Roberto Bayardo 🎩
@bayardo.eth
Apparently ENS went with a currently-capable-of-ZK-chain for Namechain over optimistic was because they seem to be under the impression that "finality takes 7 days" for optimistic rollups, an unfortunately common misconception. If you run a full node (and ENS certainly would do at least that) and you believe the fraud proof system works, then finality happens the moment the state root is posted to the L1 (5-10 min) and you confirm it matches your local root. This is no worse than a ZK rollup. The benefit of current ZK rollups over Optimistic is the shorter withdrawal window, and little else. The additional complexity they involve unfortunately brings with it a host of additional attack surface area. I believe all L2s will eventually be ZK-powered, but IMO the current generation of ZK chains are reyling mostly on security through obscurity.
2 replies
0 recast
8 reactions
RoboCopsGoneMad
@robocopsgonemad
i think about this a lot. whats the most visible way to see what is going on wrt to fraud proofs? how often do they happen? how effective are they, what are the consequences, etc.... who is out there contesting optimistic rollups?
2 replies
0 recast
1 reaction
clabby
@vex
In reality, no one attempts right now. It’s happened three times on OP mainnet to my knowledge, and no one has ever attempted to play out the dispute past just a couple moves. One of the big reasons for this is because the dispute system is rigged with stage 1; even if there was an issue, it’d still get overridden by the security council.
1 reply
0 recast
1 reaction
clabby
@vex
They’re effective enough for stage 1 and more simple (middle ground; still complex! you still have a single-cycle CPU with a mini Linux kernel implemented in Solidity, plus a big complex dispute system and off-chain challenge agent.) As another downside, as we approach stage 2, they’re _far_ more expensive (approaching the TVL of the rollup, with the current impl on the OP Stack) for the honest actors to defend in the unhappy path. I expect chain operators to trend towards using hybrid optimistic/ZK or just full ZK. The complexity tradeoff will soon be worth it, I think. The most compelling hybrid systems in my opinion is DAVE by Cartessi. Full ZK would unlock light clients, though.
0 reply
0 recast
1 reaction
