Today in the group, we're arguing about cross-chain again, to the point where I almost want to hold my wine glass and start a governance meeting... Honestly, when it comes to a cross-chain, you trust more than just the "bridge": the source chain shouldn't roll back first, and the target chain shouldn't act up either. The relay/verification of message passing in the middle (who's watching, how they see it, who signed whose name) is even more critical. The advantage of the IBC approach is that it writes rules directly on the chain, reducing the feeling of "a few people helping you carry it over," but you still have to trust that the light client code doesn't crash, relayers don't mess around, and the other chain's consensus can truly backstop everything. Looking at extreme cases like funding rates, when everyone asks whether to reverse or keep pumping the bubble, I think it's quite similar to cross-chain: you think you're betting on the direction, but really you're betting on which part of the system will break first. Anyway, I now do small-scale cross-chain tests first; if I can avoid crossing, I avoid crossing. After drinking, we'll talk.