I used to focus on whether cross-chain bridges had low "transaction fees," but now I mainly ask: who is actually confirming the transactions? Multi-signature, simply put, is just a few people signing off; even with more signers, there's still a risk if the same set of operators or the same batch of keys have issues. Oracles are more like "who reports the information"; a single error makes it clear that it's not mysticism—it's about the costs and incentives involved. The so-called "waiting for confirmation" shouldn't be considered slow; many bridge failures happen because people rush to prioritize speed, mistaking certainty for decoration. Those few minutes on the chain are actually a gamble—whether the opponent won't roll back, reorganize, or get caught in a trap.

Recently, everyone has been complaining that validators make money and that MEV (Miner Extractable Value) sorting is unfair. I can understand that, but on the bridge side, it's more direct: if the order gets mixed up, you think you're waiting for confirmation, but you're actually waiting for others to push you to the back... Anyway, I now prefer to be a bit slower so I can see clearly who has the authority to shut down, upgrade, or change parameters before deciding whether to proceed.