Communications networks are the same no matter who is plugged into the channel... The way that gossip protocols accelerate and improve distributed databases is not by themselves, there has to be a low-entropy phase in the data distribution process. You would not trust a financial database that was 99.9% correct. It has to be 100%! But you can have it 100% up to a very short time ago by using secure, trustless communication protocols. Eric Freeman's paper talks about ways of using chains and meshes of relays to prevent corruption of data in transit. Yeah, it's doing my head in a bit at the moment soaking it all up....