It is widely considered a truism that the only evidence that can provide justification for one's belief that p is evidence in one's possession. At the same time, a good many epistemologists accept another claim seemingly in tension with this "truism," to the effect that evidence not in one's possession can defeat or undermine the justification for one's belief that p. Anyone who accepts both of these claims accepts what I will call the asymmetry thesis: while evidence in one's possession (...) can either enhance or detract from justification, evidence not in one's possession can only detract from it. The asymmetry thesis is not uncontroversial; but any epistemologist who endorses the doctrine of normative defeat will be under tremendous pressure to accept it. In this paper I try to motivate the asymmetry thesis in two steps: first, by appeal to a feature that assessments of justification share with evaluative assessments generally, according to which we can distinguish generic expectations in play from the explicit criteria for satisfying the relevant evaluative standard; and second, by arguing that when it comes to epistemic assessments, the generic expectations themselves derive from our roles as epistemic agents in communities in which we depend on one another for knowledge. (shrink)