A detailed comparison is presented of five well-known crossover operators on a range of continuous optimization tasks to test longheld beliefs concerning disruption, with respect to real-coded GAs. It is suggested that, contrary to traditional assumptions, disruption is useful.