I've just seen that Donald Smith's ‘Vague Singulars, Semantic Indecision, and the Metaphysics of Persons’ is out in PPR.
Smith argues that if ‘I’ is indeed vague, and the view of vagueness as semantic indecision correct after all, then ‘I’ cannot refer to a composite material object. But his considerations would, if sound, also establish that ‘Tibbles,’ ‘Everest,’ or ‘Toronto,’ do not refer to composite material objects either—nor hence, presumably, to cats, mountains, or cities. And both considerations can be resisted, anyway.
As to the first, it suffices to observe that if ‘I’ (or ‘Tibbles’) is vague, and the view of vagueness as semantic indecision is correct, then, when I assert a sentence containing it, I do no need to take myself to having successfully referred to any particular thing—if that is understood as definitely referring to something. Rather, I aim my statement to turn out true on any admissible way of making the semantic decisions that are not (and should not, and maybe could not, be) made.
As to the second, one just has to notice that the “many” solution to the problem of the many is certainly not the only solution that defenders of the view of vagueness as semantic indecision can adopt—and have indeed adopted. One rival solution by disqualification is the so-called “supervaluationist” solution, mentioned by Lewis and more recently defended by McGee & McLaughlin, Varzi, and Weatherson. According to this alternative solution, each sharpening of ‘is a cat’ or ‘is a person’ selects just one of the many candidates—different ones in the different sharpenings, thus respecting the arbitrariness felt in denying that they all had an equal claim. ‘Tibbles is a cat’ serves as a penumbral connection, guarantying that it is rendered inadmissible any sharpening that selects a different candidate as the referent of ‘Tibbles’ from the one that is selected as belonging to the extension of ‘is a cat’—inasmuch as ‘If it is not red, then it is orange’ serves to exclude sharpenings in which borderline rose Fifí is assigned both to the extension of ‘is red’ and to that of ‘is orange.’ Thus the many candidates are indeed equally eligible as referents of ‘Tibbles,’ but it definitely the case that one and just one of them is a cat after all. Mutatis mutandis, once again, for persons.
My response will appear shortly in Sorites.