Last week, in two posts on the "tracking theory of knowledge," we reached the following point: 1) The tracking theory seems to capture many of our intuitions about what it means to know something, 2) It resolves the Gettier problems, 3) But it creates a glitch of its own -- it isn't epistemically closed. It generates circumstances in which I know p, and I know that p implies q, but I can't be said to know that q. Nozick, though, thought of the non-closure as a good thing: a feature, not a bug, because it allowed him to address global skepticism hypotheses. Suppose we concede that I don't know that I am not a brain in a vat, or that I don't know that I am not the victim of a joke by a malicious demon. Should this bother us? Specifically, should we infer from ignorance on this point that we don't know anything at all? Well, we would have to if we thought knowledge was epistemically closed. I know (we want to say) that I am now sitting in a gre...