Against Modus Ponens for Might

Suppose that you and I are at the mint and see a new coin. We have no way of knowing whether the coin will ever be flipped. Still, we know something: We know that if the coin were flipped, it might land heads. We know that if the coin were flipped, it might land tails. Surprisingly, this simple case is flatly inconsistent with theories of counterfactuals from (1971, 1973) and (1968, 1980). In this short paper I explain why, and consider how the problem might be solved.