Edward R. Tufte’s Envisioning Information

ei_bookcoverI picked up–and finished–Edward R. Tufte’s Envisioning Information yesterday. At 121 pages and filled with dozens of photographs, sketches, maps, charts, and graphs, it a mild read, but what a joy.

With illustration and insight drawn from Paul Klee’s critical writing, Galileo’s journals, and interviews and narrative excerpts from novelist Italo Calvino, Envisioning Information is designed for the humanist-statistician. Incorporating elements from graphic design, psychology, education theory, history, art, music, and architecture, Tufte makes it clear that there is no hard scientific or practical work that has value beyond its ability to be communicated to others.

Moreover, in the spirit of the Marshall McLuhan and other fundamental communications theorists, Envisioning Information is a model of the ideology it proscribes. The book is clearly written, graphics always appear on the same page on which the explanatory text appears (except in one instance to prove the importance of that point). Each page or two-page splash thoughtfully considers and balances white and dark space, text and graphics.

The book, also true to its ideology, is self-effacing. There is no grand statement that, at the close of the 20th century, the author and his field are poised on the edge of a whole new world. Computer graphics, really still in their infancy, are given special consideration for their special limitations (low resolution, and specifically the low resolution/high resolution frontier apparent at the computer-human interface) but are otherwise treated as just one more example of “flatland,” the theoretical 2D space that warps and constrains natural human thought and perception. Many, indeed most, of Tufte’s examples are from 16th-19th centuries. Certainly he recognizes advances in printing technology and overall human understanding to advances in envisioning and managing great information density; but he is also aware that these are largely mere refinements of previous methods. Tufte’s range of illustrations come from Galileo’s ingenious method of observing and tracking sunspots (in an age where conceiving of spots on the sun ran contrary to the church’s scholastic doctrine), to 18th century French dance manuals, to mid-20th century Japanese tourism pamphlets. He seems to have a special affinity for train schedules.

Tufte’s interdisciplinary, human-centered approach to visualizing data is not confined to superficial uses of examples from across time and cultures. It is his rare vision that allows him to see that a map and a table communicate the same information in the same way; it is his gift to be able to describe what he means when he says that. At one point, he compares a New York->New Haven train timetable to a local courthouse. The unnecessary columns of the timetable provide a similarly false sense of order to the timetable that the Doric columns do to the courthouse: “…the real work done in backrooms.” Magnificent.

The book contains no throat clearing. The short introduction is placed uncomfortably, cut across two-pages after the dedication and before the Acknowledgements. It is not treated as Chapter 0 as it is in most books on any subject (back to self-effacement, I presume). From there it jumps immediately into the concept of “flatland” (named after a very strange geometric romance by “A. Square”). Tufte waste no words in describing it or its subtle confundations explicitly. Rather he just describes its effects and trusts his readers to do the thinking on their own–another of the books ideological principals: graphics are not used because “information is confusing” or “audiences are numbskulls.”

As the book progresses, it becomes more and more useful to the novitiate or intermediate data visualizer. Each successive chapter contains more immediately practical advice than its predecessor. Tufte first presents good visualizations and failed ones. Later he will provide contrived good and bad examples to contrast. Later still he will show modest to good examples re-envisioned by him or his team to highlight the dramatic effects of minor changes in the use, thickness, or color of gridlines (for example).

As a How-To manual, the book does not offer much, which is unfortunate because I am very much in the “how-to” phase of learning data visualization. But the book was such a joy to read its impossible to critique its value based on my needs alone. I wanted and expected more of a how-to manual. What I got instead was a snapshot of 400 years (actually more) of humans slowly learning how to be understood. I learned how a 19th century re-envisioning of Euclid’s geometry invented Modern Art. I learned how to track sunspots. I learned why the Vietnam War Memorial is so entrancing and moving. And I learned that, from space, Earth resembles the Charles Schultz character Pigpen. I probably took away about five or six invaluable, timeless, and all purpose data visualization rules as well, which ultimately is the least of the books gifts.