As for Gaunilo, that’s a bit weightier. I think the response would go on something like the following lines: Anselm’s argument depends first on our having the idea of the perfect being in our minds. A perfect being is a logically consistent concept, and therefore can be formed in our minds; but a perfect island isn’t. Consider even the property of easy access, which you mention: easy access for whom? It would seem that the perfect island for you isn’t necessarily the perfect island for me; the idea of an island which is perfect (full-stop) comes to be a logical contradiction. The same is going to be true of any entity that is finite, limited by space and time: for such an entity to be perfect is not even going to be imaginable. This isn’t true of the greatest being, because the concept of a greatest being doesn’t contain a contradiction in the same way. For that being, and that being only, we can form a conception of its being perfect in our minds, and therefore (so the argument goes) must be able to do so in reality.

This is were my problem with the argument always comes into play. I don’t know why Anselm came to the conclusion that the existent thing is somehow superior to the imagined thing. I think that this portion of the argument often goes uncontested because of our inclination toward embodiedness. It is natural to think that, say, an real apple is superior to an imaginary one. However, I do not think that this distinction is useful. The idea of an apple is not an apple. I am reminded of Gaunilo. Basically he applied the same logic to the idea of an island. Imagine an the perfect island… blah blah blah. If the same logic held we would have an island that is perfect (and I think we would know where it was too. I mean, the perfect island would have easy access don’t you think?) Anyway, while I agree that this proof does have the highest set sights of the proofs, I disagree in that I think it fails just as miserably, if not more so, than the others.

Whatever the argument’s weaknesses are, I don’t think that ignoring the gap between imagination and reality is one of them. Much of the argument rests on bridging exactly that gap. God can exist in our imaginations; that’s the first premise of the argument. The remainder of the argument is to show that that existence in imagination implies an existence in reality.

But I think the rest of your post is on the right track. The argument depends on a very large number of disputable metaphysical and epistemological assumptions, including among them the idea that greatness is itself something really existing, not something that differs in the mind from person to person – it’s certainly not compatible with a subjectivist conception of value. However – and this is why the argument doesn’t just immediately fall flat – any of those assumptions can also be (and have been) argued for on grounds acceptable to an atheist, grounds that don’t require God’s existence.

I’ve always felt that the logic goes wrong because it ignores the gap between imagination and reality. Maybe there is no gap between the logical/conceptual and the actual, but that’s certainly not an assumption an atheist (whom Anselm is effectively arguing against) would accept. An equally applicable final sentence of the argument could be, “…therefore it’s impossible to adequately imagine god,” or a less denigrating version, “…therefore any god you imagine would not be as great as a real god.” (Though I confess I haven’t read the full/original version.)

Regardless, this certainly provides a highly *interesting* proof of god, as you lay out in the rest of your post. But there’s a flip side: it also produces a personalized god, one that’s the best each imaginer can imagine. Which I think may speak to a subtler flaw in Anselm’s argument: if god exists because he’s the greatest thing I can imagine (which includes the quality of existing), what happens if you and I imagine radically different “greatest things”?