Edward R. Tufte course in Boston - lecture notes and followup thoughts
On March 31st I attended a day-long course by [Edward Tufte](http://www.edwardtufte.com/tufte/) at the Copley Fairmont Hotel in Boston, as [mentioned](http://trainque.com/index.php/2005/03/11/attending-a-one-day-course-by-edward-tufte/) a few weeks back. Here I’ll jot down some followup thoughts.
I found the lecture rather interesting for a number of reasons. First, having been out of college for nearly 4 years, I found it refreshing to hear an academic expert elucidate on a topic of personal interest. I guess you could say I’ve been stuck in a business environment where presentations are often limited by time and so you don’t get a lot of tangents or thought-provoking asides. Also, in a business presentation you see a lot of bulleted items riddled with biz buzz-words, and so poetic eloquence is often set aside. Makes me want to go back to school a bit.
**Thoughts/Criticism**
Generally I agreed strongly with many of his proclamations, especially with respect to cherry-picked data (e.g. data chosen to support ones agenda, regardless of accuracy) and the need for content over presentation. I also found his remarks on resolution, user interface, and graphical data representation rather fascinating, since I already have an interest in this area.
However, I feel like there were some points that made more sense in an academic environment than in a business environment.
For example, take Tufte’s apparent abhorance of Powerpoint. He states “presentations should be a content experience, not a presentation experience,�? and while I agree wholeheartedly with this belief, in a sense I feel like he is preaching to the choir. While I would love to see standard company-branded, template-driven Powerpoint presentations replaced, I still feel that those attending a presentation have a certain level of expectations for what constitutes a “proper presentation.” Choosing to remove company branding in internal presentation in favor of allow content to speak for itself comes with an added risk, namely that viewers will notice the absence of branding and consider it an error on the presenter’s part. In other words, although following Tufte’s recommendations for the presentation of information allows individuals to more effectively communicate, there are downsides to taking an academic approach in a business environment. The two don’t always play well together, and I would have liked so hear Tufte address those issues alongside his preaching for better communication methods.
On the same note, I feel like Tufte could have addressed some of the reasoning for cherry-picking data. In a business environment, there are some good motivations for selecting data that supports one’s stance, or the work they’ve done. For example, someone responsible for enacting a plan that had a desired and predictable end result might selectively choose data that supports their actions or position. Rarely do you see someone present data that plainly states that their plan was unsuccessful or inconclusive, even though that may very well be the case. One might attribute that to the self-serving nature of humans, or something along those lines, but that’s really beside the point. Tufte’s ideas work when you make the blanket assumption that all individuals value the inherent goodness of The Truth, especially with respect to data. Sadly, the real world doesn’t operate by such standards, and so while adoption of Tufte’s advice makes very good sense in a scientific or academic environment, I suspect it will be a long time coming before business adopts his approaches, namely because it requires people to change so significantly. That is to say, in order for Tufte’s ideas to really catch on in a business environment, people will need to abandon that self-serving part of their behavior, become considerably more objective, and most importantly, begin to truly value and make the distinction of accurate data over data which has been “spun” by the presenter.
That said, I would have liked to have heard his take on how this will happen, but perhaps this is out of the scope of his expertise. Bottom line is that business isn’t science, and although the scientific method could (and should) be used in a business environment (especially when working with data analytics), the reality is that the rules of the game differ significantly enough for me to stay skeptical.
But don’t get me wrong: Most of Tufte’s ideas are great and right on mark, and they will have tremendous signifance over time. It is definitely worth attending one of his lectures if you get a chance (and becoming one of the converted).
**Notes**
For anyone that missed it, I’ve included my notes below, which include some key points from the lecture:
Key themes introduced during intro:
1. escaping flatland - inherent challenges in representing 3 or more dimensions of data in a 2d medium (paper/screen)
2. the effect of “resolution” (and the limits thereof) in visual communication
maximize content reasoning time & get viewer spending less time decoding puzzle. Use inline labeling, not “Q = variable A”-type legends.
Analytical Design Principles:
1. visual design should always answer a question: “compared to what?” - make comparisons
2. show causality & make explanations for data
3. show multivariate information
4. integrate words, numbers, images
5. document everything & tell people about it. Cite info sources. (Increases credibility).
