Or to be more verbose, the explanation of any phenomenon should make as many lols as possible, eliminating, or "shaving off," those that make no lols in the observable predictions of the explanatory hypothesis or theory.