Tufte lecture

Originally published to my Tumblr on 2011-06-11

This past week, Nike was kind enough to send me to Edward Tufte’s lecture on Presenting Data and Information. Though the lecture was catered to the larger crowd of powerpoint power users from accounting departments, Tufte’s principles towards presenting information have universal application, especially to those of us interested in the ever growing field of dataviz and infographics. Tufte’s lecture was the epitome of his definition of great information presentation; dry, straight forward, absent of superfluities, and extremely informative.

Below, I’ve attempted to type up my notes in a way that makes sense. And where applicable & available, I’ve included the reference that accompanied it.

https://vimeo.com/204519

We started by examining Stephen Malinowski’s Music Animation Machine. The beauty of this animation, besides the fact it takes something auditory and makes it visual, is that it allows us to simultaneously examine the past, present, and future, all at once. Rarely is this achievable. 

Tufte argues that a good diagram is not meant to be understood instantly. Rather, it should be studied at length to gain new knowledge.

Too often, diagrams rely solely on one type of data or stay on one level of analysis. No new knowledge is ever gleaned from this level of examination.

Next, we examined a flow chart of the SARS outbreak across Asia (Beautiful Evidence, pg 78). Tufte pointed out several things that make this chart effective. They were

Causality - The primary purpose of this diagram is helping the audience gain insight about the cause and origin of the outbreak. Epidemiology charts must be clear in this to be effective. Notice the dotted lines where the scientists weren’t sure. A simple visual tool to differentiate between a certainty, and a hypothesis.

Annotation - Always annotate linking lines. Nouns are linked by lines(verbs). The verbs should always be labeled and never implied.

Efficiency - There are no superfluities in this diagram. Boxes around every element would clutter the diagram, here they’re used sparingly. Typography is clean. The baseline comparison for every diagram should be a map. A map is not only a metaphor, but the standard.

Credibility - Every presentation has two elements. The story you want to tell, and why people can believe you. Consumers of your information want to see evidence, not evidence selection.

Random tangents from this section - The underlying intellectual thought in examining all data is to compare it to something. We as presenters can help shape this comparison. 

Pre specifications of datasets and visual executions, ie method and approach, blind us to seeing the correct story and answer in information presentation. Approach every problem wanting to solve the problem, not the execution.

The next graphic was Steve Chapple and Reebee Garofalo’s Rock ‘N’ Roll is Here to Pay (Visual Explanations, pg 90)

This is what Tufte refers to as a ‘Super Graphic.’ A high resolution design that are inherently and genuinely interactive. People are permitted to explore with their own narrative. Its impossible to put these into a Power Point. Rather, they must exist as paper handouts, or on tablet devices. Another example of a super graphic is aerial photographs, which allow people to choose freely what they look at, and tell their own story. I think Nike+ does a great job of creating supergraphics of running maps, so too does Google Maps with their endless exploration of satellite images.

We then examined another ‘super graphic’, various maps of cancer rates throughout the US. Here, this map of stomach cancer rates takes a previously overwhelming amount of information and makes it approachable by assigning a second level of data to it - location. The consumer of this information can begin to make inferences and deductions.

For example, two seemingly unrelated places, New Mexico and Minnesota, both have extremely high rates of stomach cancer. These regions both have high native american populations. Native Americans often preserve large amounts of food by smoking it. We can now deduce that smoked foods lead to stomach cancer.

There’s no such thing as information overload, just poor design of that info”

Apparently Tufte walked into MLB offices one day, uninvited, and offered them a critique of their site. I love this guy. 

His criticism of the site was that over 25% of the site is wasted on lines, boxes, and borders that “fortify” pieces of content. Internal politics of different content departments had lead to this turf war on the homepage. Users don’t care about the politics behind it, they just know this is a terrible site and will grimace at the thought of returning to it.

We then dove into Tufte’s forte, sparklines (I didn’t realize until this lecture that Tufte actually invented sparklines…I just thought he perfected using them).

The purpose of sparklines is to offer high resolution graphics in the place of typography. 

Sparklines offer quick comparisons that an examination of typographic numbers won’t offer.

The above graphic condenses 4,856 win/loss outcomes in a single MLB season. There are several stories contained within, all very digestible. American League West was tight until the end with 3 teams, while Seattle never stood a chance. The AL East saw a dramatic surge by the Red Sox to challenge the Yankees in the AL East. The Cardinals never really faced competition.

Tufte believes that sports pages are the best examples for an “annual report” format. Each day, people can turn to the sports page and understand exactly what happened throughout the country just by reading tables (box scores).

Finally, Tufte left us with his six principles of analytical design. To illustrate these points, he used what he believes to be the best supergraphic ever made, Minard’s illustration of the Napoleon’s Russian campaign.

1) Show comparisons, contrasts, and differences. 

2) Show causality, mechanism, explanation, and systematic structure

3) Show multivariate data, ie more than 2 variables

4) Completely integrate words, numbers, images, and diagrams. AKA Nick Felton’s golden rule :: No keys, no legends

5) Throughly describe the evidence. Provide a detailed title, indicate authors, document data sources, show complete measurement scales, etc. 

All of which lead to

6) Analytical presentations ultimately stand or fall depending on the quality, relevance, and integrity of their content. 

And, perhaps taking a queue from the late Billy Mays, Tufte through in a 7th principle absolutely FREE.

7) Data should be presented adjacent in space and time (supergraphics), not “stacked” such as pages in a book. It makes it impossible to make good comparisons.

I’d love to tell you that Tufte left us on this high note, but he then devoted way too much time to discussing interface design, which was greatly outdated (almost every complaint he had about the iPhone was from ios 1 and has since been fixed). I’ll leave you like I wish he had left us, with the principles of great data viz.