The non-maximality-solution to counterfactual scepticism

The following semantics for counterfactuals is fairly standard: for a counterfactual
to be true, the closest antecedent worlds have to be consequent worlds. Closeness is
measured by overall similarity of worlds to an evaluation world. There is a range of
interrelated challenges to this account: counterfactual scepticism, ‘Hegel’-, ‘Sobel’-,
and ‘Heim’-sequences. So far there is no unified solution to these challenges. I discuss
a solution that preserves the standard semantics by writing the shifty parameter into
pragmatics. The solution has been suggested by Križ for Sobel- and Heim-sequences,
yet I argue that it can be generalized to counterfactual scepticism. Conditionals are
subject to a pattern which is familiar from descriptions. Everyday counterfactuals are
semantically homogeneous and pragmatically non-maximal. Homogeneity: a counterfactual
is neither true nor false if only some but not all closest antecedent worlds are
consequent worlds. Non-maximality: in many contexts, not all but only practically all
closest antecedent worlds have to be consequent worlds for the utterance of a counterfactual
to say something true if the difference does not matter for the purposes of
conversation.