pull down to refresh

Op_return being 4x as expensive means there are zero incentives to use it.
Witness data can't be attached to outputs, whereas op_returns can only be attached to outputs. In order to (indirectly) attach witness data to an output, you need a second transaction which spends that output and then you attach the witness data of the input in the second transaction.
Therefore, usually people need two transactions in order to put arbitrary data into the witness data.
In some use cases, they can't use two transactions. They have only one transaction, and they are determined to attach data to the outputs. They therefore have just two choices: op_return or unspendable-utxos. The latter is terrible in the long run, as it bloats the UTXO set. If high fees are failing to discourage the use of unspendable utxos, then we have to be realistic about our options.
Unspendable outputs are actually a good example of a working relay policy, as massive data storage only started when the taproot bug was discovered, which lowered cost and slipped by policy.
reply
"unspendable outputs" are absolutely to be avoided.
All technical actors and advisors on both sides agree that unspendable outputs increase the UTXO set unnecessarily and are to be avoided.
Numerous propositions and positions on both sides of the discussion want to address unspendable outputs... saying they are "a good example of a working relay policy..."
Is exactly the opposite of the truth, sorry.
reply