i think it'll be a problem to the prog to find a proof in those to cases: a. the shortest prove ( the one with the minimal logicl stages) includes a big unsolved theorm. b. includes defining a new sets of axioms. suggestion to solve the first problem: the prog will build the math without defining what are the pupse of it (what is the target theorm)