some notes from edward tufte

“A good diagram isn’t necessarily meant to be taken in at a glance. We should read a good diagram as seriously as we might read two or three paragraphs or even a couple of thousand words of text.”

“It starts with Patient One, and that gives insiders some important information; because of course these charts usually start with Patient Zero. What happened was, the Chinese government suppressed information, so Patient Zero can’t be traced. So the chart pointedly starts with Patient One in Hong Kong.

“An important principle is not just to concentrate on descriptions but also their relationships. Show causality.

“We have evidence at the molecular level, the clinical level, the human level, the patient level, because of the detective work. Finally there’s some public health information. The evidence exists at multiple levels. And that’s one of the important principles of evidence: whatever it takes.

“Most people show material that they’re good at, that happens to be convenient, that happens to be inexpensive… but the evidence should be at multiple levels. From molecules up to the nation state. Whatever it takes.

“Notice that the linking lines are annotated. Look at an organization chart with a few names at the top and an increasing number in the mighty pyramid. The nouns are quite important, but the linking lines – I can’t believe it! – are all the same. Think what that says: that every pairwise relationship in this organization is exactly the same as every other. That simply cannot be the case.

“So how can we give those linking lines some texture? Put some words on them. Put some numbers on them. We have intense specificity in the nouns. Let’s add some specificity to the linking lines. Look at Patient 2C. That’s probably how the virus got to Vietnam. That’s the mechanism. Or look at 2E. Travel to Singapore. Or Patient 2D. Travel to Canada. Instead of just information that tries to tell us the bad news about the epidemic, this shows the flow of causation. So tha we can try to intervene.

“This texture of annotation on the linking lines and on the nouns helps the credibility of this diagram a great deal. It shows off the hard work of the investigators. This stuff is not overload or clutter. In general, there is no such thing as information overload. There is only bad design. Don’t blame the victim. Don’t blame the audience for being stupid. Blame the design.

“We don’t go around a city complaining about information overload, and that’s because we have a very powerful perceptual system. We perceive in 16-bit colour, three dimensions, all day every day. That’s massive bitflow. One of the things we are trying to do with information design, here, is to bring it up to the routine capacity of the human perceptual system!”

Twenty minutes in and I am IN LOVE.

“A map is a wonderful metaphor. Cartography has done it mostly right. Hold up a paper map next to a computer screen, see how you’re doing compared to a map. How come my diagram’s not operating at the same perceptual level as a map?

“If on the other hand your display looks like a knockoff from a corporate chart or a powerpoint pitch… START OVER.

“Everything you do should provide reasons to believe. The audience is trying to figure out what your story is and can they believe you? What are the reasons to believe the graphic on page 78? We’ve already heard some. The texture of detail. The annotation of many patients. There’s a coherent story. Diverse data. They’re talking about causality. Who’s endorsed this material? It was published in the Lancet, one of the top five or ten medical journals in the world.

“In real science, every statement about evidence comes with that parenthetical remark at the end… ‘At least until better evidence comes along.’ Now this doesn’t mean that if we’re giving a report on astronomy, we’re leaving the door open for astrology. The alternative explanation has to play in the same league as the original explanation.”

He looks a bit like Bill Murray.

“We want an open mind but not an empty head. We don

Leave a Reply

Comments are closed.