The main feature of PoW over PoS is being completely objective and uncensorable. There will always be networks using it because of this even when most are using and better off with PoS.
PoS is inherently self referential. To know the state at time N you must agree to a subjective view of validators within at least some subjectivity period of N. PoW has no such requirement and a chain of headers needs no additional information to verify.
> If a malicious group gets 51% of PoS ETH you can fork the chain to a new one that removes the malicious actor stake.
This kind of censorship is exactly what I mean. You can't fork away a PoW miners hasherate without blowing up everyone's hashrate. The fact that this may be wanted in some form is one reason many protocols are better off with PoS, but it's still a censorship vector that isn't available in PoW.
You're arguing reasons that PoS censorship points are ideal not that they don't exist.
Considering that the weak subjectivity period is so long (months), having to sync "out of band" is not really a big deal.
And unlike PoW, where you need to wait some confirmations to make up for network latency, with PoS you have actual finality assuming no 33% attack, which can be detected and punished.
> This kind of censorship is exactly what I mean. You can't fork away a PoW miners hasherate without blowing up everyone's hashrate. The fact that this may be wanted in some form is one reason many protocols are better off with PoS, but it's still a censorship vector that isn't available in PoW.
> You're arguing reasons that PoS censorship points are ideal not that they don't exist.
PoS or PoW aren't any more or less hard forkable. Any network can be hard forked (ie people agreeing to run new software with new rules) if people agree so. This is basically the social consensus, which is always above any other kind of consensus.
The "censorship" that you mention here is not a matter of a network currently being PoW or PoS. You could hard fork Bitcoin to PoS and take away some people's money (therefore they couldn't validate) if peopled agreed to do so.
There are two types of censorship, which are being confused here
A) Miners/validators censoring certain addresses (not allowing their TXs to go through) B) Not allowing certain people to become miners/validators
A is the real problem. However B can lead to A so it's also important.
In a PoW network where a malicious actor is doing A and has a lot of hardware and money to buy new hardware, you can't fix your network even with hardforks (outside of moving to PoS)
In a PoS if people agree to do so, they can hard fork and keep everything the same but take away their stake.
Now, your point is that people can do B via hardforks that take away people's stake. True but that is kind of irrelevant. With hardforks you can do whatever you want, steal anybodies money, switch the network to whatever validation system, etc. But why would people care about your hardfork? It's all about what hardfork people agree is the one that matters. Which is not something technical, is something social.
Needing to do it at all, at the beginning, means there is no mathematically pure way to validate the consensus state and nobody can affect the concensus unless it's rooted in this initial distribution.
> In a PoW network where a malicious actor is doing A and has a lot of hardware and money to buy new hardware, you can't fix your network even with hardforks (outside of moving to PoS)
Yes, this is exactly the censorship that is available in PoS and not PoW. I help minority PoW protocols integrate PoS features for this reason, and it's also the reason there will always be PoW chains. Because some social networks demand uncensorable participation that can't be selectively hard forked away.
You also need to validate the software binary or source code out of bound at the beginning, so having to additionally validate out of bound the state makes little difference.
> Yes, this is exactly the censorship that is available in PoS and not PoW. I help minority PoW protocols integrate PoS features for this reason, and it's also the reason there will always be PoW chains. Because some social networks demand uncensorable participation that can't be selectively hard forked away.
As I said, a PoW chain isn't any more resistant to censorship via hardforks. You can literally hardfork a PoW chain into a PoS chain. It is impossible to make anything secure or whatever if you consider hardforks as something an attacker can execute at will and force people to move to the new fork.
Its okay if we control all the supply because we have it in the premine to our friends and they have the best interest in keeping Ethereum safe.
In contrast for POW you have to be able to source specialized hardware and an electricity supply cheap enough to make the hardware profitable, both of which are much more noticable to the outside world.
The second, probably more important way, is that validators in PoS can decide to ignore or slash a particular validators stake. This can't be done in PoW because the hashes are fungible and stake is not.