The dominant tradition in Western philosophy sees rationality as dictating. Thus rationality may require that we believe the best explanation and simple conceptual truths and that we infer in accordance with evident rules of inference. I argue that, given what we know about the growth of knowledge, this authoritarian concept of rationality leads to absurdities and should be abandoned. I then outline a libertarian concept of rationality, derived from Popper, which eschews the dictates and which sees a rational agent as one who questions, criticises, conjectures and experiments. I argue that, while the libertarian approach escapes the absurdities of the authoritarian, it requires two significant developments and an important clarification to be made fully consistent with itself.