Thirty years ago, we used to ask: Can a computer simulate all processes of logic? The answer was yes, but the question was surely wrong. We should have asked: Can logic simulate all sequences of cause and effect? And the answer would have been no.