The State of Information Visualization, 2012
Another year has come and gone, and many exciting things have happened in information visualization. Here is a look back at some interesting events from last year, as well as what I expect for 2012 and the next few years.
2011: What Was
The launch of Protovis was a big deal in 2010, but it was bettered by D3 last year. Unfortunately, that also means that Protovis was abandoned only about one year after being publicly introduced. D3 is clearly the more powerful and versatile system, but it is also a lot more generic than Protovis. Rather than providing clean, visualization-specific primitives, D3 is a general-purpose DOM manipulation tool that requires designers and programmers to dig around in ugly, XML-infested SVG. The approach is clearly the better one from an architectural computer-science point of view, but for many users, it's a step in the wrong direction.
In my last State of Information Visualization, I expressed my hopes that IE9 would be adopted quickly by businesses, which would make it easier for visualization projects to rely on HTML5 rather than Flash. While IE9 does not appear to be quite as common as I had hoped, something similar (and perhaps even better) has happened instead: more people are dropping Internet Explorer altogether. Chrome in particular has been gaining steadily over the last year, mostly at the expense of Internet Explorer. This development makes rich, powerful visualization possible in the browser, and I am seeing more and more of that every day. Showing static charts as images on the web is nothing new, but for the user to be able to talk back and work with the visualization is. It's a bit like the move from static websites to more user-centered pages with comments, etc.: visualization on the web 2.0.
Data journalism keeps getting attention in visualization. There was another workshop at VisWeek on storytelling, as well as two papers on the topic (more on those in a future posting). The academic community is still mostly concerned with data analysis rather than communication, though there is clearly interest. But the world out there is not waiting. The Guardian datablog is collecting and releasing data every day, and more and more stories are now based on numbers and use some form of visualization. It's easy to complain about the bad ones, but we also need to come up with ideas how to tell stories around data in a constructive way.
That is also what got me excited about the launch of Visual.ly. Their current infographics directory is nice, but their more interesting and ambitious project is still to launch. Being able to build graphics-rich visualization (or data-rich infographics) will be incredibly powerful and could make a big difference in the way stories are told with data.
On a more academic subject, we saw some great visualization work in bioinformatics at the new BioVis Symposium at VisWeek. This is a trend that started a while ago and that is undoubtedly here to stay. Most of the work was interesting because it solved real problems, not because it was necessarily pointing into new directions in terms of visualization. However, I think that the connections made at the symposium and the awareness of the wealth of data and questions in bioinformatics is bound to lead to very interesting new work.
Another topic that seems to be heating up is graphs. A lot of network visualization was (and still is) centered around variations of the node-link-diagram. It's no secret that these visualizations don't scale beyond a few dozen nodes, and the amount of actual information that can be gained from them is really rather small. There are already some approaches in the right direction, like Martin Wattenberg's PivotGraph and matrix-based graph visualization techniques. None of them have seen wide adoption so far, but there were several papers at VisWeek last year that looked very promising. Whether they visualize networks in a hybrid way (using both node-link and the matrix) or use networks to analyze relational data, it's all starting to get much more interesting than yet another graph layout algorithm.
2012: What Will Be
The big deal in 2012 will be visualization in data journalism. A lot of attention will be coming our way as part of the 20-year anniversary of the Malofiej Awards, aka The Pulitzer Prize for Infographics (where I will be a judge this year). New, well-funded Data Journalism Awards were also just announced that are backed by Google and that include a visualization and storytelling category. This is a reflection of both the importance of data-based journalism and the wave of new data stories that are about to hit us. And many of these will at least include visualizations, if not be entirely built around them.
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. There will be more data than ever before, and it's not like there was no data last time. This will be huge, and it's mostly based on numbers. No visuals is not an option, so it will be touch-screen maps vs. holograms (and, hopefully, some better alternatives).
While it's exciting to see this happening, my concern is that the academic community will miss the boat. I don't want to end up with a situation like in human-computer interaction, where all the constructive work is done in industry and all academics seem to do is study what others build. We need to stay on top of what is going on and think hard about how we think things should be done. If we fail to do so, we'll have no right to complain if the results end up not being what we wanted them to be.
My other big hope for this year is networks. That networks are important is not a question, but I just don't think they've been done well so far. The ubiquitous hairball may be nice to look at for a bit, but to actually get information out of these things, we need better techniques. I'm excited about what has happened over the last year or two, and I believe that we are on the verge of seeing some really interesting new techniques that will make it not only possible to work with large graphs and understand them, but do so in a way that is easier to grasp than what has been done so far.
Data journalism will keep getting more interesting. I don't think that we have even really scratched the surface on this so far. Another area that I think is ripe for change, and where we're going to hit a wall soon, are information graphics. You can only look at so many colorful, long-aspect pictures before you get bored and realize that you're not actually getting anything out of them. There is a lot more that can be done here, and I will write about those in a future posting. This is not simply about arguing how much chart junk is good or bad, it's about the basic idea of what an infographic or visualization even is. In the longer run, I think that these two seemingly irreconcilable fields will merge, or at least end up using the same ideas (and many of the same techniques).
The visualization field also needs to become smarter, work on more advanced techniques (rather than simply new ways of doing the same things), and organize more effectively around the key issues. We're doing fairly well so far, but we need to keep it up. The competition from the corporate world is increasing with more and more companies, including some of the biggest software makers, discovering visualization. We need to make sure that we're heard and that we are able to contribute meaningfully to what is being developed, and don't just work on our toy datasets and problems inside our ivory towers.