I hear often variations of the following complaint against Kripke's firm commitment to the distinction between denoting something and whatever surrounds the process of describing something:
"The indiscernibility of identicals is too strong a principle to be accepted in all times and circumstances. It assumes that identity is always already given - that identities precede any act of identification - and takes identity as an arch-structure capable to ensure respect for singularities (to allow Socrates not to fulfill any of the descriptions we attach to him), as identity is already there no matter the work we do with our descriptions. This arch-structure, however, is just too heavy to be swallowed."
I tend to both agree and disagree with the complaint. On the one hand, Kripke makes justice to the independence between things and their descriptions in a way that Frege or even Russell do not (even though Russell allow for existence to play some role in his semantics). On the other hand, I don't like the metaphysical assumption that identity precedes identification and, in particular, that individuals precede individualisation. I don't want to take individuals as ready-made, archaic items and it is clear that no commitment to realism entails that. But I also wouldn't like to have singularities hostage to their description. Is there a way to preserve the thrust of Kripke's effort without accepting an identity metaphysics (or an identity-based logic)?
Maybe we can start considering an example similar to Quine's in "On what there is". Pointing at an empty doorstep one utters:
(1) The bald fat man at the doorstep is waiting
(2) The bald fat man at the doorstep is fat
(3) The bald fat man at the doorstep is not fat
Quine's (Russellian) strategy is to say that (1) is false. Then, also (2) and (3) are both false as there is no bald fat man in the doorstep. If, of course, there were such a man in the doorstep, (3) could be taken to be true and (2) false if we accept Donnellan's very convincing idea that definite descriptions have a referential use. In fact, I guess there is much to this idea and maybe what we're after is to restore some of its original force (in a way that has been also tried by the so called two-dimensionalist theories).
The Meinongian strategy would be to associate (1) with an object. Such object would lack the open horizon of modal life (borrowing the beautiful phrase of Grace Paley in her "Conversation with my father") because it is dependent on a description in exactly the way Kripke tried to avoid. They would be dependent on the descriptions that introduced them, say (1). Tomas Ribeiro Cardoso, my favourite Meinongian, would then appeal to Donnellan to say: it is not that bad to be hostage to a description if we take descriptions to have a parallel referential life and that they can be used in a non-attributive way. But it becomes unclear what could be the referential use of an empty description - if there is no one at the doorstep. It would be a kind of a conditional referential use, I suppose. Meinongians, I understand, tend to exorcise any consideration of existence from their notion of object.
There should be other ways to think about the open horizon of modal life of an non-existing object. My hunch is that we can be more stringent on objecthood and then allow for a more relaxed notion of existence where existence is not a matter of all or nothing but rather a matter of degrees or - as I prefer - a matter of varieties, as in Souriau's existential pluralism. I guess there is no object associated to a description like that in (1). To utter (1) is not enough to bring about (my translation of Souriau's word instaurer) an object. Bringing about involves some existing material, fragments of other compositions, in my terms. But maybe we can allow the act of bring about to take place without this concreteness. A character of fiction, say Harry Potter, can be in a crossroad of many descriptions, so that there is something singular that is at this crossroad.
Fiction is typically made out of singularities. It is not about the thief and the owner of the house and the lover and the logician. The logician is also the thief and the lover is also the owner of the house etc. No one is fulfilling just one role. Existence is a bit like this, it is a common plane - the one that Deleuze and Guattari at the note on page 44 of their last book together refer to as a plan d'instauration, talking about Souriau. The lover happen to be the owner of the house etc. Identities given by descriptions are not exhaustive. Fiction is often like this as well, except maybe in Novarina's theatre... In any case, fiction often mimics existing things by associating a critical mass of description to the characters, they're at a crossroad of descriptions as well.
But at this point one could say: but this vaguely Davidsonian strategy of appealing to a critical mass of description is not just revamping good old cluster theory of reference that Kripke was clear at criticising? I don't know but I would venture to say no, for reasons that will please Tomas, I guess. The critical mass is needed not for existence but rather for referential use. That is, if we have enough descriptions, they become usable not only in attributive but also in referential ways. Referential usage can then be a criterion of objecthood. In other words, we can start considering that some descriptions can be false and yet be about Harry Potter. It takes more than one description to bring something about. In general, I would say it needs embodiment, but in the case of fiction, we could allow for more flexibility. Or is this still a different form of the criticised cluster theory?