And, nobody would correct them? If a programs output is random, what use is it.

Isn't the original point, that natural language isn't sufficient for proofs? Aren't you agreeing now?