3 ms·

What do you think is not-state-of-the-art about LibraBFT?

• sgeisler 7y ago
  My main disappointment was that it's synchronous. Synchronous BFT algorithms seem like an old and already solved problem, with many already existing implementations. So I don't see exactly why you'd develop your own, especially since it's quite hard to get right. Maybe I missed some revolutionary performance increases back when I read the paper. Imo async BFT algos are much more interesting since they can actually fit the attacker model of an open network like the internet. Nobody can give you any guarantees (in an absolute sense) about message delivery times and hops can arbitrarily delay your messages.