ANOTHER FORMULATION OF THE ARGUMENT

OK, I've presented the traditional version because that's the one that keeps coming up in the literature, and so students should know about it. But there's a better version.

Does God exist by definition? While it is part of linguistic convention that the word "God" is used to denote a personal almighty being, I don't think that existence is part of the meaning of the word (in contrast sisters, by definition, are female, and hexagons, by definition, are six-sided).

However, it is a striking feature of human language that we can always, whenever we want, invent new terms, so long as our meaning is clear to our audience. For instance, I am free to stipulate: from now on, "yod", by definition, shall refer to any three-legged horse (whether or not yods have to be roan, and whether or not yods exist, are features that are not built into my definition). Likewise, I am free to stipulate: from now on, "hod", by definition, shall refer to an existing God (whether or not a hod wants to keep the seventh day for resting is a feature not built into the concept).

Just as a yod must be three-legged, a sister must be female, and a hexagon must be six-sided, so must a hod exist (it's true by definition). And just as my spotting a yod entails my spotting a horse, and my having a sister entails my having a sibling, so too does a hod's existence entail God's existence.

Objection 1: Existence is not a predicate

Kant admits that we can generally make the inference from (a) to (b):

(a)  "X", by definition, signifies property F.
(b)  An X is necessarily F.

However, there's one restriction: "F" must be a genuine predicate, and existence is not a genuine predicate.

Kant's position amounts to a naked rejection of the Ontological Argument. Kant claims that "exists" is the one verb phrase, out of all verb phrases, that does not express predication. But why should existence be exceptional? (Maybe it is, but you would need an argument for it.) Furthermore, Kant claims that lack of genuine predication renders the Ontological Argument unsound. (Maybe so, but again you would need an argument for it.)

In fairness to Kant, the view that "exist" is not a predicate is enshrined, for independent reasons, in standard (Fregean) logic.

Objection 2: Reductio ad absurdum

Regardless of how the Ontological Argument goes wrong, it clearly does; for by it, we could prove the existence of anything! In criticizing St Anselm's original argument, for instance, a fellow monk by the name of Gaunilo observed that Anselm's reasoning proves the existence of a perfect island. Let "faradise" refer to an actually existing tropical isle easily accessible from my bedroom closet. Voila, I should now be able to step into my own private eden! Since I can't, something has to be wrong with Anselmian reasoning, even if we can't locate the problem precisely.

Incidentally, Anselm's argument is badly named (though centuries of tradition will prove irresistable). Granted, the conclusion reached (that God exists) involves ontology (that which exists). However, in this sense all theistic arguments are ontological. No, what is distinctive about Anselm's argument is that it, unlike others, is based on a premise that attempts to bridge the gap between words (or concepts) and their referents. For this reason it would be better to call St Anselm's argument the Semantic Argument.