After four long days had passed, sure. But the rollback could have been initiated much earlier (even after one block). My point was less about the economic feasibility, and more about the social infeasibility: we witnessed a resounding an unambiguous decry from the public against the possibility of a bribe-encouraged rollback, even to save user funds. The only “rollback” we’ve seen (on a meaningful chain, that wasn’t due to some weird in-protocol bug) was the irregular state transition to recover TheDAO funds, which was (debatably) initiated by the community, not accomplished by bribing miners.
Interesting counter-counter-measure. This would, however, make the blacklisting more obvious, which makes it easier to coordinate a drivechain-like UASF. The issue with game theory is that for any n-th order move, there is an n+1-th order counter-move, ad infinitum (a valid argument for not using game theory and instead only relying on cryptographic properties and universal composability). If only mechanism design actually worked in the blockchain space…
Yes. I personally used 2 weeks when implementing my prototype code of optimistic rollups. I’ve even suggested the possibility of months, if not a year, to withdraw funds, with the majority of user funds being “withdrawn” through atomic swaps instantly, rather than having to wait for the withdrawal period.
Pretty much. We don’t want to do a rollback/irregular state transition to save funds, but we could do a UASF (especially with PoS, as burning attacker’s funds is made easy)!