>it really depends on the situation and the speakers' 'subjective attitude'
This can be used to account for any utterance.
>Let me filter it down. Say ,initially, we got “ he can be hiding anywhere'
vs “ he must be hiding somewhere' and we don't question so far ( ok ?) the 'felicity condition' for the positive use of 'epistemic can' in the former.
We don't questions that if we take that to refer to dynamic modality (he knows how to hide anywhere). If we also know that he's not left the scene, we can conclude that he must be hiding somewhere. So far, so good, but it only works one way.
>So I was trying to violate ( theoretically ) the “possibility or ability"
of 'can' and pop slightly the “logical assumption" of 'must 'into the former on the assumption that 'he must be hiding somewhere so he can( not may ) be hiding anywhere then' or he 'can be hiding anywhere so he must be hiding somewhere then'.
No, if we're sure that he's hiding somewhere, we might say 'he may be hiding anywhere' if we fear so. However, to say 'he can be hiding anywhere' requires that we know he has the know-how of being able to hide anywhere. That's what I meant when I said 'one-way'.
>I know it's a bit far-fetched but we're not here to exclusively deal
(excuse the split infinitive ).
No. To leniently excuse that would go against my beliefs.
Quote , “ .. modal auxiliaries have been grammaticized into expressing many different types of meaning , deontic, dynamic and epistemic, and are in themselves ambiguous as to their exact semantic meaning. The polysemy of the modals have been noted by Leech and Coats ( 1980) and Coates ( 1983),for example, who claim that a modal verb may be ambiguous especially between an epistemic and non-epistemic meaning, and also indeterminate between two meanings, in which case even the context may fail to excude on of the possible meanings", close quote .
Does your source give examples? Not specifically of can, but it would just be interesting to see such na-dwoje-babka-wróżyła sort of examples.