@vitalik.eth

It's looking really interesting! My fear going in was that switching to a protocol like this *really* reduces security by almost half, the last thing ethereum should do when it's one of the very few chains aiming for bitcoin-like resilience and decentralization. But in reality: * It increases the finality threshold from 67% to 80%, which is actually something I have been advocating for a while (because I think "being in limbo and entering an inactivity leak" is a much better failure mode than "finalizing the wrong thing") * It reduces the "percent slashed" threshold from 33% to 20%, but imo that's totally fine, 33% (today: ~$30B slashed in a 51% attack) is massive overkill anyway * If you combine it with an available chain in the natural way, it does reduce the threshold needed to dominate the fork choice (and thus censor) from 50% to 40%, but... If you make a non-standard choice for the thresholds: 50% to notarize, 83% to finalize, 33% to advance rounds without progress, then: * The fork choice domination (ie. censorship) threshold stays at 50% * Percent slashed in a successful finality reversion drops from 33% to 17% (but that's fine, 33% is way overkill) * Percent that need to be offline for finality to stop drops from 33% to 17% (but that's fine, as the chain keeps going and inactivity leak recovers finality within hours-days anyway) Tentatively looking very good! https://x.com/_patrickogrady/status/1993340409138102482

22 replies

53 recasts

364 reactions