Certain” represents an assessment of the evidence for a conclusion; it is usually contrasted with two other broad types of assessment: “possible” and “probable.” . . .
Idea X is “certain” if, in a given context of knowledge, the evidence for X is conclusive. In such a context, all the evidence supports X and there is no evidence to support any alternative . . . .
You cannot challenge a claim to certainty by means of an arbitrary declaration of a counter-possibility, . . . you cannot manufacture possibilities without evidence . . . .
All the main attacks on certainty depend on evading its contextual character . . . .
The alternative is not to feign omniscience, erecting every discovery into an out-of-context absolute, or to embrace skepticism and claim that knowledge is impossible. Both these policies accept omniscience as the standard: the dogmatists pretend to have it, the skeptics bemoan their lack of it. The rational policy is to discard the very notion of omniscience. Knowledge is contextual—it is knowledge, it is valid, contextually.
“Don’t be so sure—nobody can be certain of anything.” Bertrand Russell’s gibberish to the contrary notwithstanding, that pronouncement includes itself; therefore, one cannot be sure that one cannot be sure of anything. The pronouncement means that no knowledge of any kind is possible to man, i.e., that man is not conscious. Furthermore, if one tried to accept that catch phrase, one would find that its second part contradicts its first: if nobody can be certain of anything, then everybody can be certain of everything he pleases—since it cannot be refuted, and he can claim he is not certain he is certain (which is the purpose of that notion).