This is a very nice protocol to tackle the problem of leakage, but it is not perfect either. It is practically the same as mine with ZK proofs or gmaxwell's based on Schnorr, because leakage can still happen via u.
The value u never leaves the user computer, so it would be impossible to a backdoored hardware wallet to communicate u to the malicious party.
But it is better than mine because it is easily computable. It is better than gmaxwell's because it does not require a protocol change.
Could you provide the link here to your method ?