Christopher Hookway correctly suggests that I am extremely uneasy about the oft-stated claim that epistemic judgments are normative. The issue is potentially important because a great many self-proclaimed internalists argue against externalism by claiming that the externalist cannot account for certain alleged normative aspects of our epistemic judgments. Thus, for example, some philosophers would argue against the reliabilist's construal of epistemic justification by appealing to evil thought experiments. We can compare two believers, each of whom has identical experience of the world, but where the subjective experience has radically different causes. The one believer is the victim of a powerful deceiver who amuses himself by inducing radically false beliefs about a non-existent material world. The other is a believer lucky enough to inhabit precisely the world of objects our common sense takes to be the causes of sensations. Can we really maintain that the demon-induced beliefs are any less justified than the world beliefs? To characterize a belief as unjustified is to criticize the belief, is to blame the subject for having the belief-justification is a normative notion. But the demon victim is blameless. The demon victim believes just what anyone else would and should believe given the available evidence. Reliabilists and tracking theorists, however, seem committed to the view that the demon victim's beliefs (produced by a massively unreliable process) are unjustified, while perceptual beliefs in the world are justified (produced, we may suppose, by generally reliable processes, processes that track effectively the relevant facts). The argument has certainly been influential. Indeed, the most notable reliabilist was clearly worried by the problem in his seminal paper What is Justified Belief? Indeed, he was so worried that in his book Epistemology and Cognition he took a disastrous detour into normal worlds reliabilism in