This talk will focus on developing a theory of risk for the normative assessment of an agent's credence functions, within the framework of epistemic utility theory. In particular, I propose a general theory of epistemic risk in terms of relative sensitivity to different types of graded error. While this account is analogous in important respects to contemporary approaches to risk in ordinary expected utility theory, it has a uniquely epistemic interpretation, which has its roots in Peirce's ``economy of research''. I express this framework in information-theoretic terms and show that epistemic risk, so understood, is a scaled reflection of information entropy. As a result, every unit increase in risk comes with a corresponding unit decrease in information entropy and epistemic risk may be expressed in terms of entropic change. I explain the significance of this for the choice of scoring rule, the selection of priors, and the Laplacian principle of indifference.