Since the user can't prove that they are using this technique, or petertodd's timelock encryption for that matter, an attacker has little incentive to stop physically attacking until they have a spendable UTXO.

I believe you can get the same effect with on-chain timelocks, or delete-the-bits plus a rangeproof and a zero-knowledge proof that the rangeproof corresponds to some secret that can be used to derive the expected public key. I think Jeremy Rubin had an idea for such a proof.

Also, adam3us has described a similar thought here:

- Bryan