[globalreset]

Except in in PoW it only means delayed transactions, because only one miner has to "sign off" the block - everyone else accept it passively. A single miner, even with 0.1% mining power is enough to keep the network censorship resistant.

In PoS majority of validators has to actively approve a block containing "illegal" transactions, leading to permanent censorship.

The exact interpretation of validation vs mining responsibility in face of law and passive vs active is fuzzy , but it leaves miners is better legal position.

[randomran01234]

Interesting, so this is a problem if 66% or more decide to reject a transaction. How many blocks are currently signed by solo miners in PoW? Almost all blocks are signed by pools which looks like the PoS committees in practice yo my naive eyes. A miner with 0.1% hash power would probably be a very long delay.

I do agree this is probably the most concerning thing that Tornado Cash sanctions have shown us. Still don’t think the energy cost of PoW is worth it, and would rather see PBS and crList solve this problem.

https://notes.ethereum.org/@fradamt/H1ZqdtrBF

[TakeBlaster16]

The difference is that PoW pools are made up of individual miners who are free to go elsewhere if their pool misbehaves. Pools do not have their own mining hardware, and so they have an economic incentive to be well-behaved. If they misbehave, everybody leaves and they vaporize their business overnight.

PoS is tyranny of the majority: You cannot take your business elsewhere.

[randomran01234]

It’s the same with staking. If you aren’t solo staking, you are just delegating to a staking pool. And you can withdraw that and deposit it elsewhere if they do not align with your values.

[TakeBlaster16]

Again, it is not possible for staking pools to differ on "values". If a validator does not vote with the majority, their funds will be slashed by the protocol.

[randomran01234]

Trying to understand how this differs in practice. Anyone delegating to a staking pool is forced into their values. Ethermine recently started blocking OFAC transactions, so does the majority of miners in that pool agree with this? If so, what is the defence against censorship in this scenario except for some to exit the pool and use another that aligns with their values?

[TakeBlaster16]

Ethereum is still using PoW as of today, so anything Ethermine did recently isn't really relevant to discussions about PoS. As long as Ethereum is using PoW, pools of any size can pick and choose what transactions to include in a block. Once Ethereum switches to PoS, that will no longer be possible (without further protocol changes)

[brian_cloutier]

PoS seems very similar to PoW, and if one is fine (as we seem to think), the other is fine too.

In POW there are only block producers, and producing a block requires actively choosing transactions and transaction ordering.

Ethereum POS includes an additional attestation step where validators confirm that they have seen blocks without exercising any control over the contents of those blocks.

But this is exactly what PoW miners do when they extend the chain: every block is a vote for all previous blocks in the chain.

[robcohen]

A better legal position, until that is it's made illegal to accept unapproved blocks. PoW is not immune to the exact same regulatory pressure. The US gov can just force a fork of the chain by requiring US miners to only accept approved blocks.

[0x64]

You are wrong about the majority of validators having to approve a certain transaction for them to be included. Even if 80 % of the network were censoring, those transactions would, on average, make it into every fifth block.

[randomran01234]

Can you point to any sources? I would like to read about how this will be accepted even if the majority is against inclusion.

[0x64]

Because not attesting is against the network's fork-choice rules. If that censoring majority chooses to avoid said block, because it contains transactions that should be censored, they will perform an illegal re-org around the block. Sure, it won't be "illegal" as a majority of the network follows said re-org.

It's an extremely nuanced topic, and it was extensively discussed in the last core-developer call on Thursday. Every single operator has been warned that going against the fork-choice rules could, and ultimately will, result in social mitigations against spec-deviating behavior.

This could be set of socially executed slashings of validators that don't respect the fork-choice rules, or an honest-minority executed hard-fork that effectively causes misbehaving validators to bleed until they start respecting the protocol's rules.

I could be wrong on multiple counts, especially regarding re-orgs, but this is how I've understood the issue.