More Guns => More Crime or Less Crime?

Megan McArdle has what are at least the first steps towards a theoretical model relating the prevalence of guns to crime levels. As far as I know, it's untested against real-world data, and I don't necessarily agree with all the assumptions being made, but as a first-order approximation, it makes a certain amount of sense:

In other words, the curve will peak near the point where the majority of criminals who want guns have them, and the majority of law-abiding citizens do not.

In my view the biggest assumption being made here is the idea that we can neatly divide the world into criminals (who use guns for nefarious purposes) and law-abiding citizens, who do not do so. We can do "more likely" vs. "less likely," but I'm worried that this model neglects how easy access to guns can enable a formerly-law-abiding citizen to elevate a fit of passion to greater lethality.

UPDATE: This discussion on BoingBoing led to this academic paper analyzing data on multiple shootings vs. state laws readily allowing concealed weapons. The data strongly support (more so than I had expected) the theory that "shall carry" laws prevent mass shootings. Virginia has such a law, but handguns were not allowed on campus.

UPDATE 2: Check the comments for a link to an article pointing out serious flaws in the aforementioned paper's research. I've only given a quick read to both the original paper and the response and haven't scrutinized the data in either one; I'll leave the links up to let you make your own conclusions.

UPDATE 3: Meteor Blades on Daily Kos has some very well-reasoned thoughts about what level of gun control vs. gun freedom he'd like to see; his positions closely mirror my own, with the caveat that I don't actually own any guns myself.