I still am concerned about the Benacerraf Dilemma; that it works.
If we say that there is an entity or a class of entities which are abstract entities or propositions, it seems to me that there has to be a sort of closed criteria to differentiate that single or that single kind of entity. Imagine if we thought that there was such thing as a cat or a kind of thing as cats, but that all sorts of other animals could be identified as cats.
It seems to me like having ten photographs of alleged Sasquatches. Now some true believer wants to say that in any given photo there is a Sasquatch, and in the totality of photos a kind of thing that is a Sasquatch. If the skeptic says: That could be anything in the photo!, can the true believer really get away with saying: Oh well, I guess there is a Sasquatch structure or function which many different things and kinds of things can fill! This seems absurd.
Or imagine finding a dead body. This seems like compelling proof of a murder to the conspiracy theorist. He can construct a theory based on evidence to identify the murderers as Smith or Jones. But the skeptic can point out that the theory is so loose that anyone can fit this role. There are alternative explanation for the dead body for the skeptic. There is no single killer here, nor even a single kind of killer for this body.
If "The dude who wrote Naming and Necessity", "One half of Kripkenstein, the half who isn't Wittgenstein", and "Saul Kripke" are all equally good linguistic representations of Saul Kripke, don't we have to abandon the notion that there is a single linguistic representation of Kripke, or a single kind of linguistic representation of Kripke? I'm not even sure that there is a single type or class of linguistic expressions that pick out Kripke. There are many equally good linguistic representations, many equally good kinds of linguistic representations.
This all seems like a Wittgensteinean move against Universals, with the example of what makes all games games. It seems there is just a family resemblance, a messy cross-section of related but distinct criteria. If we can say that there are cat entities and a kind of entities that are cats, we surely avoid the B.D. If we cannot say that there are proposition entities and a kind of entities that are propositions in the same manner, we should probably drop the notion that they are entities like cats are entities. I don't see this as overly devastating except to Platonism, which is false anyhow.