@wizkid057 I've carefully read your messages as wells as the other messages from your colleagues at ocean and I have failed to find any clear explanation of the harm we could expect to experience from removing this limit.

It's all well and true that NFT/shitcoin stuff is bad, but there is no reason to expect that this should increase that activity: Anything that can be done with op_return can be done just as well (for the data embedder) with 'fake addresses'. And fake addresses are both far worse for Bitcoin due to bloating the utxo set and are essentially impossible to block. Moreover, parties that want to bypass this limit at scale and particularly for abusive purposes have an easy avenue to do so by directly handing large miners the transactions, which is now a reliable method for getting transactions mined that violate policy.

I've also found no counter to the benefits of removing the behavior: that at least some of the fake address traffic has indicated it will switch, that any discrepancy between what nodes relay and what miners mine hurts block propagation which advantages large miners at the expense of smaller miners, that direct-to-miner relationships also favor large operations over small, that an incomplete mempool reduces your ability to look at it and tell what price may will get your transactions in the next block, and that a fruitless game of wack-a-mole trying to block transactions creates a dangerously muddled message about the ability of Bitcoin participants to blacklist transactions/addresses such as those on state set blacklists. And of course, simpler code with fewer options and thus combinations of options to test.