In the Meno, Plato argues that the value of knowledge over true belief lies in the fact that, unlike true belief, knowledge is more stable. This idea is echoed in Williamson's discussion in Knowledge and Its Limits, in which he argues that the probability that I will have a given true belief tomorrow if it is knowledge is greater than the probability that I will have that belief tomorrow if it is merely a true belief that does not amount to knowledge. Williamson's position seems to assume something like the following reductivist thesis concerning belief stability: the warrant or justification of a belief completely and fully accounts for contribution of belief stability to knowledge. Some such assumption, in fact, seems to underlie the appeal of the traditional analysis of knowledge as warranted/justified true belief.
In my papers on the analysis of knowledge (see here and here), I reject this reductivist thesis. I propose that, pace reductivism, knowledge should be analyzed as justified true belief plus a context senstive temporal-cum-modal stability operator which takes scope over all three of the traditional clauses. In such an analysis, there is no presumption that the stability of one's belief is entirely accounted for by one's justification for it. (Obviously the stability of the truth of one's belief is not accounted for by one's justification!)
The view opens up the following possibility: how stable a belief must be in order for it to count as knowledge may be determined, in part, by one's practical interests. Thus, if the proposed theory is correct, it explains Jason Stanley's recent anti-intellectualist observations about how knowledge can depend on practical interests.