According to robust virtue epistemology (RVE), knowledge is type-identical with a particular species of cognitive achievement. The identification itself is subject to some criticism on the (alleged) grounds that it fails to account for the anti-luck features of knowledge. Although critics have largely focused on environmental luck, the fundamental philosophical problem facing RVE is that it is not clear why it should be a distinctive feature of cognitive abilities that they ordinarily produce beliefs in a way that is safe. We propose a novel way to resolve this problem. Key to our proposal will be an appreciation of different representational states beholden to truth. We suggest these different representational states are distinguished by how, in the proper governance of these states, the twin goods of attaining truth and avoiding error are weighted. Moreover, we explain how varieties of representational states line up with varieties of cognitive achievement such that knowledge, cum cognitive achievement, must be (ordinarily) safe because of the kind of attempt at success that belief is-namely, an attempt that places the premium it does on avoiding error.