Why? PoW can only afford to be attacked twice. It even has a name: “spawn camping.” In proof of stake, this is pretty easy to defend against repeatedly.
If you are accepting that the network can hard fork to change the hash algorithm, then why not accept that they can do a soft fork to blacklist UTXOs or any one of who-knows-how many possible defenses? Changing the hashing algorithm is a nuclear option and probably not necessary in the astronomically unlikely event that someone managed to get 51% of hash rate for any amount of time.
edit: Blacklisting UTXOs is entirely possible as I mentioned, and as I said, it should be achievable by soft fork. This would be the proof of work equivalent of slashing. I have a hard time believing Vitalik doesn’t know this. Is he being dishonest? I don’t know.
You can target the UXTO so that their value in crypto starts back at zero. But because they still have the mining hardware, they still control 51% of the hash power that secures the network. They can just mine from a different address and it won’t be clear who to target until they once again begin to produce majority blocks on the chain. This way they can continually attack the chain and render it useless.
When you burn a PoW miner’s crypto addresses, you aren’t burning their capital and stake, because it’s in the form of mining hardware, not tokens. When you burn a PoS staker’s crypto, you are burning their capital and stake which is in the form of tokens.
I don't understand how PoS changes this. Also, that blog post does not consider CPU-mined PoW which is unprofitable for miners and sustained attackers.
In PoW, if you control 51% of mining the users can defend by switching to another hash algorithm, meaning the attacker needs to re-buy a lot of new hardware. If they’ve anticipated this, and attack again with new hardware, there is not much further defence. In PoS, users can coordinate a soft fork to burn the attackers funds, each time they re-attack the chain, until they eventually run out of capital.
edit: Blacklisting UTXOs is entirely possible as I mentioned, and as I said, it should be achievable by soft fork. This would be the proof of work equivalent of slashing. I have a hard time believing Vitalik doesn’t know this. Is he being dishonest? I don’t know.