ment, that wasn't how it was deployed.[8-27] To large ISPs like AOL, the ends
are important only in the way customers in a mall are important -- they con-
sume the commerce but don't participate in its design.

More recently, however, a change in the Net's architecture has occurred.
As machines have become more powerful, and as they have become more
reliably and permanently connected to the Net, the possibility of using
peers to process and forward data on the Net has increased. Peer-to-peer
services are returning to the Internet as machines mature and are persis-
tently on the Net. The character of what can happen is changing, and the
potential -- if left free to develop -- is extraordinary.

Napster is the most famous peer-to-peer technology, even though it is not
exactly peer-to-peer. (There is a central server that keeps a database of who
has what; the music itself is kept on other people's machines.) But Napster
is the horse and buggy in this transportation system. It is only the beginning.

Consider the SETI project. SETI -- the Search for Extraterrestrial
Intelligence -- scans the radio waves for evidence of intelligent life some-
where else in the universe. It does this by recording the noise of the radio
spectrum that we receive on planet Earth. This noise is then analyzed by
computers, looking for telltale signs of something unexplained.[8-28]

Who cares about wandering _X-Files_ types?, you might ask. Is this really it?
But the point is the potential that SETI evinces. The computation involved
in the SETI project is immense, and they soon discovered that the cost of
renting computers to process these recorded radio waves was increasingly
prohibitive. So researchers at Berkeley had an idea: Facilitate the distribu-
tion of chunks of this recorded data to machines across the Net, then enable
these machines across the Net to do the computation required on this data.
A package of data would be delivered to the participating computer along
with a program to be run; that program would run on the data and send it
back to the mother ship.

When the SETI@home project first began, within ten days it had
350,000 participants in 203 countries. In four months, it broke a million
users. The service grew so fast that it had to stop processing data for a while.
The speed at which data was being collected had surpassed the processing
speed.

In mid-2000, the system could boast the equivalent of 280,000 years of
processing time devoted to the SETI mission.[8-29]

Just as Napster had latched on to unused disk space, SETI@home had
latched on to unused computer cycles living at the edge of the Net. Idle ma-
chines could be turned to large-scale cooperative projects.

[[135]]

p134 _ -chap- _ toc-1 _ p135w _ toc-2 _ +chap+ _ p136

v?