Proving the Impossible Is Impossible, with Applications to Robot Worlds