I read it that way at first too. But if you read a little further, the article centers on providing evidence for the model. Specifically he shows evidence that a a state with higher G has linearly higher D (which is exactly what the model states).

And then when the evidence does not match up, he fudges with one of his parameters, V, to "explain" it.

This seems like a non-sequitur. The gp's criticism was that the article doesn't cite any evidence for the model, which he does.

Besides, he doesn't "fudge" V. He shows evidence: that violent crime decreased over the relevant period and argues that less violent crime implies lower average tendency to violence (V).

Well, the Australian cases of drug laws / murders provides enough validation for his case.

Except that in both Australia and the UK the gun murder rates were already incredibly low before their respective gun bans. Their bans didn't actually lead to less violence than before their bans; in fact, violence actually increased post-ban.

This would only be scientifically defensible if there was another Australia, identical in every respect, with only one thing changed -- the thing being measured.