TL;DR Read Tufte’s books and follow his advice and attend one of his courses if you have the opportunity.

Who is Edward Tufte and what does he have to do with CFD?

Last week two co-workers and I attended Tufte’s 1-day course on Presenting Data and Information. This is the third time I’ve attended one of Tufte’s courses, each time with a different cohort of colleagues.

Who is Tufte (pronounced TUFF-tee)? He is Emeritus Professor, Yale University where he taught for 22 years (political science, statistics, computer science). He is an expert on information visualization and publisher and author of four seminal books on the topic (with a fifth in work). He is a sculptor and print maker. He is also a presidential appointee to the Recovery Independent Advisory Panel, Recovery Accountability and Transparency Board, 2010-2012.

Why have I gone to hear him speak three times? I think his ideas on visualizing information, especially scientific information, are impactful, actionable, and just plain good.

As for what he has to do with CFD, it’s a bit more indirect. In my opinion, if more of us in CFD followed his advice, presentations would be better and audiences would be happier.

Tufte’s 7 Principles of Information Design

Let’s begin with his seven principles of information design.

1. Show comparisons, contrasts, and differences. Answer the question “Compared to what?”
2. Show causality, mechanism, structure, and explanation.
3. Show multivariate data. The world is complex so illustrate that.
4. Completely integrate words, numbers, and pictures. Be mode indifferent.
5. Provide documentation to establish the credibility of your data (and you).
6. Above all, protect the quality, relevance, and integrity of your content.
7. Allow important visual analysis to be done adjacent in space, not stacked in time.

How Do You Implement His Principles?

So those seven things sound good, but what do they mean? And how do you implement them?

This is where it gets interesting. I’m a huge fan of practical learning. Sure, sometimes it’s great to learn for the sake of learning. But even Tufte advised us to shed the learning style we all adopted in college and “read to hack the material” and “plunder ideas you can use.” I like that approach a lot.

So since his course is presented as a sometimes rambling verbal walk through his four books and life experiences, I’ll just share the things I noted.

• “The information is the interface.” Tufte is very much against extraneous user interface (UI) elements (scroll bars, buttons, pop-ups).
• On a related note, Tufte calls the touch screen “the friend of data” because all UI elements are banished from the screen.
• Again on the topic of UI, maximize the time the consumer spends on content reasoning as opposed to format deciphering. For example, Tufte thinks with five Excel templates you can make all the graphics you’ll need. Each graphic need not have a unique display.
• “Put your name on your work.” Take responsibility.
• On a related note, quote authorities to support your position. Don’t state it yourself.
• “Design is a research problem, not a creative act.” This is why I think engineers should respond well to Tufte’s concepts. He’s not about aesthetics; he’s about ensuring the audience understands your data.
• When you have “little data” present it as text and numbers in a tabular format, not in a paragraph. Tufte is a big fan of the richness and conciseness of how ESPN.com displays data, especially box scores.
• Never alphabetize data. There has to be a more meaningful and illustrative sorting criterion.
• “Story telling is fashionable.” The use of story telling as a presentation device is indeed widely recommended. But Tufte wants us to “get to the point” and let the data tell the story.

How About Running a Meeting and Making a Presentation?

Because meetings and presentations go hand-in-hand, Tufte offers several suggestions for running a meeting.

1. Every meeting begins with a document. Prepare this document in advance and start the meeting by giving your audience 15 minutes or so to read it. They’ll all immediately seek out and find the information of most importance to them. Then they’ll link it to the rest of your content. The principle here is that people consume information 2.5 times faster by reading than by taking.
2. As the presenter, you no longer have to read through every point of the presentation. Verbally annotate the document. Verbally prioritize the document and identify what’s important or what requires action.
3. Have a rigorous Q&A session.
4. Finish in 24 minutes. In other words, this format should allow you to finish meetings early (which everyone likes), giving your audience time to get to their next appointment.

You know, I would be a happy man if I never again had to sit through a CFD conference presentation that looked like this one. (Whomever is telling you to give presentations like this is doing you a disservice. This is fine for your paper but your presentation is not your paper.)

• Logo slide
• Outline (this slide)
• Introduction
• Motivation
• Deriving the Navier-Stokes Equations (8 slides, one of which is the new stuff)
• Results (time allotted for presentation runs out here after only 2 slides – jump to last slide about funding agency because money)
• Conclusions (repeats outline)
• Future Work (i.e. vaporware)
• Funding Agency Acknowledgment

Back to Tufte.

When it comes to the actual presentation, Tufte says to focus on sharing your content and establishing your credibility. The latter is important as he cites a study saying that the average amount of “self cheating” is 8%. (This must fall into the category of little white lies.)

