I find Kant's refutation most convincing. His claim is: "Existence is not a predicate." My intuition on his point is this. Let's consider the following statement: "Imagine a unicorn, with the property that it happens to be brown." There is an implicit and unstated assertion at the beginning of the above statement, that we are for the moment imagining that that the unicorn exists. We might rephrase as follows: "Imagine for the moment that a particular unicorn exists, and that it has the property that it is brown." Now we attempt to attribute to this unicorn the "property" of non-existence: "Imagine that a particular unicorn exists, and that it has the property that it does not exist." The statement is logically ill-formed, since we assert of a hypothetical entity that it simultaneously has and does not have a "property", namely that of existence. Hence, "existence" cannot be considered to be a property whose presence or absence can be attributed to a hypothetical entity. Kant argues that since Anselm's Ontological Argument depends on this logical fallacy, it is shown to be invalid.
Has anyone evaluated Godel's reformulation of the Ontological Argument to see if it avoids Kant's critique? I can't help but imagine that Godel was aware of Kant's (and others') analysis of the Ontological Argument.