Abstract

We argue that small miracles and transition periods, notions that play part in Lewis's standard resolution of vagueness of counterfactuals, lead his theory to errors, either by giving conditionals wrong truth values, or by going against other Lewis's views. In fact, we do not believe that Lewis needs these notions. We suggest a way how his theory could be reformulated without them, in a way that does not disturb other parts of his system. More precisely, we argue that the asymmetry of counterfactual dependence can be reduced to the asymmetry of overdetermination even if the relation of similarity is determined without notions of small miracles and transition periods.