This was one of my favorite panels of the conference. If my rambling pseudo transcripts are too much, click some of the cool links. We listened to the leaders in data visualization talking about the craft. They all start by first referencing a true pioneer and guru in this field, Edward Tufte, who said “Minimize interface design to get to the content.”
Martin Wattenberg, Head of “big picture” Data, visualization group, Google. Blog
Joe Ward, Sports Graphics Editor, NYTimes. He’s responsible for some gems like this one
Ben Fry, Principal, fathom.info. If there’s a top ten for “coolest visualizations ever,” I bet this ste gets a few spots. This one is awesome.
Moderated by Sam Hinkie, Executive VP of Basketball operations, Houston Rockets.
Q: OK, visualizations are cute, even beautiful. But where’s the usability to make decisions?
Martin: One great example is at babynamewizard.com . People will spend time on this site for hours, deciding on the right name. But there are thousands of time series embedded in this.
Q: What are some of the most common mistakes you see with visualizations?
Joe: Overuse of color: As little as possible is better. Typography: less is more.
Martin: We discovered a good rule at Google: when you find a horrible error you didn’t know existed in your data, then your visualization is valuable. You discover things that you never anticipated. Ben Fry’s Windmap shows many unanticipated patterns we never expected. Also high-end dashboards have a sense of uncertainty about our data. Very seldom do we see beautiful demonstrations of uncertainty. I think great visuals show the uncertainty.
Martin: Some are happy with pure numbers. Very rarely does visualization lead to a very different decision, but reaching those decisions often happens much much faster. The speed of its effectiveness is its value. As long as they can get back to the data, they trust it.
Q: We’ve pulled the vizualizations from a few of the research papers being presented at the Sloan Conference. What do the experts think?
Tennis Research Project:
Joe: Lots of noise introduced beyond your signal. I would delete the surrounding courts, people, any distraction.
Martin: This is an excellent first step…. Now let’s ask: what outcomes do I care about? What shots are most important? Can I start making comparisons directly.
Joe: More of the court to show you’re using the service boxes. Get rid of grass stripes. Edit. Take out as much as possible. The data are your story.
Martin: A sense of importance is clear. There’s no sense of uncertainty. ..I’d like to see that. I’d love to see many versions of these to compare.
They show a visual of data Larry Sanders near the basket:
Joe: This is really good. Clean, gets to the point. Clear comparison of Sanders/ Lee. We did this with teams, and added Performance relative to NBA averages.
In comparing Steve Nash / Dwight Howard:
Joe: drop shadows on dots changing the color of the dots, and changing the color, which has meaning. I’d get rid of that.
Q: How does a team think about augmenting their staff with resources with data visualization?
Martin: if you get number guys, they will organically go there. When you show things to fans, then graphic designers needed. First find people who love numbers, and that’s 75% of the way there.
Joe: If you have a design problem to solve, somebody has probably solved it.Example: Cholera Map. They did this 1850. Last Year, Washington Post created a similar map gun store and homicides.
Ben: Anybody can pick up a road atlas and find their way. And online is not quite there yet. The cartographers are ahead of us still.
Martin: You should make sure your information tech folks are happy to work with out a final goal; it’s an iterative process on constant tweaking.