But regarding your credibility: avoid cherry picking data, share your data sources, resist the urge to draw conclusions, and avoid incompetence. The latter he says is a much larger problem than conspiracies. In other words, never ascribe to malice that which can be equally explained by ignorance.

Tufte’s Books

I love Tufte’s books. If nothing else, his courses are motivation to delve deeply into his books (that are included in the course fee) to explore and discover his work. They’re beautifully and exquisitely published. (Caveat: I have the older hardbound versions and now he seems to be mostly selling the paperbacks.)

Beautiful Evidence, 2006. Tufte loves the graphic shown on these pages illustrating Napolean’s march to and from Moscow.

Visual Explanations, 1997. This page illustrates that rainbow color maps are vastly inferior to 1- or 2-color maps.

Envisioning Information, 1990. Love seeing a Mondrian tossed in to this illustration of geometric principles.

The Visual Display of Quantitative Information, 1983. The granddaddy of them all. Note the CFD graphic with velocity vectors in the upper right.

Tufte’s Visualization Techniques

Tufte has invented many interesting visualization techniques. I’ll highlight three of them here.

Sparklines

A sparkline is a word-sized graphic that can be easily embedded in text. Perfect for being mode indifferent.

Examples of sparklines. Image from edwardtufte.com.

Small Multiples

The technique of small multiples involves an array of graphics sequenced by a variable in the data.

Example of the small multiples technique. Image from Juice Analytics.

Image Quilts

Image quilts are a larger variant of a small multiple and show maybe 100 images or so. I share the one below as an example because I am a fan of Agnes Martin’s paintings.

Example of an image quilt. Image from imagequilts.com.

What Kinds of Things Does Tufte Hate?

Tufte’s mortal enemy is PowerPoint. He says to use it only as a projection device. According to him, “PowerPoint is about convenience for the presenter, not the needs of the a data or audience.”

In my opinion, the true focus of his ire are PowerPoint’s bullet-oriented, built-in templates that force us to dumb-down content. The result, according to him, are presentations that are a laborious march, bullet by bullet, slide by slide. He refers to this as “stacking nouns in time” and says that it’s a lousy way to present information.

He advocates “arraying verbs in space.” The implication here is that PowerPoint’s information density (in the bullet-point approach) is incredibly low per slide (6 facts?). Whereas, an information-rich visualization on a single page produced on a high-res display or his favorite P.A.P.E.R. technology can include 1,000 times more information.

The other benefit of presenting an array of verbs is that it allows each audience member to easily and immediately find what’s uniquely important to them. With the stack of nouns, you have to hope that the audience is still interested and awake by the time you get to the slide and bullet that has the nugget of information they seek.

His bile isn’t reserved for PowerPoint; poorly generated graphics (usually from Excel) take a hit too. Please note that he’s actually warmed up to Excel as a tool and thinks it’s OK. His advice is more about getting us users to avoid temptation. His term for it is chart junk: bar charts with 3D effects, grotesque cross hatching, excessive axis lines, garish colors, pie charts, table cell borders, etc. You could probably have a day-long clinic on what not to do in an Excel chart.  But for a more rigorous and academic study of some of these issues Tufte recommends (and I do too) the book Interaction of Color by Josef Albers (now available as an iPad app). Albers is not only a color  theorist but an abstract painter.

Frankly, I’m glad to see Tufte warm up to Excel and wish he’d do the same for PowerPoint. I understand his points. But I believe in the adage “Only a poor musician blames his instrument.” As an illustration of this pithy saying, consider the classic jazz album Live at Massey Hall by The Quintet: Dizzy Gillespie, Charlie Parker, Bud Powell, Charles Mingus, and Max Roach – a supergroup by any definition of the term. Arriving late at the venue, Charlie Parker had to borrow a saxophone, and a plastic one at that. Despite that, Parker and the band performed superbly.

Bonus Info

One of the high points of Tufte’s courses for me is when he shares two books from his personal collection to illustrate certain ideas about information design. One book is a first edition by Euclid from the 1500’s and the other is a first edition Galileo from the 1600’s. You don’t get to see things like that too often, carried around the room reverently by his white-gloved assistants. Sorry, no photos.

Closing Thoughts

This is a case of “do what I say not what I do.” It takes time to rigorously adhere to Tufte’s principles. Even I like putting drop shadows on images. You have to throw out a lot of stuff and start over. But I advocate that we all try. And if you see me doing something that violates one his principles, don’t hesitate to call me out.

Resources

• Tufte’s website: www.edwardtufte.com
• Tufte on Twitter: @edwardtufte
• Image Quilts website: imagequilts.com