We are problem solving active guess-workers, capable of an indefinite number of anticipations, which we have to criticize through the analysis of their deductive consequences. "Inductive inferences' are an illusion; either they are pure guesses, or enthymematic deductive reasonings, with an implicit "fair sample hypothesis." Our aim is not justification or even consensus, but truth. And it is because our theories are true or truthlike that they have some successes. This can be generalized to the realm of moral and political concerns: our aim is not consensus, but the invention or safeguarding of just institutions. Popper's theory of democracy appears as a deliberative theory of representative democracy. Its rationality is inscribed in its method, the liberal theory of public critical discussion. Consensus, after dissent, is only a fallible sign of our possible grasping of some part of what should be justice in an Open Society.