[drdeca]

Would you? Are you saying that an oracle which gets some statistical information from such zk-proofs could be used to solve NP-complete problems in (randomized?) polynomial time?