"The Cognitive Style of PowerPoint: Pitching Out Corrupts Within" is little more than a booklet, at 32 pages. It is, however, quite a bit of fun. Tufte writes qualitatively about improving the presentation of quantitative data, and so I always end up wondering how much he's really said when I read his work. On the other hand, he does present his main thesis pretty sensibly, even if it feels a little superficial.
His argument is that PowerPoint is a terrible tool for presenting the analysis of data for decisions. At least, I think it's really this he's after. The argument that a tool designed for selling stuff is probably a really bad tool for presenting statistical information to make important decisions with is pretty solid. My heart sinks remembering some of the internal presentations where a .ppt file has been the generator of a Reality Distortion Field.
So, what's the right alternative? His argument is that for most cases, scientific work should be presented through some form of paper. This seems fairly reasonable, at least in that there should be a longhand verion of the information available somewhere, given the way that a bunch of PowerPoint slide out of context convey pretty much nothing. However, I'm not sure what the right tools are for trying to transfer ideas and concepts rather than just the results of scientific tests. Furthermore, it's becoming increasingly difficult to break away from PowerPoint - in a business setting pretty much anything other than a slide show seems "unprofessional", and a proper pedogogic approach, well, requires a lot more effort.
All in all, this booklet emphasises what's bad in PowerPoint, but I'm not particularly impressed by the alternatives it offers. Having said that, I'm hoping that it provides at least the nucleus of a business case for switching away from using PowerPoint for every single presentation.