I've been devoting quite a few cycles lately to p2p networks, including popping my head into the p2p-hackers mailing list. It's a frustrating field for me, because of my belief that real progress in p2p network design is going to come only from serious intellectual work. The usual methodology of cobbling together a little network, then crossing your fingers and hoping it will scale will lead to lots of toy networks that won't scale. Unfortunately, the impression I get the p2p-hackers mailing list is that most people are still there.
I played a bit more with Mojo Nation over the weekend. It's one of the few p2p networks good enough to criticize. Unfortunately, it left me with the impression of still being pretty far from prime time. It's still got scaling problems up the ying-yang, and I believe it's vulnerable to all kinds of fun attacks as well. Even more disappointingly, there seem to be all kinds of low-level implementation problems, including memory leaks, and still-poor handling of failures (the logs show lots of failed "pass this along" messages). Block availability is still spotty, so having file downloads fail because of unavailable blocks is still common. That said, it is possible to use the network to pirate music and stuff. I'm not giving up on it yet, I just wished it was a little farther along.
I think I've figured out how to combine the attack resistance of Advogato with Simic's Chord protocol. I've sent an email about the subject to the few people I know who would even know what that means, or why you'd want to do it. Feel free to prove me wrong!
I am such an elitist.