The Aporetics of Existence and Self-Identity

Andrew B. made some powerful objections to a recent existence post.  His remarks suggest the following argument:

Argument A

1. Existence is self-identity
2. My existence is contingent:  (∃x)(x = I) & Poss ~(∃x) (x = I)
Therefore
3. My self-identity is contingent:  I = I & Poss ~ (I = I)

Argument A may be supplemented by the following consideration.  Since I am contingent, there are possible worlds in which I do not exist.  Not being in those worlds, I cannot have properties in them, including the property of self-identity. So it is not the case that I am necessarily self-identical; I am self-identical only in those worlds in which I exist, which is to say: I am contingently self-identical.  I am self-identical in some but not all worlds.

The argument can be rationally resisted. 

Consider a possible world w in which I do not exist.  In w, the proposition expressed by an utterance by me of 'I am not self-identical' is true.  But if it is true in w, then the proposition exists in w.  Now if the proposition exists in w, then so do its constituents.  On a Russellian view of propositions, I am one of the proposition's  constituents.  So for the proposition  *I am not self-identical* to be true in w, I must exist in w.  But if I exist in w, then of course I am self-identical in w, and the proposition is false in w.  But the same goes for every world in which I do not exist.  It follows that I am self-identical in every world and I exist in every world.

Of course, one needn't take a Russellian line on propositions.  One could take a Fregean view according to which propositions about me do not have me as a constituent but an abstract representative of me, a sense or mode of presentation.  But the first-person singular pronoun 'I' has the peculiarity that it cannot be replaced salva significatione by any description; so even if there is an abstract representative of me in the Fregean proposition expressed by my utterance of  'I am not self-identical,' there still has to be a referent of the representative external to the proposition.  So I have to exist in w for the proposition *I am not self-identical* to be true in w.  But if I exist in w then I am self-identical in w.  This in turn implies that the proposition is not true.  

The cognoscenti will appreciate that what I have been doing in a rough and dirty way is reproducing some of the thoughts in Timothy Williamson's paper Necessary Existents.  I am doing so to show that Argument A is not convincing.  Making use of materials from Williamson's paper, we can 'throw Argument A into reverse':

Argument B

1. Existence is self-identity
~3. My self-identity is necessary: Nec (I = I)
Therefore
~2. My existence is necessary.

In point of validity, there is nothing to choose between A and B: both are valid.  And both, I submit, have counterintuitive conclusions.  It seems to me that the arguments cancel each other out.  So I propose that we think very skeptically about the common premise that existence is self-identity, and the Quinean thin theory that commits us to it.