Counterfactual Mugging and Logical Uncertainty — LessWrong