Wrong hypotheses, rightly worked from, have produced more useful results than unguided observations.