Recently, I saw someone complain again about validators/miners eating too much, and MEV making the ordering look like a lottery... Actually, once you do cross-chain (whether it's IBC or message passing/bridges), the trust surface automatically expands. To "cross over" once, there are quite a few things you need to trust: the source chain itself shouldn't go down or be excessively reorganized; the set of light clients/relays responsible for delivering messages shouldn't drop the ball; the target chain must verify your proof according to the rules; if there are multi-signatures/oracles/executors in the middle, that adds another layer of "people" and "operations" variables. More practically, who holds the ordering power, who can cut in line—many times, that determines whether your bridge is normal or gets a slippage slip.

My current approach is somewhat conservative: before large cross-chain transfers, I first check the mempool and blocks for obvious signs of front-running, and I prefer to do it in batches and slower. Next time, I might just write a checklist of "trustworthy components." Which part of cross-chain are you most afraid will go wrong?