Tufte: Beautiful Evidence, a Partial Review
By pmonday on Dec 15, 2006
Its important to know what you don't know. In fact, this can be an architect's greatest strength (unless, of course, they know they don't know anything and make up for it by dressing well). I know I don't know visualization. As a result, it is a good idea to surround myself with people that know their stuff with visualization and also arm myself with good books, like Edward Tufte's "Beautiful Evidence" so that I at least have an idea of what people are talking about.
Beautiful Evidence is a book about presenting evidence in an effective way. Sounds simple enough. Well, its not. Consider how often you sit staring at graphs and information that purport to mean something, yet the "evidence" that is being presented is spread across multiple slides and backing documentation. By the end of the presentations and discussions, you get a queasy feeling that you don't have a complete picture of the evidence or that you wasted an hour for a single nugget of information.
Tufte's book takes you through several techniques for displaying evidence:
- Mapped Pictures: Images as Evidence and Explanation (relevant for displaying maps, product layouts, and sketches that must be coupled with real-world information)
- Sparklines: Intense, Simple, Word-sized Graphics (very relevant to succinctly displaying large amounts of data concisely)
- Links and Causal Arrows: Ambiguity in Action (particularly relevant to modelers)
- Words, Numbers, Images - Together (relevant for all of us who author technical documentation that weaves together images and technical data)
He then has broader sections of the book that assemble the techniques and cross them with common perceptions and mistakes with presenting evidence. These sections consist of:
- The Fundamental Principles of Analytical Design
- Corruption of Evidence Presentations: Effects Without Causes, Cherry Picking, Overreaching, Chartjunk, and the Rage to Conclude
- The Cognitive Style of PowerPoint: Pitching Out Corrupts Within
- Sculptural Pedestals: Meaning, Practice, Depedestalization
- Landscape Sculptures
To be honest, I am still working my way through the latter chapters, though I've skimmed many of the parts while I am deeply working through the book. The book is fascinating to read AND discover. Tufte integrates his writing with the visual facts and evidence that backs up his conclusions (this is an important point in the chapter on "The Fundamental Principles of Analytical Design" with excellent examples in "Corruption of Evidence Presentations"). The examples provided in the book range from some of the first printed writings, to sketches from da Vinci, to maps from cartographers, to modern day "evidence" in the 9/11 reports. Where examples are good, Tufte goes to detail in explaining WHY the example is good. When examples are bad, Tufte explains why the examples are bad. One of the many reasons the book is good is that Tufte then goes on to show how the bad presentation of evidence can be improved! So the book is very illustrative and contructive.
Often, people have the impression that Tufte is just for user interface people. This is simply not the case. Architects are responsible for delivering information to many different audiences. Often, this information is in the form of evidence for why certain decisions must be made
- Why did you choose this off the shelf product vs. another
- Why are you using certain throughput rates as your design point
- How did you come to the conclusion that a product / subsystem was beyond repair and needed to be rewritten
Unfortunately, many of our audiences ONLY want slide presentations (I'm being honest). As you work through Tufte's book, you realize ways that your presentation of evidence could be made better and more effective.
Let me give you an example that I just lived through.
My new boss is a number's guy. I went through about 4 iterations on a slide set before he started to nod that I was on the right track. My first slides were highly bulletized on some slides, with graphics on others. To be honest, I was skewered in the meeting (rightfully so). While I had worked on the graphics to integrate them with the surrounding text, the integration turned out flawed in many different ways, this became obvious because my boss, it turns out, follows one of Tufte's important points: "To maintain standards of quality, relevance, and integrity for evidence, consumers of presentations should insist that presenters be held intellectually and ethically responsible for what they show and tell".
While I had not engaged in corruption of evidence, I had included "Chartjunk" and in including that Chartjunk, I was not exacting in my choice of chartjunk (there were other problems, but this is the one that bothered me most). My causal lines were also not very informative. It turns out, my boss is an active consumer of information and the causal lines and visual evidence are very important.
I went through several iterations on the presentation. It turns out, he wanted 4 slides and those slides needed to tell the tale of the cause of where we were, and the effect it was having on my decisions and, in turn, his decisions. The slides were simple tables with compact information. The evidence had to be succinct and telling of the cause and effect.
Hopefully my presentations will get better and better from here on out, but it will take time. It is obvious that these things should not be thrown together.
I should tell you about the BEST part of the book, in my opinion. The chapter on sparklines was completely enjoyable. I had never really put a lot of thought into sparklines (since I didn't have a name for them anyway). Edward Tufte's web site has a discussion on sparklines with examples. Sparklines facilitate a high compression of data into a readable format. For example, in addition to a win / loss display of a sports team, sparklines could be used to show the entire season and give the reader the ability to see trends, additional information about intra vs. inter league play, and more. All of this information can be compressed into a very tiny, yet readable space.
What is even MORE impressive about this book, is that Edward Tufte CLEARLY articulates the limitations of our own presentation technologies in portraying information. During the discussion on Sparklines, Tufte shows how flaws in rendering the information can lead to mistakes in reading the information, ineffectual display of the information, and annoyances that can occur from the information rendering.
This is an ALL AROUND fascinating read. The burden of the presenter of evidence is high, especially if the consumers of the information live up to their requirements. On the one hand, I regret reading it...I know my presentation of evidence style is extremely flawed at this point. On the other hand, Tufte has provided me with a foundation for improvement...and I have lots of oppurtunities to practice the skill.