The State of Information Visualization, 2013

Well, the world hasn’t ended, so here’s a look back at what happened in visualization in 2012, and a look ahead in case the world is still around a year from now.

2012: What Was

Last year was an exciting one, at least for me. First I started my sabbatical year at Tableau, then I decided to stay there. I went to SxSW and Malofiej. I met a lot of smart people.

2012 was the year visualization in the news took off. Forget Flash and slideshows, the real breakthrough has been D3 and its developer and champion, Mike Bostock. Alberto Cairo also published his book, and is teaching thousands of people how to turn data into interesting stories.

And you know who called it a year ago? Hm? Who? That’s right, this guy:

If you don’t believe that data journalism will be big in 2012, I have one word for you: U.S. Presidential Elections. Polls, primaries, more polls, ads, counter-ads, and then election night.

The New York Times, which used to hide its interactive pieces online, has posted a collection of the amazing work they did in 2012, including such pieces as 512 Paths to the White House. Since Mike Bostock works there now, many of them were done in D3.

This marks a sea change not only in terms of interactivity on the web, but in how newspapers are starting to approach the issue of browser support. Rather than play to decade-old technology (remember IE6?), they now work on the cutting edge and provide limited fall-backs for people on prehistoric browsers. Technology moves on, and news media have finally started to get it. This isn’t just true of papers like The New York Times or The Washington Post, I’ve heard similar things from much smaller papers.

2013: What Will Be

My predictions have at times been self-serving. I knew when I predicted more theory work in 2010 that I was going to be doing some of that myself.

So this time, I will not just predict that storytelling in visualization will be big, but outright say that I have every intention of making that happen. I will be at Computation+Journalism (on a panel with Alberto Cairo) in two weeks, then there’s the Tapestry conference, and right after I will be at NICAR (on a panel with Amanda Cox). Jock Mackinlay and I also have a paper on storytelling coming out in the special issue on future research directions for visualization of IEEE Computer (May issue), and I have a few more things in the pipeline.

Communicating data to people using visualization is an exciting and important direction for visualization, and one I deeply believe in. And with the growing availability of data, as well as journalists’ increasing ability to build exciting and informative pieces (and the continuing need for good, accessible information), we will see a lot more interesting visualizations this year and in the years to come.

In less self-centric future developments, Many Eyes is coming back! I know that IBM is rebuilding a new visualization group after shamefully neglecting and abandoning the one they had. They have finally come to their senses, perhaps after some nudging from yours truly, and are about to release a revamped version of Many Eyes. I believe the targeted release date is in March, if I’m not mistaken. Given some of the people involved, I’m cautiously optimistic.

Beyond 2013

Maybe it’s my new perspective in my new job, but I’m starting to lose patience with research that doesn’t even try to solve realistic problems. I’m not talking about theory, I’m talking about doing work for its own sake rather than because it actually addresses a problem or provides any kind of real insight.

And I’m not talking about purely applied work, either. Look around who is currently killing it with visualization: newspapers. They’re where it’s at. Not academia. The most exciting work last year was done by people who mostly don’t even have advanced degrees; and yet they also publish or perish (see what I did there?). Academics should take note, and they should be scared.

If you think I’m ranting, wait until I unload my actual rant on the topic here soon. But this is not a new thing. I actually mentioned this in last year’s posting. Academic visualization research needs to stop messing around and figure out where the interesting work is happening, and get into the game.

There is plenty of opportunity, too. BioVis is still up and coming, but it’s showing how application-specific work is done right. Graph visualization sucks. Interaction in visualization is mostly an afterthought, and terrible. Too much work is still tied to the idea that every individual data point needs to be shown, even if most questions are asked in terms of sets and subsets. And visualizing those isn’t exactly a done deal, either.

So let’s get started!

Published by

Robert Kosara

Robert Kosara is Senior Research Scientist at Tableau Software, and formerly Associate Professor of Computer Science. His research focus is the communication of data using visualization. In addition to blogging, Robert also runs and tweets.

8 thoughts on “The State of Information Visualization, 2013”

  1. “Graph visualization sucks”. No argument about that being true for traditional techniques and current common practice. But I contend that by taking the approach, demonstrated by BioFabric, that nodes can be successfully represented as higher-dimensional geometric primitives (i.e. lines) instead of as points in a node-link diagram, there is now a new way to think about and experiment with graph visualization. In other words, maybe “In 2013, maybe graph visualization doesn’t have to suck as bad anymore”?

  2. One of the interesting things in newspapers graphics is how interdisciplinary the teams are. We hear that from Amanda Cox and Kevin Quealy every time they speak or post about the NYT process. Academic work — and we see this in the papers — is NOT interdisciplinary in the same sense. And it really shows in the output, which will usually be strong on one dimension but really bad on the others. The other issue with academic research is that the tendency towards building frameworks and systems, at least for vis tools, rather dilutes perceived value from the outside. Until someone creates something amazing with it – like the NYT with Mike’s help.

  3. It is exciting to see infovis technology move into the mainstream. Storytelling of course is fascinating and is often essential to communicate with any audience. However I have some qualms. Some folks spent many years learning to read and write academic papers, read and write code, understand and occasionally invent algorithms, use some discrete math, and even build usable large scale systems. Some people spent years studying and learning to create graphic designs or even art, or learning to write or tell stories. They are both necessary skills but it is a fact of life that what the math/code nerds do and what the art/journalism kids do are very different, they are usually not good at doing what the other group does, and the pay scales are very different.

    As for the criticism that graph visualization, er, is lacking, but BioVis is great. Sure, in the end, domain-specific representations and tools are really what you want. But how do you get there? BioVis is built on more fundamental techniques. The best, most beautiful systems are going to have sophisticated underlying algorithms that generate beautiful representations automatically, because not many of us can get 20 designers from the NYT to design special purpose HTML5 graphics for one particular data set. It doesn’t scale. Interaction doesn’t make up for lousy visualization algorithms – “I’m sorry it looks bad, but maybe you can use these controls to get something better.” More basic, domain-independent techniques will also be in demand there’s the question of what to do when a data set doesn’t fit some canned domain-specific tool. Some people that have worked in network visualization for a long time do realize that the audience doesn’t want to draw networks – they want to create diagrams, but the technology is really lagging. If anything there should be more investment in this area, not less.

    Interaction is great to explore details, to use visualization as a human interface into some underlying data set or process, no question there.

    Thank you for writing the article – it’s a stimulating discussion.

    1. You make too many points to easily answer in a comment response, but I’ll say some more about them in a coming posting. We agree on more things than it might seem.

      I picked graph visualization and biovis as examples, there are many more. The thing about successful news graphics isn’t just their aesthetics, but also the thought processes that go into them and the fact that they simply work extremely well. That is way beyond prettiness, it’s about understanding the purpose of what is to be shown.

      And yes, I understand the difference between academic and news publishing, but I think we can learn a lot from the way news works. More on that soon.

  4. Robert,
    Love your realist points of view, about so many aspects of data visualization.
    Your site is a wonderful resource and I’m looking forward to mining it for ideas for telling the stories locked in my huge datasets.

Leave a Reply