InfoVis impressions, part 1

I got back from InfoVis 2007 a week and a half ago, and after catching up on the week I missed back home, I am just getting around to putting my impressions in order. Since I have a lot of stuff to post, I decided to serialize it. This is part 1.

Overall, it was a fantastic experience, and really gave me some much-needed insight in to what’s going on in the visualization community. There were many compelling presentations, panels, and posters there, but I’m mostly going to focus on the “for the masses” elements of the conference, since that is what is most related to my work. As the unofficial theme of InfoVis this year, the focus on the popularization of information visualization provided a fascinating and extended look in to not only what work is being done in this nascent “genre” of infovis, but also how the information visualization community as a whole is responding to its ideas and incorporating them in to the existing infovis domain.

To clarify that last point, I’ll start by saying that, as a first-time attendee of the InfoVis conference, I was quite surprised by the type of visualization discourse that dominated the event and, by extension, the infovis community that it represents. The outstanding impression I got from this conference is that the majority of research being done in the field of information visualization is focused on the technical details of designing visualizations (layout algorithms, data management, etc.), and is driven by a methodology that emphasizes a traditional scientific approach to this research. In other words, it presents itself (to this admittedly naïve observer) primarily as a branch of computer science, obviously reflecting and reinforcing its origins in that field.

That said, my exposure to the field originally came through my work developing interactive scientific visualizations for educational purposes (teaching principles of physics and biology to undergrads and high school students), the design of which focused primarily on supporting informal learning rather than, or in addition to, “focused episodes of work” (to borrow a term from Pousman, et al.). Though these educational tools were often used to help solve specific problems, we were more concerned with making them approachable to non-experts and promoting a casual exploratory usage model. As I began to study information visualization, most of my encounters with it have come through the internet, and though I’ve read most of the “bible” texts of the field that are clearly representative of the “computer science” approach, the “live,” publicly available, high-profile examples of infovis that you typically see on the net would generally be characterized as “casual infovis,” or “information aesthetics,” or “data art,” or whatever you want to call it. This would include systems and tools like Many Eyes, the Baby Name Voyager, the ubiquitous work of design firms like Stamen Design, the various visual web search tools like TouchGraph, art projects like wefeelfine.org, probably some business intelligence related visualizations like the Map of the Market, and even advertising-related infovis like the now defunct Coca-Cola WorldChill visualization (there are of course many more examples, these are just the ones I can think of off the top of my head). So, going in to the conference, my impression was that visualizations like these, regardless of the quality of their implementation, represented a legitimate area of study within the field of “information visualization.”

As it turns out, though, it appears that most of these examples probably wouldn’t be considered “information visualization” by the “information visualization” community represented by the InfoVis conference, presumably because, for the most part, they aren’t designed as tools with which you do rigorous analytic work. I saw a lot of evidence for this (which I’ll get to below), but it was most explicitly stated by Stephen Few in his capstone presentation: Stephen pointed to these kinds of examples on the web (in addition to some that I agreed were legitimately horrifying) as presenting a “primitive,” misleading view of what “information visualization” is, and suggested that it was the job of the conference attendees to be “model thinkers and communicators” that “take up residence in the real world” to show the “outsiders” what infovis is really all about. And this was framed as one of the more progressive viewpoints on visualization at InfoVis.

What was also surprising to me, based on the panels and presentations that explicitly addressed the idea of popular visualization, was that there didn’t seem to be much agreement on what “infovis for the masses” even referred to, or what its concerns should be. After the recap of the keynote below, the following posts are my observations on how this played out in the various related panels.

I already discussed Matthew Ericson’s Sunday morning keynote on info-graphics at the New York Times, so I won’t say much more about it other than reiterating how useful I thought it was. Matt described a design process that was very much “in the trenches,” emphasizing the special considerations that need to be made when designing for a mass, non-expert audience (i.e. readers of the New York Times). This included their development of a vernacular theory of visualization design that at times broke with what traditional infovis theory would suggest; examples included his observation that scatter-plots tend to confuse non-expert readers, because they are not familiar with graphs in which time is not on the x-axis. He also stressed the value of embedding of text annotations into their graphics to help guide a user through the data – something that a Tuftean purist might consider chartjunk. I was very impressed with his presentation and was happy to have a chance to speak with him afterwards.

InfoVis impressions, part 2: "Infovis for the Masses."
InfoVis impressions, part 3: "The Impact of Social Data Visualization."
InfoVis impressions, part 4: "Infovis as Seen by the World Out There."