This is where small, iterative feedback loops come helpful! It is better to tell a train beforehand that it might go on the wrong-track, than to tell it when it has already gone on the wrong one.