How many small wrongs for the results to get out of control?