The utxoset size is *permanent* it can’t be pruned like other block data unless you consolidate the spend into a smaller set of utxos basically think of them as a coin purse where if you put two coins in, the only way to shrink the bag is to spend to coins with no change. JPEGs in witness data means that they will likely be unspendable, meaning that there is a permanent storage increase requirement on all nodes. But they are no provably unspendable so you can’t discard them from the coin purse. This is really bad. OP_RETURNs are *provably* unspendable, meaning they can be ignored from the utxoset perspective (never goes in the coin purse) By trying to stop both witness jpegs and large OP_RETURN pushes, it will push people to do even worse things like large multisigs that stores data in the signatures. This is how the whitepaper is permanently stored in the utxoset. This is even worse for utxo bloat. At this point the censor proponents would say well thats not economically viable… but none of these methods really is. Some are cheaper than others, sure, but overall it’s still the most expensive data storage out there. People have to burn the hardest money on the planet if they want to play stupid games. The point is people are going to store data anyway, the *least bad* is OP_RETURN, because it minimizes the *permanent* storage burden on pruned nodes.