Maybe eventually he’ll arrive at close group concensus
By the way close group consensus still leaves the choice of consensus algorithm open.
Corda Ledger also does.
The current one you use I think fails if 33% or more nodes are malicious inside the section. But as they get kicked around, hopefully the section will become less malicious, unless we have a google scale attack.
Correct, and also worth noting that David said this is the best theoretically possible.
How does he know for sure? Would love to see that theoretical result. I have been discussing this with the chief cryptographer of Ripple for a while.