These days, I've been looking at the old problem of cross-chain bridges again: is multi-signature really "safe"?

I'm now leaning more towards treating it as "distributing single point of failure," not eliminating risk.
Who the signers are, whether they are from the same company or the same batch of machines, how to use emergency pause permissions—basically, all of these can determine whether you're waiting for technical confirmation on the bridge or just waiting for people's trust to shift.
Oracles are the same—on-chain, it looks like price feeds/status feeds, but in reality, it's still a trust chain.
If the data source is biased or delayed significantly, the bridge starts to go haywire.

So I increasingly understand the significance of "waiting for confirmation"—it's not procrastination, but giving yourself a calm window: waiting for finality on the chain, waiting for risk control alerts, waiting to see if there's any strange large inflow or outflow.
Especially recently, hardware wallets are out of stock, and a bunch of phishing links are flying everywhere.
People's security awareness has improved, but a quick hand can still lead to mistakes.

That's all for now.