To construct a scientific theory from the data and to be able to recognize that it is a reasonable theory is possible only if there are some very sharp restrictive principles that lead you to go in one direction and not in another direction. Otherwise, you wouldn't have science at all, merely randomly chosen hypotheses.