pull down to refresh

My understanding was that fedimint does have some ability to handle byzantine faults.
Fedimint does say in their docs that "a BFT consensus algorithm is used to agree on a set of consensus items."
On another page they say
The consensus protocols we are discussing, asynchronous ones, can only handle about 1/3 faulty nodes, so this will also be our assumption when building our protocol on top if not stated otherwise.
But the reason dpc is working on this (aside from it being really interesting) is that they perhaps would like it to be more robust.