In line with that ideal, a competent witness to a mathematical or experimental demonstration should be able directly to follow the steps in a proof, but in the case of testimony (whether by a witness or an experimental arrangement) proof is indirect and requires fallible practical judgment.