> Notice that the original code hasn't changed. The only information transmitted in the answer is the corrected diagram. That is because to the person asking the question, the diagram is a better representation of their mental model. As such, getting an corrected diagram has an effect on their mental model but looking at the code doesn't.

This argument (that he tries to make several times in the article) does not hold.

Almost every time, the diagram is a _lower level_ representation of the program than the code is. And then he says "look! you can't figure this out from the code" (so therefore diagrams are better), but if the code was similarly represented in a lower level, you totally could.

And similarly, if the diagram happens to _not_ contain this extra lower level information, you can't figure it out from the diagram either.

I'm not saying diagrams aren't good, they can be great, it's just the reasons in this article aren't particularly compelling. But maybe I'm missing the point.