Paul Graham does a good job outlining the different forms a disagreement can take and evaluating their effectiveness. I agree with his analysis. I'm going to expand on it.
It seems this piece was written in response to the various criticisms he received because for his "You Weren't Meant to Have a Boss" essay. In addition to objecting to the form these criticisms took, I think he could also have talked about one of the fundamental causes of most disagreements on the internet - lack of charitable interpretation.
Many interpreted his essay as an attack on big-company programmers - using as evidence the analogies he used, the frankness of his writing style, etc. - and reached the conclusion that he had an elitist attitude about his profession. Now, this may or may not be true (having met him, my opinion is a firm "not true"). But the point is that it doesn't matter. If we interpret charitably, we remain agnostic on what his argument implies about him (thus abstaining from DH1 and DH2 attacks) and instead focus on the actual argument. Some critics managed to do that. But many didn't. And the worst part is that the lack of charitable interpretation obscured their actual disagreements, so now we've wasted time talking about "how to disagree" instead of "what are the correlations/effects of company type on programmer happiness/ability/prospects".
Of course, an author should be careful to frame his arguments in a way that minimizes the danger from misinterpretation. In other words, an author should try to write well.