Let me illustrate my objection to the basis of this proposal:
You spend $1000 on a safe. Worried that a burglar might damage it by trying to get at the valuables inside, you take out the $1,000,000 of gold it contains and leave it in the street.
Breaking down why this is a poor strategy:
- The safe is worth less than the gold, so you need to protect the gold not the safe.
- The safe has no utility if you don’t use it to protect the gold.
Block proposer/builder secures the empty safe (blockchain structure) at the expense of the gold (blockchain content).
It formally omits transaction inclusion and ordering from the consensus on the basis that it threatens the structural security of the blockchain, but the reason it increases the risk of a consensus attack is precisely because it is worth protecting from those trying to do so.
The proposal (correctly) admits that MEV extraction is centralizing, hence trying to mitigate it in the structural layer.
But in doing so it actually facilitates the centralization of content, the endgame of which is its monopolization by a handful of wealthy, well resourced actors with their own agenda, ie: the antithesis of blockchain technology.
Our security assumptions will have fallen from requiring 51% of the hashpower (or thereabouts) to co-opt the chain, to simply being the best at extracting MEV, having the most money or even just having the greatest motive (or being close enough to all three).
No-one will ever have 51% of the stake/hash power (hopefully). Someone is always best at extracting MEV, and someone is always the richest, therefore it is guaranteed that at least one actor is always in a position to monopolize the content of the network according to their agenda.
@fradamt spoke similarly in this excellent post. Your response that such as attack is uneconomical does not account for the fact that censoring transactions can have large and unquantifiable private value to the censor and that once established that value may be self-reinforcing. Allowing the proposer to add a small number of transactions to mitigate this will simply incentivize an informal secondary auction market which they can also dominate.
Rather than excluding transaction ordering and inclusion ever further from consensus, it seems to me that we must look at consensus mechanisms that include it.