A friend of mine from graduate school, Dylan Dodd, discusses a new paradox (and how it relates to a similar paradox called the "Surprise Exam" paradox) in one of his new papers. You can find it at this link. Here's an abstract of his paper:
We've all been at parties where there's one cookie left on what was once a plate full of cookies, a cookie no one will eat simply because everyone is following a rule of etiquette, according to which you're not supposed to eat the last cookie. Or at least we think everyone is following this rule, but maybe not. In this paper I present a new paradox, the Cookie Paradox, which is an argument that seems to prove that in any situation in which everyone is truly following the rule, no one eats any cookies at all, no matter how many there are to be eaten. The `Cookie Argument' resemblance the more familiar argument that surprise exams are impossible, but it's not exactly the same. I argue that ultimately the biggest di difference is that, while every argument against the possibility of surprise exams (I discuss three) contains a subtle mistake, the Cookie Argument is actually sound! In addition to presenting this new paradox, I also present a new solution to a particularly difficult version of the Surprise Exam Paradox, the Conditionalized Exam Paradox.