Temporal loops in which two agents are involved in a circle of reactive attitudes create a puzzle concerning who harms whom. I argue that this puzzle, which has been developed by Stephen Kearns in a recent paper in this journal, should be solved by accepting that the situation involves normative indeterminacy. A supervaluationist treatment of this indeterminacy allows us to maintain that the normative supervenes on the non-normative and that the involved agents are in normatively symmetric situations. It further allows (...) to maintain that it is determinately the case that one agent wrongs the other just in case the latter does not wrong the former. I conclude by tentatively arguing that the indeterminacy involved should not be understood as semantic and that the puzzle provides us with a novel case of indeterminacy in reality. (shrink)