Why Not Napster?
Why Not Gnutella?
Why Not Freenet?
Since we're bashing on the competition...
Some important distictions must be made between the ANet project and Napster. First, ANet is open-source and was not made to become a commercial product in any way, including from distribution and support. Also, ANet was not made specifically to share files, including MPEG 1 layer 3 (MP3) files. Finally, while both ANet and Napster are both peer-to-peer networking systems, only ANet has the ability to be able to implement a true distributed network.
Now, we can begin our analysis of Napster.
Napster was(1) the first peer-to-peer system made specifically to simplify file distribution. Its power came from the fact that all the file information was stored at a single source of information. This makes the process of distributing the file information and searching for a specific file particularly easy. Too easy, some would say.
From the start, Napster could have been simply a file information repository for any kind of file. But what made Napster so popular for its users, and unpopular for the music industry, was that it was made specifically for MP3 files. Thus it was optimized for the storage of the ID3 tags commonly found in the MP3 files, making it easy for the server to categorize the information by title, album, artist, year and so on. On a generic file repository, the information would be based on file names only, which wouldn't be particularly useful for searches(2).
So, even if the Napster servers only stored information about the MP3s, and not the MP3s themselves, it over-simplified and helped illegal distribution of copyrighted data. Not only that, but Napster didn't try to "filter" information about copyrighted files, with the result that they greatly helped illegal distributions. This resulted in Napster (the company) being sued by the music companies, and Napster lost.
Here, the estimated amout of money lost by the music companies are in the order of several millions of US dollars. This resulted in huge awereness of this kind of illegal activity by the general public that was somehow ignited by Napster(3).
True, illegal piracy in computing was and is always there, but never to this kind of magniture before. Usually, piracy is done within smaller groups, and is more the result of "sharing": you copy a file to someone that does not have it, and you receive back a copy of a file that you didn't had before. So, the servers that were set up to enable that "file sharing" are in great number, but in smaller, independant groups, making impractical to the authorities to close them and to sue the owners of the servers. With Napster, the music, the press and the general public had a single ennemy (or friend): Napster itself.
So, the problem that Napster had was that it had sole liability over their servers. Even if they were not copying themselves the MP3s, they had the total responsability of knowing what was going on through the help of their servers. The point is that even if computers are automated machines, their users are responsible of what their computers do through the user's actions.
Since the users were the ones illegally the information, why shouldn't they host themselves the servers? True, this is possible, but the answer to this question is a technical one, not simply a legal one. If the users are responsible, legally speaking, for the data they produce, they should also be responsible, technically speaking, of the distribution of their data, removing the need for a third party or a corporation (in this case, Napster) to be the source of the data distribution.
Clearly, the solution to legal responsability in data distribution must be to let the users themselves do the data distribution. But then, how do you do that?
The technical solution is something that a new generation of peer-to-peer network systems do that is called "distributed networking". While, in any terms, less efficient and more complex than any peer-to-peer system based on a central server, as was the case with Napster, it does offer something that Napster, or any other company or individual with a similar system, could do.
When Napster was temporarely closed, all users were not able to use the service anymore. With a distributed network, most, if not all the servers that take part in the distributed network must be closed to make the service unavailable. So, distributed networks are not only good for data distribution, but more importantly they are perfect for systems were availability of the service is important. This makes implementations of distributed networks span over much more than MP3 copying, as it was the case with Napster.
While distributed networks can be succesfully used for illegal data distribution, they have technical advantages so important that they should be considered in many other areas, making them a worthwhile area of research and development.
(1) The "Napster" that is described here is the original one, not the one based on a "subscription service". So, the "Napster" described here is, as offered by the company that designed it, "dead", hence the past term. Though, the protocol is still available in open-source project, but from what I know, the servers are never hosted by companies, by fear of a retribution similar to Napster's case.
(2) One of the indirect goals of the ANet project is to offer a distributed directory-based file information repository, though its distribution rules are quite different of those of Napster and more similar to those of Usenet. This would allow the information of a file to be extended to its full path, which conveys more "human-readable" information.
(3) I believe that it couldn't be done alone from the MPEG 1 layer 3 format, since the problem was much more caused by how easy it was to distribute the information from one computer to many computer (which was done by Napster) than about distributing the data from one computer to another (which was helped by the MP3 format).
Last update for this document: November 12, 2001, at 0:53:50 PST