Edward Tufte

Author of ...

Which correspond to:

EdwardTufte has one main idea, and it's simple, but powerful: show people as much data as possible with as little ornamentation as possible. Let the data speak for itself. He lays out some design principles in TheVisualDisplayOfQuantitativeInformation:


If Tufte ever comes your way, or if you get the chance to go out of your way to attend one of his day-long courses, jump on the opportunity. Tufte is an exceptional lecturer, and has a wonderfully wry sense of humor.

A key point is that good presentation of data answers a question about the link between cause and effect. Too often, graphs answer the wrong question, or they're so cluttered up with "chart junk" that meaning is obscured. (Tufte is not a fan of PowerPoint.)

To drill this point in, he showed how ineffective presentation contributed to the ill-fated decision to launch the Challenger. Morton Thiokol initially provided the right data to indicate that low temperature increased the risk of O-ring failure to NASA. (Copies of the faxed notes Thiokol engineers sent in VisualExplanations). They failed to effectively support Thiokol's initial no-launch recommendation - the presentation was confusing, and it didn't establish the causal link between temperature and failure. Tufte then showed how a simple rearranging of the available data, clearly linking failure to temperature, provides a far more compelling presentation.

His course is one of the best I've been to in recent years. I now look back on presentations I've made about software schedules, and cringe.

-- DaveSmith

To get on the mailing list for future tours, contact Graphics Press at 800-822-2454 (9am-5pm EST).

Envisioning Information is printed in 23 colors. A beautiful book, filled with all sorts of interesting examples from throughout time and around the world. Here's a taste of Tufte:

"Through an elegant chain of visual reasoning and with characteristic bluntness, Galileo, writing from Florence in August 1612, converts empirical observation into focused evidence supporting conclusions. His argument unfolds the raw data ("what the eye of the forehead" registers) into a luminous explanation of mechanism ("what the eye of the mind" envisions), a deeply visual logic that produced precise insights far beyond those achieved by others who had also observed sunspots in the early 1600s." (p. 19)

As a counterpoint, while I appreciate his ideas, I found his books harder to read than they could have been. He has a very good sense for choosing images that demonstrate his points, but often he doesn't explicitly tie those images to specific text passages. This left me wondering "should I look at this diagram yet?" whenever I turned a page. The nadir came on page 155 of TheVisualDisplayOfQuantitativeInformation (15th printing), in the first paragraph: "Look at all the different levels of detail created by this population density map [...]". What map? There's no map on p. 154-155, nor any indication that the map he's referring to is in fact spread across the following two pages. That's not good design.

Also, Tufte seems to gear his visual representations to the higher end of the reader sight BellCurve?. For instance, chapter 6 of TheVisualDisplayOfQuantitativeInformation offers a redesigned box plot. The redesign is so subtle (using single-pixel offsets & gaps) that it requires very good sight to recognize that what appears to be a straight line isn't.

There's much wisdom in the three books, and I strongly recommend them. It was saddening, though, to see his books marred by a few poor design choices. They are obviously a labor of love, but do not have the QualityWithoutaName that suffuses ChristopherAlexander et al's pattern language books.


View edit of March 20, 2004 or FindPage with title or text search