This paper responds to a recently popular objection to non-naturalist, robust moral realism. The objection is that moral realism is morally objectionable, because realists are committed to taking evidence about the distribution of non-natural properties to be relevant to their first-order moral commitments. I argue that such objections fail. The moral realist is indeed committed to conditionals such as “If there are no non-natural properties, then no action is wrong.” But the realist is not committed to using this conditional in (...) a modus-ponens inference upon coming to believe its antecedent. Placing the discussion in a wider epistemological discussion—here, that of “junk-knowledge”, and of how background knowledge determines the relevance of purported evidence—shows that this objection does not exert a price from the realist. (shrink)