Review: Edward Tufte's Beautiful Evidence

A recommendation for any lover of beauty

When I picked up Edward Tufte’s Beautiful Evidence, I expected a rather dry book about making good graphs. As someone who aspires to put more graphs in my blog posts (someday!), I thought this might be helpful.

Beautiful Evidence is not like that.

The experience of reading Beautiful Evidence is most like the experience of being infodumped at by a very smart, very informed autistic person with a special interest in data visualization. Every word your autistic friend says drips with passion. As far as they are concerned, data visualization is the most important and interesting thing in the world, and if you disagree with them it is because you have not had the importance and interestingness explained to you properly. Every so often they grab a medieval star chart or nineteenth-century natural history print from their book collection—they have thousands of rare books, obviously, which they often buy instead of food—in order to illustrate some point. Their infodump doesn’t so much have a “structure” or “organization,” instead wandering off on a tangent as soon as something new occurs to them. Every so often they are reminded that PowerPoint exists and spend twenty minutes expressing the sort of vicious hatred normally directed at genocidal dictators and bands beloved of teenage girls.

I’m not diagnosing Tufte from his book, I’m just saying there’s a vibe.

In retrospect, I didn’t expect a book about data visualization to be beautiful. I expect charts and diagrams to be workmanlike, to give information clearly without getting in the way too much. It is as if I spent my entire life reading Isaac Asimov and thinking that that was what prose is, and then I read Vladimir Nabokov for the first time. Tufte showed me that diagrams could be art.

(If any billionaires read this and want to express their approval of my blog, I am accepting gifts of Edward Tufte’s fine art prints.)¹

I’m not sure that Beautiful Evidence is helpful with regards to avoiding the usual problems of data visualization—misleading charts, ugly colors, unreadable or confusing labels, etc. Tufte like most autistics is strongly opinionated about matters that aren’t as dogmatic as he makes them look: sometimes one might want to have a picture without a contextualizing measurement, sometimes you might want to put your diagrams at the back of the book for a good reason, and even the dread PowerPoint is sometimes useful. You’re not going to find practical rules here.

What you’re going to find is something more important: developing good taste.

Once you read Beautiful Evidence, it becomes obvious that most designers of data visualizations have bad taste.

Once you have read Beautiful Evidence this chart becomes physically painful. Source.

Depicted: me after reading Beautiful Evidence

In no field of art do you develop good taste by memorizing a set of rules. You don’t acquire taste in fiction by learning that you shouldn’t write in second person and should use active verbs and should have a dark moment immediately before the climax which is about ninety percent of the way through the book. You acquire taste by reading great fiction, and lots of it: old and new, from your culture and other cultures, genre and literary, ambitious and just trying to show the reader a good time. You read poetry and nonfiction and drama and myth. You read bad fiction, too; no amount of lectures about said bookism teach you as much as reading a webnovel where people are instead constantly yelping, whistling, hissing, spitting, booming, shrieking, and gurgling. And then you analyze what you read: what worked, what didn’t, how the author used their tools to achieve their goal, the structure and style and symbolism and how they all interact to create an overall experience. You talk with others about what they got out of it. You read literary critics and creators of meta and working writers; you turn over their points in their head, taking what is good and leaving what is bad.

They say “once you know the rules you can break them.” What this means is that, after a while, you develop a sense for good work, and your sense for good work is far more reliable than any list of rules anyone has conjured up. In general, good work includes active verbs and climaxes and points of view that don’t draw unnecessary attention to themselves; but none of those are necessary.

Beautiful Evidence is about developing good taste in data visualization. Tufte shows you piece after piece of good work. He doesn’t limit himself to what we think of as data visualization: he ranges across the centuries, showing you the most beautiful data visualizations of the past thousand years. He breaks them down, showing what works and what doesn’t. He has strong opinions, but that just prompts the reader to go “…no, I don’t think that’s quite right” and develop their own sense.

What makes for good taste in data visualization? Well, my own taste is still mediocre, so I will only venture a few guesses. Some of it is captured in the quotes at the beginning:

What was observed by us is the nature or matter of the Milky Way itself, which with the aid of the spyglass, may be observed so well that all the disputes that for so many generations have vexed philosophers are destroyed by visible certainty, and we are liberated from wordy arguments. --Galileo Galilei

Medical statistics will be our standard of measurement: we will weigh life for life and see where the dead lie thicker, among the workers or among the privileged. -Rudolf Virchow

If you look after truth and goodness, beauty looks after itself. —Eric Gill

The watchword of good data visualization, as I understand it from Beautiful Evidence, is elegance. The creator of a data visualization should be an ascetic in the temple of Truth; every line, every color, every flourish must justify itself by the information it conveys to the reader. This is not to say that they should be spare, but the richness comes from the richness of the data—not the desire of the creator to show off their clipart collection.

Similarly, good data visualization is simple and clear. The reader doesn’t have to peer at the image to make out what you’re saying; the main details can be understood in a glance, and it is easy to answer further questions that the reader might have. Again, it is not that beauty is not a concern, but beauty comes from the precise use of design techniques to convey information. The beauty of data visualization is the beauty of a proof or a poem: nothing superfluous included, nothing necessary left out, everything chosen precisely to convey the meaning intended.

1. I am also accepting the HPLHS’s Angell Box and Dark Adventure Radio Theater Collector’s Set. This is not relevant to anything about this review, I just want to make sure the hypothetical billionaires are fully informed.