“The scientist thus learns truth experimentally or mathematically; the strategist reasons at least partly by analogy with the past – first establishing which events are comparable and which prior conclusions remain relevant.”