I mentioned this, idk about a hundred pages back, in the form of talking about what mining is actually for.

The only reason we need a P2P network and mining at all is to resolve double spending. Most of the work of validating transactions is stateless, except that two proofs are needed which require a mined blockchain to produce:

• The inputs to the transaction exist
• No transaction exists which spends the same inputs

There are really only two ways that mining can fail:

• A miner can perform a double spend
• A miner can execute a denial of service against valid transactions

The point of proof of work is to raise the cost of both of those attacks. Bitcoin never has made either of those attacks impossible in a mathmatical sense (and doing so is probably impossible) - all Bitcoin ever did was put a defined cost on those two attacks.

If people would stop arguing about undefined terms like "decentralization", then maybe instead we could talk about attacks in terms of ways an attacker might reduce the proof of work cost for performing double spending or DoS attacks.

It should be possible to stop worrying about miners and all the ways they might behave sub-optimally, as long as they don't have any way to avoid paying the specified PoW cost for any attacks they might perform.

View in these terms, having any protocol-mandated block size limit at all is a built-in denial of service attack and so should be removed as soon as possible.