July 19, 2007

Disappointing reading

After crunching through three rounds of papers, it's time to take a break and review Edward Tufte's Beautiful Evidence (2006), the fourth in a series of coffeetable books on presenting information: The Visual Display of Quantitative Information (1983), Envisioning Information (1990), and Visual Explanations (1997). Tufte's website is a cornucopia of fascinating discussion of graphics and shameless huckstering. As Tufte's reputation for sharp-witted analysis has grown, I wondered whether his latest book would be as effective as Envisioning Information or cement his reputation as a true info-guru, the Font of Wisdom for followers but all too clay-footed for the rest of us.

Tufte walks along the ground. With the exception of one serious contribution, this book is largely superfluous. Go to Envisioning Information, drain it for all it's worth, and then read up on sparklines and find tools for creating them online. But you can safely skip Tufte's latest book.

Tufte's central point in the book is that one can present quantitative information effectively, elegantly, and even beautifully while making arguments. To this task he brings a lifetime of knowledge, a catalog of famous displays of information and infamous corruption, and odd examples that show Tufte's skill in spotting and explaining technique. So you will find in the book Galileo's drawings of Jupiter and its moons, a horrifically clinical depiction of slave galleys, and an effective demonstration of Cezanne's multiple-perspective cubism, as well as Tufte's praise of a skiing manual, an 18th century dancing diagram, and a NYC police departmental poster showing how to spot hidden handguns. Tufte brings Galileo's dispersed drawings together to show the moons' path through several dozen observations, something Galileo didn't do (and Tufte implies he could have, though he admires Galileo's integration of text and image).

Given the intervals between the books' publication dates, one should neither be surprised nor too distressed that many of Tufte's ideas recur in the various books. So too do some images. Charles Joseph Minard's heart-stopping graph/map of how Napoleon's army disintegrated in the Russian campaign appears once again... in part because Minard's image deserves such study (and has been reworked by several statisticians and others). In itself, such repetition is neither surprising nor disturbing.

Nor is Tufte's true contribution in the book, a chapter introducing a type of condensed charting he calls sparklines. Tufte's idea is interesting and a legitimate contribution, even if I think it is better suited to data-rich time-series that one wants condensed for small multiples than a panacea for presenting data. Fortunately, even for us crass users of Excel, there are add-ins to create sparklines where appropriate.

Beyond the introduction of sparklines, however, the book contains nothing new for those who care about presenting information clearly. The book has some odd sequences that are distracting rather than informative and at least one ill-informed screed that consists largely of overblown rhetoric. First and briefly, the bizarre bits: In a chapter putatively about the integration of text, images, and numbers, Tufte includes an interlude discussing the Hypnerotomachia Poliphili, a 15th century erotic novel with woodcuts, and a 2006 prosecutorial sentencing memorandum used in the bribery case against Randy Cunningham. In the same chapter, Tufte interrupts his discussion of Galileo with a two-page spread from Matisse's Po├ęsies Antillaises. He finishes the book with unexplained color plates of his own sculptures. Beautiful? Some of these. But a reader must wonder what this is all evidence of. Certainly not information design.

The last substantive chapter is a 29-page jeremiad against powerpoint that was put better in a three-panel Dilbert cartoon August 16, 2000, where one audience member succumbs to "powerpoint poisoning" after seeing slide 397 (search for the phrase to find probably-copyright-violating images of it). Okay, so we know presentation software is often abused. But Tufte takes a ludicrous leap from the banal uses of presentation software to the claim that the overuse of PowerPoint doomed the Columbia shuttle in 2001. Tufte spends seven pages on this argument.

The central theme of all the journalists reporting on the Columbia accident investigation was that the problem was a NASA culture that put safety behind all sorts of political and bureaucratic considerations. How much did the Columbia Accident Investigation Board spend on the topic in their first report volume? The accident report included a sidebar on p. 191 of Volume I that includes Tufte's analytical text, wording similar to what eventually appeared in Beautiful Evidence. The board's conclusion in the sidebar is important enough to quote in its entirety:

At many points during its investigation, the Board was surprised to receive similar presentation slides from NASA officials in place of technical reports. The Board views the endemic use of PowerPoint briefing slides instead of technical papers as an illustration of the problematic methods of technical communication at NASA.

The word illustration is important here. NASA culture killed the Columbia astronauts. The same culture had existed 15 years before, when Challenger exploded. PowerPoint had not. Why should we blame PowerPoint when its use is a symptom and not a cause?

That sin of commission is matched by a sin of omission: Tufte did not explore any redesign of visual presentation software. In his decades-long fight against chartjunk, Tufte advises redesigning the presentation of information. Lighten the lines, reduce obnoxious color contrasts, eliminate shadows and other distorting 3-D effects, keep quantities visible, etc. Those are specific steps to take. Tufte's advice on PowerPoint? Don't. That's all. Just don't. Use handouts instead (though too many handouts and the audience sleeps through a talk). Forget that you have 50-70 square feet of visual space you could use to help communicate.  Just fuhgeddaboudit.

My advice on Tufte's latest book? Fuhgeddaboudit. Go read up on sparklines and get his other books instead.

Listen to this article
Posted in Writing and editing on July 19, 2007 9:51 PM |