The threat of determinism suggests that every action, including the action of holding morally responsible, is both the best and worst of all possible alternatives. This seems to pose a problem for consequentialist approaches to determinism, and moral responsibility. The solution is to appeal to the conversational context of praising, blaming, judging right and wrong, holding responsible, and the like. Even if, strictly speaking, an agent couldn’t have done otherwise, conversational context may select certain counterpossible alternatives as the relevant ones with which to compare the action. The non-identity “problem,” popularized by Parfit, suggests that the existential dependence of people on our actions creates puzzles, problems even, for some common approaches to ethics. A scalar version of consequentialism, combined with a contextualist semantics for some moral terms, dissolves the apparent problem.