The point of his article is that we run into an absurdity so long as we understand akrasia to be 'knowingly acting against your self interest' (or some equivalent variation thereof). Suppose I have before me actions A and B, and I judge that A has greater utility. Then I do B.

If this is my problem, we can as easily solve it by raising the utility of B (until my doing B instead of A is no longer irrational) as we can by lowering the utility of B until it is no longer tempting. But it's manifestly absurd to think that I can cure akrasia by raising the utility of B (as Sorensen ironically recommends). Yet nothing about our understanding of akrasia explains this absurdity.

So it must be that our understanding of akrasia is faulty. That's the point of the article.