Seeing someone criticize IBC as a "bridge," I can't help but want to laugh... Honestly, cross-chain transactions can't be trustless; it's just that you trust different components. When a message goes from chain A to chain B, the core trust points are: chain A not rolling back (finality), the proof/consensus mechanism that packages the message, the code on chain B verifying this proof isn't faulty, and the upgrade compatibility of the light clients/verification modules on both sides. The real danger isn't usually "the bridge gets hacked," but rather a certain upgrade changing the verification logic, and you still think everything is normal.

Recently, with staking/shared security and that "compound yield" model, I see the controversy is similar: you think you're just stacking yields, but actually you're stacking trust chains. Any problem in one layer could spread. Anyway, when I look at cross-chain, the first question I ask is: who is actually backing this message, and if the backer messes up, who pays the price? That's all for now.