Matthew Toseland wrote: > I'm not sure that this would solve the problem though. A hostile node could > probably get most of the keyspace, and therefore most of the local requests, > without interfering with the target's specialisation, which is where most of > its remote requests come in...?
Hmm, good point. That also applies to any fixed limit like 20% though, doesn't it? This brings us back to the old problem that any request that's far from the previous hop's specialisation probably originated at the previous hop... Cheers, Michael
