Maybe that should be ‘Yablo on Hartshorne on God’. Stephen Yablo (see, for instance, ‘Textbook Kripkeanism and the Open Texture of Concepts’) points up the inevitable reply to Anselmian ontological arguments that the theist is equivocating on ‘possibility’. The Anselmian God is epistemically possible, sure, but that Anselmian God is not metaphysically possible. Yablo’s point is that this textbook Kripkean distinction is not available to the textbook Kripkean. For Kripkeans, conceivability and possibility pull apart in cases where the presentation of some statement S is possible, but the proposition expressed by S is not possible. The problem is that this does not (or does not obviously) happen in the Anselmian argument. The standard example goes as follows.
1. Water is not H2O. (metaphysically impossible)
2. The watery stuff is not H2O. (metaphysically possible)
If the reference of ‘water’ is fixed by ‘the watery stuff’ (or the liquid potable stuff, or what have you), then it is no more than *contingent apriori* that water is the watery stuff. When we are tempted to assert that (1) is possible, what tempts us is the genuine possibility of (2). Indeed, it is because (2) is metaphysically possible that (1) is epistemically possible. Hence the illusion that (1) is possible. Notice that you cannot substitute ‘water’ for ‘the watery stuff’ in (2) to get (3).
(3) Water is not H2O (metaphysically possible)
Try doing the same thing with the Anselmian God. What gives the illusion of possibility?
(4) God necessarily exists. (metaphysically impossible)
(5) The being instantiating the attributes of omnipotence, omniscience, &
omnibenevolence necessarily exists. (metaphysically possible)
What is supposed to happen is that the metaphysical possibility of (5) provides the illusion that (4) is possible. Indeed, it is the metaphysical possibility of (5) that makes (4) epistemically possible. But the Anselmian denies that the name ‘God’ is fixed by the description ‘the being instantiating the attributes of omnipotence, omniscience, & omnibenevolence’. Rather they take it to be *necessary apriori* (not contingent apriori) that God is the being instantiating the attributes of omnipotence, omniscience, & omnibenevolence. *So, if it is conceivable that God necessarily exists, then it is possible that God necessarily exists*. There is no room (or no obvious room) for conceivability and possibility to pull apart. That is, given the apriori necessity that God is the being instantiating the attributes of omnipotence, omniscience, & omnibenevolence, the substitution of these terms goes through, and (6) is true.
(6) God necessarily exists. (metaphysically possible)
So it looks like Hartshorne’s Anselmian argument from conceivability to possibility goes through! But, as Yablo notes, the traditional argument seems to fail anyway. He points up the problem that it seems equally true that we can conceive of the Anselmian God not existing. Maybe that’s right. But I think the problem is with the claim that (5) is metaphysically possible. That is, the problem is with the epistemic possibility that Hartshorne’s Anselmian God exists–certainly this is where the atheist is on pretty good dialectical ground. If Yablo is right, then the reasons the argument fails are non-Kripkean. But it looks to me like the reasons the argument fails are indeed Kripkean. It is not that conceivability is failing to entail possibility *when a Kripkean says it should*. It is rather that we have a failure of conceivability.
One thing interesting here is that we are not getting some ordinary inference failure from conceivability to possibility. Nor do we in general get such a failure when the inference is from the conceivability of a proposed conceptual truth that G is P, to the possibility that G is P. We get such an inference failure only in cases where we infer from a non-conceptual truth that G is P to the possibility that G is P. In the case of a proposed conceptual truth that G is P, we at most find that there is no such conceptual truth. The objection then to the Anselmain ontological argument is that it is not a conceptual truth that God instantiates the typical divine attributes. And that can be a difficult objection to sustain.