> If someone can create a protocol clever enough to efficiently use
> the distributed computing resources of the
> internet, it will always exceed any centralized capacity
It's not a protocol problem: TCP/IP would do well enough. The problem is 1) bandwidth 2) latency
This can be addressed: 1) by much fatter nodes or 2) different network technologies (photonically switched networks instead of xDSL, or >>1 GBps local cellular wireless).
Right now the most interesting project a la distributed.net would be looking for a way to mutate machine code robustly (using a virtual machine to protect the innocents). Obviously, one needs to mutate the mutation function itself and feedback it into the pool.
Unfortunately, breaking crypto or looking for ayleehns is way more appealing to the general public.
> the government can attain. Of course, the government can do
> the same thing. Either way this allows multiple parties to
> compete in creating SI's. It could get very interesting.