6. presentations will be weighed on the quality/relevance/integrity of content
7. try to show info adjacent in space, rather than stacked in time
8. use “small multiples” (which are inherently comparative and deflect suspicions of cherry-picked data). Small multiples easily show your work. (re: Galileo’s sunspots)
Tufte showed off first printings of Euclid’s The Elements of Geometry, Galileo’s The Starry Messenger, and Newton’s Optics
“Good design is clear thinking made visible, bad design is stupidity made visible”
reduction of ornament in favor of making the content shine - content is king
avoid “chart junk,” and be skeptical of those who use it (examples “fancy” Microsoft 3d-style chart templates with only 2
dimensions of data, ornamentation that clouds the data, cleverness for the sake of being clever.)
Feynman: “Fundamental laws of nature are causal,” hence the universality (i.e. cross-cultural understanding) of good analytical design
Put your name on things - it shows that you care about the content and take responsibility for its validity
It is the responsibility of the presenter to preemptively address suspicions of cherry-picked data. People often decide what they want a chart to say, and then choose data to support their indiv. agenda. Let the data speak for itself. It is better to have inconclusive or generally unimpressive accurate data than it is to have impressive false data. Even worse to make a decision based on the latter. (re: NASA mktg dept. & CT traffic deaths chart).
“It’s better to be approximately right than exactly wrong”
get a Macintosh, and be happy (not thinking about OS & working context, focus is on content/tasks).
as viewer, ask yourself if relevant information is excluded from graphs/charts, especially if you suspect a causal relationship.
The resolution of good old paper is higher than the most advanced computer monitors
vertical scale of charts/graphs: ideal aspect ratio of graphs/charts is one where the absolute value of the average slope of all the lines is one (i.e. 45 degree & -45 degree slopes). The human mind perceives “lumpy” charts as more revealing more than “spiky” charts showing the same data.
Feynman: “Nature cannot be fooled” (see discussion in Visual Explanations, p. 53)
If a chart or table or object needs a label, label it inline - don’t use legends/keys that require “back-and-forths”
by studying the field of magic, it is possible to learn what NOT to do when presenting info (examples: magicians never id
problem, relevance, and solution up front. Magicians never repeat, because that assists in understanding (e.g.
Particular-General-Particular approach particularly useful)).
lessening detail doesn’t make graphs easier to read.
Don’t use footnotes, use sidenotes - they’ll be closer to the content you’re referencing
simple underlying designs result in successful communication of ideas (ask yourself “what is the thinking task?” and the design will manifest itself, instead of coming up with a design before the thought process behind it has been worked out). The way you reduce clutter is to clarify the design and then add information.
including a zero point in a chart is not necessary unless it tells you something. Better approach is to include contextual data along the horizontal (from discussion on CT traffic deaths).
presentations: annotate everything, esp. financial data. (Is inflation considered, etc.)
When determining how to present financial data, one need only turn to well-known newspapers like NYTimes, who’ve “figured it out long ago.” Newspaper sports stats tables also effective, as readers are familiar with the format already. Use familiar approaches to your advantage.
The power of the Smallest Effective Difference - make all visual distinctions as subtle as possible, but still clear and effective
Sparklines - Tufte’s text-sized inline charts as easy ways to represent hundreds of numbers more effectively. New idea, not yet caught on. (spent some time hyping this “revolution”)
Guidelines for Presenting:
1. show up early
2. tell problem, who cares, and solution. Include an abstract on the paper copy, and use full sentences (not bullet points).
3. never apologize before presenting
4. give everyone technical style handouts (11×17 folded, double-sided) (longevity of paper medium, paper supports casual skimming
during presentation)
5. think about your audience in terms of what they read. Use high-resolution tactics whenever possible.
6. MS Powerpoint should only be used for fullscreen projections & images. It should never have company branding or bullet points.
“Presentations should be a content experience, not a presentation experience.” If not doing overhead projections, use Word.
7. All audiences deserve respect; show it to them. The “know your audience” approach more often results in “talking down” or
underestimating their intelligence. Avoid “keep it simple, stupid” approaches, they are replacements for meaningful thoughts/ideas.
8. be appropriately demonstrative, show your passion for the subject if you believe in it. Body gestures important.
9. Always finish your presentations early.
(Note that a one or two of these points were augmented by Signal vs. Noise’s notes. Apparently, Tufte’s routine is fairly consistent, as only a few of SvN’s notes weren’t mentioned in the Boston lecture).
2 Comments
RSS feed for comments on this post.
[…] If you’re interested in more of the meat of the presentation, there’s a whole lot of it here and here. […]
Tufte’s Seminar of Data Display, PowerPoint and Cognition
I saw Edwards Tufte’s seminar yesterday. I am not going to summarize his more well known ideas since his seminars have been well documented elsewhere. You can find one here and another here. Tufte’s work and thoughts on information display