Criteria for a good scientific question