On Sunday 03 September 2006 15:19, Ed Tomlinson wrote: > On Sunday 03 September 2006 07:18, Dave Baker wrote: > > I still don't see where the correlation between well-behaved nodes and > > non-malicious ones came from. > > I do not think we can ever know that a node is non-malicious.
Yes, that was rather my point. Don't get me wrong - avoiding misbehaving nodes is probably a good thing to do. I just don't think it will offer us any protection from nodes whose malice extends beyond trying to leech off the network. > Even a good friend can turn cause you pain (consider divorce)... I think we > can assume that most nodes a not malicious. Given this if we look for nodes > that behave or respond abnormally and decide not to trust them we probably > will at the very least aviod hacked > freenets... > > I would say that new nodes are trusted - with just enough that other nodes > will allow them to connect. If the new node does not behave, it quickly > will lose trust and will not get integrated. On the other hand if it > behaves the converse will happen. That seems plausible, so a misbehaving node will have less trust than he started with, which at least forces him to go to the hassle of generating a new identity. New nodes can still connect, albeit slightly slower. That's fair. Seems to me like this could be considered for opennet, provided that this 'misbehaviour' heuristic doesn't cripple those with the best of intentions, but that just don't have very reliable Internet connections.
