Yesterday I attended Ed Tufte’s one day course on Presenting Data and Information. His book, The Visual Display of Quantitative Information, completely changed the way I think about data. If you love his book, I highly recommend his course. He doesn’t cover a lot of new ground, but he puts emphasis on a few things I didn’t pick up on before.
Key points:
- Show causality
- Don’t pre-specify the medium of the presentation. Use whatever it takes to show causality.
- Annotate linking lines.
- Be inspired by maps.
- Web design is too influenced by internal hierarchies and ends up being a turf war. Make the interface flat and filled with content.
- Your presentations should strive to be as data dense as the sports page.
- No zebra stripes in tables.
- Most interesting data is multivariate. Supergraphics like Minard’s Napoleon’s March show 6 or more variables.
- Progress in most fields is measured by information resolution and throughput. Why are our power point slides limited to 4 or 5 bullets?
- Put important analysis and comparison in a common eyespan (no flipping or scrolling)
- Be wary of focus groups. Good design is not a democracy.
- Start every software project with the interface.
- Make the data the interface.
- Instead of trickling in data during a presentation, dump a ton of data in their lap, have them read it, and have them cross examine you.
I got to talk to ET himself for a few minutes before the course started about his work on the stimulus bill. I mentioned some work I’ve been doing on making sparklines in HTML 5 and he said to make sure I paid attention to the length and width proportions. I got to meet a lot of interesting people and even convinced PMMI to send Jorge and Paula. So glad I went!