> the prize for a 51% attacker would quickly become worthless since the value of Bitcoin would fall when people realized that mining was controlled by one entity.

I’m more optimistic, personally, because of the possible market responses to such an attack. Censored parties can increase fees to incentivize non-censoring miners to join the network, for example.

You’re right that honest miners are disincentivized from establishing a standing majority to avoid this risk under the status quo. We have seen this empirically as miners move hash between pools when any gets too large.

I don’t know if the existential risk is sufficient to dissuade an attacker from attempting a majority escrow sweep. It is a different category of prize from previous 51% attacks, which require history rewriting.

An otherwise honest pool (or cohort of cooperating pools) can sweep escrow without rewriting history. The game theory of these interactions is not clear cut, IMO, and what we’re gambling with here is the finest money ever devised.