Tufte is often wrong about what constitutes good communication. Indeed, I am surprised he likes the Napoleon map so much because it has, in his terms, superfluous chart chunk - those drawings of soldiers. This is indeed an excellent graphic, but much of his work does not have this character.
Tufte is not the only statistician who has addressed the problems of representing graphical material. In my opinion, Bertin is the best.
Tufte preaches. I entered into a discussion with him about this once and tried to present some experimental data that one of my students had collected. he refused even to look at it. That is, it isn’t that he looked at the data and disagreed with the interpretation or even the collection— that would be permissible. No, he refused even to look.