Sorry, I mixed in a bit of what I have been seeing in other discussions here. There seem to be many people claiming that relaxing the OP_RETURN limits would make it more expensive to run nodes. I am trying to push back on that argument, as I don’t consider it tenable.

AFAIA, some people at Ocean feel that the blocksize generally should be reduced to 300 kB because it is already too expensive to IBD.

Yeah, I think that would be a fair characterization of an argument in this debate. There are a bunch of 2nd-layer protocols, sidechains, and rollups, etc. in development that aim to use the Bitcoin blockchain for anchoring, proofs, or data storage in some other way. Dropping the OP_RETURN limit would ensure that they can at least use OP_RETURNs if they must write data to the blockchain instead of writing in more harmful ways like fake pubkeys or fake pubkey hashes. OP_RETURNs would need to be part of a complete copy of the blockchain, but would not live in the UTXO set which needs to be kept highly available. Fake pubkeys/pubkey hashes live in both the UTXO set and the blockchain.

A pruned node keeps the last two days of blockchain and the entire UTXO set. A node with a complete copy of the blockchain would retain the data in the blockchain and will also keep the UTXO set of course. The UTXO set is also stored on disk and usually only recent outputs are in the cache in the memory. A bloated UTXO set especially affects IBD and generally slows down loading UTXOs from disk.