I arrived at the hotel's conference room in San Jose well in advance, as instructed, and was handed a beautiful case containing Tufte's four self-published books. The room was immense. There were at least five hundreds of us, quietly seated, discovering these four pieces of art while classical music was floating around us. The books were so beautifully made that, at first, I didn't dare flipping the pages by fear of damaging them. My heart nearly stopped when my neighbor spilled coffee on the table inches away from my copy of Beautiful Evidence. I got up to have two of my books autographed by the master of ceremony. He was greeting everyone with the same few words, which sounded like a rhyme: "What do you do? Where do you work?" And people would tell him a few things about them. All I shyly came up with was: "I studied two chapters of your books," to which he answered: "Well, you have a lot more to read!"
Back in my seat, the course began. We listened to a piano piece by Chopin while little colored segments were displayed on a screen in rhythm, one next to the other, at a height proportional to the pitch of the notes. (There is an example here.) I could see the music even better than I could hear it: its rhythm, its melody, its intertwined patterns, and its overall "shape"... I could anticipate what I would hear next. It was breathtaking. And it introduced all the key ideas of the course: how the abundance of detail (data, not decoration), presented all at once, clarifies the design and gives the audience a multitude of points of view, enabling them to compare data at various levels, and draw conclusions.
We spent the rest of the day reading excerpts of Tufte's books and analyzing their myriad of wonderful graphics, some as old as Galileo's sunspot drawings. Each example reinforced the principles introduced with the piano piece, and added depth to them. Here are a few sentences I wrote down during the course:
- Graphics should provide two things: information (the reader must understand the story) and reasons to believe the information (the reader will assess the story's credibility).
- High-resolution designs are genuinely interactive.
- "Whatever it takes" principle. Show as many data as it takes to let the evidence speak for itself.
- Comparison is at the heart of the intellectual task of processing information.
- In a design, try to maximize the signal-to-noise ratio. Avoid boxes around titles, dark lines, and clutter (noise) in general.
- Never sort tables alphabetically. Order by performance.
- Maximize content.
- The principles of analytical design are based on the principles of analytical thinking: causality, comparison of multiple variants, credibility, etc.
- Help comparisons by presenting data adjacent in space rather than stacked in time. (For example, present all on one page rather than on two pages, front and back. Use high-resolution print-outs rather than PowerPoint slide shows.)
- All presentations should provide a super-graphic.
- Don't attract attention to the method used to present the data.
- Use good models to avoid amateur design.
Tufte's love for rare books, art, and science was really contagious. The day went by quickly despite the overwhelming amount of images and information that was presented. I went home with my four precious books, the head full of new perspectives on statistical design, knowing that, indeed, I had far more to read.