Illustrating Data

I just finished reading Edward Tufte’s book entitled The Visual Display of Quantitative Information. As the title implies, this book is about designing statistical graphics. What a topic, right? The subject sounds about as dry as a Mojave riverbed in August. However, this book is exactly the opposite.

Tufte explains the principles of good (and bad) statistical graphic design with a sense of enthusiasm that borders on zealotry. He compares examples of illustrations throughout history and shows by example how some simple modifications to a graphic can turn it from a pedantic, heavy-handed blob to a refined, articulate illustration that conveys a meaningful message.

The principles in this book are repeated throughout. Be concise in your illustrations, uphold the integrity of the data, and avoid doing things that detract from the message. Use simple graphics to convey complex data sets, and never the other way around. Use multiple variables when possible, to show more meaning with your data.

I didn’t necessarily agree with every point this book. Tufte frequently extolls one particular graphic, a grayscale plot of galaxy distribution, which I find uninteresting for two reasons: First of all, the graphic itself does not give the viewer a deeper insight into the distribution because the shades of gray are often in dots too small and can be hard to distinguish from the darker or black dots. Second, the issue of perspective in a galactic view of the universe makes little sense in a two dimensional plane because your point of perspective at any given time and place in the universe and what you are observing is completely different than the next; distance and time distort greatly the perceived spacings and distances between galactic objects, and if Hawking’s theory of an infinite universe that loops back on itself is correct, then placing a four-dimensional galactic distribution (space+time) into a two-dimensional plane is meaningless, or at the very least not effective in this way. The filaments that appear in the graphic are no different than a puff of smoke from a firecracker – but then perhaps that is significant in itself…

Also, the writing style can be a bit presumptuous at times. Often complex terms are thrown out in the text without a definition, while minute historical details about obscure graphics appear in the side notes. If I were editing this book, I would have placed a few key definitions in there among the notations. Occasionally, ideas are thrown out there without supporting arguments and we are expected to take his word for it. I got a lot more out of the actual illustrated comparisons and the supporting explanations, which always clearly demonstrated the point.

Speaking of which, I love how the pages in this book are laid out, with the footnotes on the side, as Tufte said in his lecture ‘the way God likes them’. These notes are level visually with the illustration they describe, and it is much easier for the eye to travel left to right and associate the text with the object it describes. The book itself is beautifully done, with nice thick opaque non-glaring pages, gorgeous typography and illustrations, and a solid old-school style of sturdy book binding. It is a good example of placing your integrity behind your work by focusing on the details and the core essences of your product.

Overall, the book is gorgeous and I loved reading it – despite some of my disagreements with the details. It has made me think differently about how I go about designing these graphics in my everyday work. I am looking forward to reading the next two of his books, Envisioning Information and Visual Explanations. But first, I must get through Eric Meyer on CSS, Flash MX Training from the Source by Chrissy Rey, and Jeffrey Zeldman’s Designing with Web Standards. So much to do, so little time…