Tufte, Beautiful Evidence

Edward Tufte’s most recent book Beautiful Evidence follows in the tradition of his previous books, arguing for visual displays with high information density. Much of the argument is by example, annotated reproductions or redrawings of images Tufte identifies as “graphical excellence.”

Beautiful Evidence introduces the idea of sparklines: “Why not also construct data graphics that work at the resolution of routine typography? Thus the idea of sparklines: small, intense, wordlike graphics” (48). Examples include medical, financial, and sport data (for example, the win-loss history for a baseball season). Here’s the federal budget deficit from 1993 to 2003: Deficit 1993-2003 On the one hand, I’m very hip to this. But on the other, they often seem too small. High-resolution display of information doesn’t equal information delivery. As Tufte points out, paper can deliver hundreds of data points in the space of a word, and we can certainly perceive it (that’s how we quickly differentiate typefaces). But that doesn’t mean it’s possible to make meaningful use of very small graphics. I think Tufte’s suggestion to consider sparklines as fractals plays well; each should link to larger, ideally interactive graphics of the same data. In practice, this is less difficult than it sounds, since for a display of multiple sparklines like this one only one software implementation is needed. Indeed, several software implementations of sparklines have appeared; the one above is generated using a PHP library, and this Python-based generator includes a web interface.

“Words, Numbers, Images—Together” looks at several classical works (e.g. Galileo’s Starry Messenger) which integrated image and text heavily. Why isn’t this more common today? Tufte points out digital tools discourage it–we use different programs to store data, create graphics from it, and integrate those graphics in text. Today few books emulate top-notch field guides, which show “a sense of craft, detail, and credibility that comes from gathering and displaying good evidence all together” regardless of its textual, pictorial, or graphical nature (115). And as we know, our curricula discourage it, too.

Beautiful Evidence goes on the offensive in two chapters which read like a unit, “Corruption in Evidence Presentations” and “The Cognitive Style of PowerPoint: Pitching Out Corrupts Within.” The latter updates the 2005 pamphlet “The Cognitive Style of PowerPoint,” an all-out assault on PowerPoint which includes criticism of PowerPoint slides related to the 2003 loss of the shuttle Columbia. Many readers have focused on the shortcomings of PowerPoint, which for Tufte epitomizes the modern software corporation, a bureaucracy focused on programming and marketing (161). Yes, the information density is abysmal, the typography nearly unreadable, and the AutoContent wizard encourages changing the message to fit the template (all flaws Tufte discusses). But Tufte also shows that PowerPoint presentations replace technical reports followed by discussion, best summarized by the IBM anecdote which leads off the chapter, where Lou Gerstner cut off a formal presentation and said, “Let’s just talk about your business.” PowerPoint is replacement for discussion of literature, following on the heels of other presentation, meeting, or teaching formats which replace talking about reading. Eliminating PowerPoint wouldn’t eliminate the problem, which is cultural, as the loss of two shuttles and fourteen astronauts painfully shows. What underlies the problem PowerPoint? Failure to consider complexity; the drive to find a simple, conclusive answer, often via secondary or tertiary repackaging of information.

However, some material just doesn’t seem to fit. The chapter on sculpture pedestals and the accompanying photo essay are interesting, but I just don’t see the connection to evidentiary matters. Tufte also recycles examples, reading Minard’s Grand Army graphic to extrapolate six principles of analytical design. This would be very useful as a standalone pamphlet, or a nice way to epitomize Tufte in a course pack, but for folks familiar with Tufte, it feels like a retread. Beautiful Evidence has many valuable insights, but I don’t see a book’s worth–at least not a book as powerful as The Visual Display of Quantitative Information, still by far Tufte’s best.

Side note: Tufte has updated his list of courses for 2008. I’ve signed up for the Thu Jun 19 course in Chicago.

This entry was posted in Reading. Bookmark the permalink.