Narrow-minded Data Visualization
I was going to let this one slide, but people kept commenting, essentially trashing FlowingData, and that's just not cool. As you might recall, I put in my picks for the best data visualization projects of 2008 a while back. They were the fine work of statisticians, designers, and computer scientists, all of them beautiful, and all of them built to tell an interesting story with the dataset at hand. None of them were traditional graphs or charts.
In a post titled Better late than never earlier this week, a friend of Andrew Gelman's responded to my picks: "Does this stuff suck? Or am I missing something?"
Andrew replied, "Yes, I agree. They all suck (for the purpose of data display)."
I didn't think all that much of it. I knew what Andrew meant (after I got over my initial shock). My project picks don't work as analysis tools, and that's true, because they were made for presentation more than anything else. The comments that followed, however, were what got me going. They serve as yet another reason for me to believe that statisticians (and a lot of the analytically-minded), in general, are clueless about (or unjustifiably against) visualization outside their own field of expertise. It's this sort of narrow-mindedness that has kept statistical visualization looking pretty much the same for the past few decades. Sure, add some interaction here, change a color there, but take that away and you've got the same stuff John Tukey was writing about in Exploratory Data Analysis (1970). Don't get me wrong. Tukey is great. This was before computers though. I just got done reading about how to draw my graphs with a pencil and to use a pen for extra emphasis.
Data has changed since then my fellow statisticians. We don't draw our graphs with pencils and use a pen to make things bold anymore. We have computers, and believe it or not there is software and programming languages that work better than R in some situations. Again, don't get me wrong. R is excellent. I'm just saying you don't have to do everything with it. Like data parsing. Python is pretty good at that last time I checked.
Look at the cool stuff coming out of the MIT Media Lab or HCIL at Maryland. We should be doing that type of stuff. Experimenting. Playing. Trying new things. Computer scientists and designers are doing it every day. Some of the stuff might be a bit clunky or useless, yes, but at least they're thinking about (and implementing) new ways to explore. That's how we learn. Why not statisticians? But no. It's all about bar graphs, scatterplots, time series charts and ways to combine methods we already know. Everything else is labeled chart junk, which seems to have developed in to the visualization equivalent of meh and epic fail.
Maybe this is because of a lack of technological know-how. After all, statistical computing is still a relatively new idea, but still, open up your mind to the possibility of something new. Traditional statistical visualization will only take you so far, and sooner more than later, the statistical tools that you are so accustomed to won't be enough. You shouldn't have to fetch a computer scientist to set things in motion. Don't knock something just because it takes more than a moment to understand.
There was even one commenter who went so far as to call most of what appears on FlowingData garbage, which is absolutely unacceptable. Andrew, who I feel is more on my side on this one, disagreed. I, of course, completely disagree. This isn't just an insult to me and what I do, but to every person and group I've ever posted about. Stamen Design. Jonathan Harris. Martin Wattenberg. The New York Times. Brad Paley. Google. Seriously? Garbage? No way. I don't need to explain why.
As for one of those ending comments that said yesterday's guest post was "possibly the worst, least informative, most gimmicky presentation of data that I've ever seen," well, what can I say? That commenter hasn't seen much.
Statisticians, of all people, you should understand there's more to data than just the numbers. I know I do, but if it's sparklines that you want, go ahead and stick to those, and I'll continue with my garbage. Let's compare notes in 10 years.