Physics, and science generally, rarely function according to the mechanist tradition of founding all scientific knowledge on "shaped matter in motion" of the physical parts of a system. Rather we employ a vast range of explanatory strategies a great many of which work in terms of "stripping detail" when detail is not relevant to the problem at hand. Most of these strategies involve some level of idealization, inaccuracy, or distortion, which raises the worry: When accounts in science involve distortion, how can they count as knowledge? This problem motivates reconstruing knowledge, and in particular its requirement of (unqualified) truth in its content component, in terms of the kinds of standards that require something less than perfect precision and accuracy, in analogy to the context and interest dependent standards that we apply for representational accuracy of things such as maps and pictures. The paper concludes with consideration of possible connections with pragmatism and with ways of thinking about "independent reality".