Presenting Data and Information, a Course by Edward Tufte

Mon Dec 17 12:21:00 CST 2007

Date: Dec 11, 2007 Included four books and course leaflet: * The Visual Display of Quantitative Information * Visual Explanations * Envisioning Information * Beautiful Evidence

Attendees were asked to read the following before the course began: * The Visual Display of Quantitative Information, Introduction and chapter 1. * Envisioning Information Introduction and chapter 2. * Beautiful Evidence, Introduction and pages 12-45.

I am frustrated with my notes. The course seemed to span different sections of the books and I am having to reconstruct an outline after the fact, as I did not have a take-home outline provided (which is ironic given one of my favorite take-home points is to provide something for the audience to take home). While some themes come out, they were not explicitly stated. Hmmm.

Notes

There are two long-running issues 1. Anything interesting is multivariate 2. Information resolution: bits/time and bits/area

Strong opinion for Gill Sans font for tables.

Heavily complimented iPhone and flattening of navigation while maintaining context of information.

I mentioned we were waiting out our Sprint contract before jumping and he replied this may serve us well as EDGE iPhones will be out in 2008. This has led me down a different rabbit hole of confused research on cell phone technologies.

Analytical design => analytical consumption 1. fiture out their story 2. what is their credibility 3. exactly to what, is this presentation relevant? (domain specification) 4. what should I be seeing (vs. what I AM seeing)

PowerPoint abuse an ongoing theme (even contributing to loss of space shuttle) * A size A3 page folded in half can hold from 50 to 250 PP slides. Do this instead of a PP and get it to everyone before hte meeting, and save the meeting for Q&A. Better use of everyones time and a souvenir/advertisement to take back to office. * Abusive use of Excel charts that removed cancer

  • Provide scale for numerical info, several examples of subtly (minimalistically) adding scales and annotations in an effective way.
  • Deep hierarchies for navigation are painful for the user.

Fundamental Principals

(from pp122+ Beautiful Evidence) 1. Comparisons * Show comparisons contrasts, differences * Answer the question "Compared with what?" 2. Causality, Mechanism, Structure, Explanation * annotate with explanations that cause the difference * would be good to annotate our well production logs with well events? 3. Multivariate Analysis * include more than x and y. Example shows 6 variables. 4. Integration of Evidence * Annotate in place. (Does this mean getting rid of the legend? maybe) * avoid building a table of definitions (letters connected to chart via lines, then you have to go look up what that letter means) * Integrate words, numbers, images and diagrams in place instead of off to the side * I think this makes it noisier, will be hard to do in a pleasant way, especially programatically. 5. Documentation * Credibility of evidence, source of data.
* Who did it, and who they did it for (authors and sponsors) * why? (authors' agenda) * scale of measurement 6. Content Counts Most of All * quality relevance and integrity of content * me: some work notes * '''What are the content-reasoning tasks that this display is supposed to help with?''' (p136 BE) * me: direct ramifications for project at work.

Conclusion

I enjoyed the talk, but I didn't come away with an easily-quantified new skill but more of an awareness. Talking with Tufte before hand, he did say he intended the warm-up to convey the introductory information so that the majority of the talk was a dialog. Dialogs tend to have a serpentine and chaoticly iterative structure.

There are several other reviews on the net. * http://www.stcwvc.org/galley/0201/c1.htm * http://www.mousetrap.net/~mouse/uta/PhD/tufte.html