The block size being the limit is a really myopic and naive perspective that doesn’t factor in any sort of systems/adversarial thinking. Taproot wizards proved that this was a bug.

However, I agree with you that UTXO bloat is a problem that needs to be solved economically. If we all run knots with full filters then this will become a problem again. A witness limit of 400kb with an OP_RETURN of 4kb I think is a better solution. This solves the real problem for nodes (witness unlimited) and allows the market to show what it might do with a slightly useful OP_RETURN (it would especially be useful for Nostr, which would be complementary to Bitcoin’s monetary use). I will expand on this in a longer form soon if you’re interested…