industry can smell those billions of dollars flowing untouched by their fingers in much the same way that the software industry fulminates about pirated software. Not being geeks at heart, the music people like it even less.

The Napster controversy is yet another chapter in the "open Internet" saga of free access to information (and just about everything else) versus the right of intellectual-property owners to receive reward for their efforts. It takes very little logical facility or real-world experience to see that this is an argument with a case on both sides and lots of shades of gray.

The legal issues, fascinating as they are, are not the only point of the Napster exercise. Peer-to-peer networking as a method of information distribution and multiplication of computing power may be more important. Sharing music files is hardly the ultimate use of the technology. Almost any kind of file can be shared, as can information packaged in a number of formats, such as programming objects. The Napster model, with a server that mediates resource location, is but one of several architectures for peer-to-peer, and I expect that more subtle variations will arise as a result of all the fuss over copyright protection.

The overall effect of peer-to-peer file sharing might be to blur the distinction between formal content providers (the big Web sites) and informal providers (you and me). To use two of the most awful words of the past decade, it may enfranchise the computer of anyone using the Internet and empower each person to find and share resources like never before. It could also be a mess. A key part of the music model used by Napster is that people generally know what they want. They want Shania Twain, or the Beatles, or the latest by the Backstreet Boys. However, the realm of peer-to-peer can quickly become as vast, confusing, and unusable as the Internet at large.

Then there's the use of CPU power. Peer-to-peer networks are much friendlier to distributed computing than other approaches. By distributed computing, I'm talking about using the processing time available on most PCs. From a CPU point of view, we use only a tiny fraction of their capacity. Our CPUs could easily be working on other problems and applications and still do what we want them to do when we want it. Peer-to-peer networks make it relatively easy to parcel up larger computational tasks and farm them out to thousands, if not millions, of CPUs in the network.

This is a kind of massively parallel processing that is already being used by scientific research groups (like SETI@Home) and for various kinds of commercial processing. We're starting to see a class of software, such as Applied MetaComputing Legion, that not only provides peer-to-peer services, but does so across operating systems and hardware platforms. Much of the new software is aimed at augmenting the usual peer-to-peer services, making them more sophisticated and secure.

Security is a huge issue. In a peer-to-peer network you're talking about possibly millions of open computers, all belonging to individuals, that could be compromised by a single security glitch or even a hardware error. While security control, network management, and data integrity are a big part of most peer-to-peer systems, in general it isn't nearly enough. I can easily foresee the day when accomplished hackers turn their attention to this burgeoning technology and shred it. It won't be pretty, and it will be public.

Nevertheless, it's a technology that's attracting attention. As a sign, we have the interesting sideshow of a few U.S. senators (Sen. Orrin Hatch of Utah in particular) suddenly writing letters to the Napster trial judge indicating that not all of Congress is happy to see the legal weight of the government pressing against Napster and its ilk. I thought Napster was a college kid-revolutionary free-Internet kind of thing, and Hatch is not noted for siding with the free-wheelers. What's going on here?

As usual, it's the money. It's not coincidental that the CEO of the company that makes more CPUs than anybody (Intel) said recently that peer-to-peer technology is "something IT managers are going to have to worry about" and that we need to be using the idle processing power of PCs and servers to handle business applications. While self-serving, his comments point in the direction a lot of companies would like to take peer-to-peer technology. That's why numerous computer-industry consortiums wrote friends-of-the-court letters to the Napster trial judge to warn of the damage to the concepts involved (even if they wouldn't come out directly in favor of what Napster was doing).

I really don't know how important peer-to-peer is going to be. It's not much of a technology, but what I said about the Internet doing strange things to technologies isn't all B.S. I do know that as we go along, the Internet--or more accurately, the people providing and using it--are engaged in a constant, feverish attempt to find out what it can do. Peer-to-peer services, like distributed applications, may develop some unique characteristics in the vast matrix of the Internet. We may find social or economic benefits that aren't apparent on a company network, or aren't even predictable from the basic concept of the technology. That's what the success of Napster seems to hint at.

On the other hand, peer-to-peer could be as dead as push by this time next year. (You remember push technology: It was supposed to send you information before you even knew you needed it. It didn't and you don't.) I do think peer-to-peer has some real uses. However, there are also some real barriers: legal, technical, and economic. Peer-to-peer is no joke, but I'm inclined to smirk; we're a long way from the last laugh.

Editor at Large Nelson King also writes Enterprise Pursuits every Tuesday on computeruser.com.