I was asked about the most iconic data graphics in the last ten years for an article on FastCoLabs last week (so were Andy Kirk and Matt Stiles). It’s an interesting question not only because of the actual choices, but also the criteria to use. Is something iconic because of its unique look and/or shape? Does it have to have impact? What is an iconic visualization?
One of my choices was the Bikini Chart, which is perhaps my favorite bar chart ever. It’s simple, clear, easy to read. It’s also unique in its shape. I’m not sure if it has had the impact that it was probably meant to have, though.
Another choice was 512 Paths To the White House, which I think did have an enormous impact just before the 2012 Presidential election. It finally drove home to many people that the pundits were just producing a lot of hot air, while the statisticians (in particular Nate Silver) knew that the race was practically won. It came at exactly the right time and just worked incredibly well.
Another easy choice was gapminder. Hans Rosling made a huge impact with his talks, and has set the bar for data-backed presentations.
What all these have in common is that none of them came out of academic research. The first one was published by the Obama administration, the second one by the NY Times, and third one by a small non-profit. Where is academia in all of this? A lot of new things are thrown out into the world every year, but little effort is made to get them to stick. We could make a list of the most iconic things that have shown up in visualization publications, but it would seem like a rather artificial distinction.
Also, what makes a visualization iconic? Some of the things I chose, like Hannah Fairfield’s extremely clever scatterplot, are not that well known or have a very distinct look. But they do something that hasn’t been done before. Perhaps iconic is the wrong term, anyway. But novel by itself isn’t enough. There needs to be something else: impact perhaps, or a new understanding of the data that wasn’t there before.
The overall list is quite good. You can argue about individual items, but I think we hit the big ones. An interesting question would be how to make two or three lists using slightly different criteria (impact, novelty, etc.) and then compare how the different approaches fare. That wouldn’t make for as clear a headline, but it would be illuminating.