Beautiful Evidence
Beautiful Evidence
Graphics Press, 2006 • ISBN: 0961392177
Of all Tufte’s books, I felt this was most like a curate’s egg. Good in parts, but quite a bit of stuff we’ve heard before, and some less-palatable portions.
It’s a bit of a grab-bag. One chapter, rather grandiosely titled Fundamental Principles of Analytical Design, is yet another discussion of the Minard graphic showing the retreat from Moscow. The fundamental principles, if you’re interested, are: use multivariate data, documented, integrated with explanation and text, showing causality and explanatory comparisons. There’s a chapter famously and largely justifiably condemning PowerPoint, already available for years as a separate pamphlet. There’s a nice explanation of sparklines, 
tiny trend lines that can be integrated into text, also posted online. There are photographs of Tufte’s sculptures, for some reason. And there’s a discussion of figure/text integration in the Hypnerotomachia Poliphili. Now, I confess to being a big typography nerd, so will happily read all about the Hypnerotomachia. But I wonder what most of the scientists I know with Tufte tucked dutifully on their shelves will make of it.
What I found most bothersome was the discussion of cladograms, (evolutionary trees inferred from attributes of living organisms). Tufte initially posted a much more critical analysis of a fairly typical cladogram, with jabs at the use of well-understood technical terms like “strict consensus” (Tufte wrote, “Sounds like 
marketing, not science.”) There was a subsequent pile-on by some very reputable systematists, and Tufte backpedaled, claiming he was just a curious interloper in their field. Nevertheless, the watered-down version of the critique that made it into Beautiful Evidence still accuses biologists of using feel-good pitch words with “broad cheerleading meanings” to cloak their dodgy “editorial judgements” and questionable “statistical crunching”. Yes, cladograms are information-poor, and convey undue certainty to those who aren’t familiar with them, so Tufte’s skepticism is perhaps excusable, if not really justified. Skepticism vanishes a few pages later when lauding Feynman diagrams, which to me seem to contain far more dodgy handwaving than a good cladogram.
If you already have the first three books, this is by no means necessary. There seems to be a trend towards increasing length, cost, use of color, and hodgepodgery in Tufte’s books; perhaps the fifth in the series will buck this trend when it comes out it about (I predict) 2015.
Comments
Hi Mike.
I've got all four of Tufte's books. Your review is quite accurate, even despite the potential bias of a typophile biologist.
However, i found the quirks to be endearing. Rather than thinking Tufte was overreaching at times, it just seemed to be his style. Ironic, considering his composure in the treatment of data, he can't seem to contain himself when it comes to expression (case in point: his landscape scuptures).
With the cladogram, i think he is assuming that the reason for showing the data ought to be evident in the presentation of it (without the possibility of miscommunication). But this seems to conflict with his premise: don't assume the audience is dumb, assume they are uninformed. Somehow i don't think he would be satisfied with a note "objects in the cladogram are not as certain as they seem." It is as though he wants the viewer to draw their own conclusion by way of prohibiting the infographer from presenting any conclusion. Having said that, i wonder if there isn't a better way to do the cladogram which doesn't look so explicit. Where is the line between Selling vs. Telling?
Posted by: m4tts1m | May 10, 2007 9:33 PM