Agree. What I’m saying is that our current methodology doesn’t just try to say “ensure cost(attack) > in-chain-benefit(attack), and if we reach that point rest on our laurels and say we’re done”, it says "try to come up with an algorithm that maximizes D, such that cost(attack) > in-chain-benefit(attack) + D. Then we have D as an lower bound on the minimum out-of-chain-benefit(attack) required to make a profitable attack, and it should be clear that the higher D is, the more secure the algorithm. And I believe that PoS can achieve a much higher D.
PoW is somehow less anonymous for major validators due to the possession of physical assets to play the powerfull mining position
This is both a benefit and a cost. PoW validators being known also means that they can collude more easily. Also, as I mentioned, it makes them more vulnerable to governmental and other meatspace actors.
PoW is supported by all the powerfull cryptocurrency communities so - may be (?!) - there is no interest to readress the computing power to destroy the competitors in comparison with the supporting their very own currency.
This is also ambiguous. The fact that PoW is supported by many currencies actually means that it’s more likely to expect PoW miners to attack, as they might reason that if they kill one of two competitors then the other will rise in value, and so their total ability to get revenue would not decrease by much (if the miners have maximalist empirical beliefs, they will in fact say that killing one of two competitors will increase the value of the other by more than the value of the victim, and so their revenue will rise).
Arguably, the fact that a PoS blockchain has a guaranteed hegemony over its consensus-forming resource is a very good thing for its security for its reason - if the blockchain does die, there is no place for validators to run.