Course: Presenting Data and Information, a one day course Instructor: Edward Tufte
Location: San Francisco
Edward Tufte has long been considered the godfather of data visualization. And if his one day class in San Francisco this week is any indication, he knows it. When we checked in we were handed a box full of Tufte’s books and an agenda that assigned us reading while we waited for him to start.
Then when the clock struck 10am, the lights were turned off dramatically, like kindergarten teachers do to get the attention of their wards, and two large screens came alive with a chart. It was undulating bars, lighting and beating colors to the chords of Chopin’s music. It was, Tufte said, “a project management chart for Chopin’s music.”
The crowd laughed and I settled in. Part lecture, part sermon, Tufte espouses certain principles of design and data morality.
“It’s not how quickly we get them but how much we learn from them,” Tufte said, referring to charts, illustrations or other ways of displaying data. He warned us not to specify the method or the data set but instead to do “whatever it takes,” meaning use a diversity of types of data and methods. I began to silently refer to him as the Malcolm X of data visualization – By Any Means Necessary.
Tufte is militant but in a 1950s Karl Malden sort of way although less cuddly and more sarcastic. Sometimes sneeringly so. Something I realized as he walked us through his books.
A few takeaways:
“An arrow is a verb – say what it is! No more generic linking lines!”
“Your metaphor is the map.”
“You almost never need boxes – they’re space hogs; when you put boxes around everything you just raise the noise level. When you emphasize everything you emphasize nothing.”
“Everything you do should provide reasons to believe!”
“There’s no such thing as informational overload, just bad design.”
“Clutter is a failure of design. Don’t blame confusion on the audience. Instead of throwing out information, fix the design.”
“Never do lowest common denominator design. Your role is to make people smarter.”
“Powerpoint victimizes statistical data.”
Interestingly, he was very light on design for the web. He seems to be partial to Apple products and incredibly critical of Microsoft. Powerpoint felt like a dirty word, he spit it out with disdain so much. He even advised us to ditch Powerpoint and hand out Word documents (he says they are equivalent to 50 to 250 slides in PowerPoint), leading me to believe he hasn’t been inside a company in a very long time.
He did offer this web advice, “no matter how beautiful your UI, there should be less of it.”
We learned about sparklines, data-intense, design simple word-sized graphics. He praised sports score tables and weather charts. And told us Wavefields would be the future – where every pixel comes with information – like an HD movie in terms of evolution. Other tidbits I picked up included that the Mexican peso character is the largest in the world. Who knew?
His very basic advice for presentations was this 1. 2/3rds of your report will be performance data – use sports or weather tables as your guide 2. You need a super graphic that will help relate your performance data to personal locations 3. Your intellectual model should be the WSJ, New York Times or Nature 4. Be sure to say something about your credibility
His other design principles include 1. Show comparisons, contrasts, differences 2. Show causality, mechanism, structure, explanation 3. Show multivariate data 4. Demonstrate credibility 5. Content counts most of all
So how to accomplish all these feats of beauty in data design and at the same time adhere to his principles? He advises, find good examples and copy them. Ala T.S. Elliot, “talent imitates, genius steals.” He views design as a research project problem and pointed us to Google images.
He cited Galileo as one of the best data visualizers – simple. Unfortunately, Tufte himself was not. And that was perhaps the irony of the day – whatever he thinks about Powerpoint it usually forces the presenter to get to the point. Tufte took 5 hours to make his – but at least his charts looked good.