I believe that a thoroughgoing fallibilism, which is supposed to apply actually to the statements constitutive of fallibilism themselves, is self-destructive, and consequently absurd. I wish to propose a less radical, but I think perfectly self-consistent, kind of fallibilism, which I shall argue is not self-destructive. A first-order fallibilism — applying to the natural science, history, and hermeneutics — may issue in, and in my opinion ought to issue in, a second- order infallibilism — applying to epistemology and metaphysics. More concretely, fallibilism is in general an infallible way of getting at the truth about the real world; or perhaps rather, fallibilism is an absolutely infallible way of tending towards the truth.
The issue is not merely a technical one within philosophy, but seems to be of some general cultural importance. Fallibilism is in effect an attempt to set out foundations for knowledge, or, what amounts to the same thing, to provide a way of distinguishing between knowledge which is adequately grounded, and so amounts to ‘knowledge’ properly so-called, and what merely makes some pretensions to be ‘knowledge’. As a result of the alleged failure of fallibilism and of other attempts to provide foundations for knowledge, the conviction has got about that knowledge neither has nor needs any such foundations. If this principle is once seriously accepted and consistently applied, the urgent question of how much of what anyone claims to be knowledge actually is so, and how much is not, can only be settled at the level of dogma and prejudice. If it is not, this can only be due to saving inconsistency; since to provide criteria in virtue of which any claimant to the title of knowledge can be rationally vindicated as such is nothing else than to provide foundations for knowledge.