InfoVis 2007: InfoVis for the Masses

Visualization for the People - Fernanda Viegas

The InfoVis conference this year had a theme that was not planned, but that made it even more impressive. That theme was InfoVis for the Masses, or Visualization for the People, and it was present throughout the keynote, many paper presentations, the panel, the World Visualization Day BOF, and the capstone. This is the beginning of a new era in visualization, and it is exciting to watch it happen.

Keynote: Matthew Ericson, Visualizing Data for the Masses

Matthew Ericson, deputy graphics director at the New York Times, spoke about the trade-offs between communicating complex data and packaging it in ways that a large number of people will be able to read. He showed a number of examples where they had tried quite complex visualizations, and others where they added hints to scatterplots to show what the different structures meant. Perhaps the most prevalent visualization are still maps, and Ericson gave a few examples of good and problematic uses of maps, and what they did to make them better (Enrico Bertini discusses Ericson’s keynote and this particular issue on his blog).

What strikes me as interesting is that readers of the New York Times – and more generally people who still read newspapers, for that matter – should be more inclined to learn about new things like visualization. By raising the bar a bit in a paper like the New York Times, the overall visual literacy could be raised quite effectively.

Ericson summarized his main take-away message “InfoVis for the Masses, In Short”: Find, Explain, Annotate, Design, Edit. And these were exactly the points that were discussed in presentations and discussions throughout the conference.

Paper Session: Visualization for the Masses

Ben Shneiderman chaired the session titled Visualization for the Masses, which contained four very interesting papers: one on Many Eyes, one on scented widgets for additional context in user interfaces, one on automatic configuration of visualization in Tableau, and one on casual uses of information visualization (as opposed to experts analyzing very specific data). Ben not only did the usual session chair things, but framed the session with additional remarks on the importance of the presented papers and how they related to other work (even fooling someone into thinking it was an actual panel).

Panel: Impact of Social Data Visualization

The panel The Impact of Social Data Visualization I organized continued in that vein. Fernanda Viégas (who also designed the brilliant image at the top of this article) wrote a posting on infosthetics where she described the feeling of a revolution in the room. Brent Fitzgerald of Swivel (who, like Fernanda, was one of the panelists) did not quite agree, and was also surprised by the discussion about the proper use of complex visualization tools by the unwashed masses.

This highlighted an interesting problem: academics are generally not used to seeing their work widely used, and so imagining that happen creates anxiety and even outright rejection. Ola Rosling (also on the panel) made a great comparison with photography: the tools for taking pictures are widely available, and people take lots of pictures. That doesn’t make everybody a photographer, but the general lack of skill and knowledge does not mean that people shouldn’t be taking pictures or post them on Flickr – the really good photos still stand out. Martin Wattenberg spoke about vernacular visualization, which does not have to follow the rules for “proper use” of visualization methods, but is still a valid way of looking at one’s data and can broaden our view of what visualization methods can be used for (perhaps similar to the way treemaps are now used more for categorical data than for hierarchies).

My attempts at stirring up controversy by making very political statements about and with visualization (using, among other things, my Presidential Demographics example) did not lead to the heated discussion I had hoped for, but it was a great discussion nonetheless with many interesting points: how do we know if we can trust the data, how do we get visualization tools into the hands of people with little or no access to technology, how do we create a repository of visualization techniques, etc. The panel was clearly a success, I spoke to a number of people who told me they had been inspired by it and were rethinking what their work could do for a much broader audience than they had originally thought about.

Mike Danziger has written a great account of the panel with some excellent observations.

Birds-of-a-Feather: World Visualization Day

This meeting was not publicized too well (other than here, of course), and the room was also a bit out of the way. Still, about 20 people made it there, and there was a good discussion on how to do it, who the target audience should be, etc. Stephen Few agreed to say a few opening words, and he made a good case for the importance of a broader understanding of visualization by the general public.

More needs to be posted on this topic, and more will be coming soon.

Capstone: Stephen Few, InfoVis As Seen By The World Out There

Finally, Stephen Few‘s capstone talk (which Fernanda also wrote about) on InfoVis as seen by the world out there rounded off the conference by covering the things people identify as visualization, and the ideas about visualization they get from the media and blogs. Examples include Few’s favorite punch-bag, information dashboards, as well as articles that depict InfoVis as mostly shiny eye candy, and finally products like the ambient orb.

I felt that the examples he used could have been a bit closer to InfoVis, though. It was a bit too easy for the audience to point fingers and say “Look at those dashboards!” or “Look at those 3D charts!”, and not understand that we are guilty of many of the same mistakes. Stephen did use some InfoVis examples in the tutorial he taught (which I wasn’t able to attend, but the tutorial notes are on the conference DVD).

Perhaps the closest thing to the conference he criticized was Swivel’s “bling” feature, and I’m a bit torn on that issue. While the name is certainly unfortunate (and perhaps betrays their own opinion of the feature), and the background images don’t add anything to the graphs, they are generally done in a way that is not obtrusive and that doesn’t make it impossible to read the graphs. Sure, they could have spent their time better on including more types of visualization; but then again, they probably reach a lot more business people who can use blinged Swivel graphs in their presentations and who wouldn’t want to use treemaps or other more complex visualization techniques, anyway.

Stephen has posted his thoughts as well as the slides from his talk, and it seems that he liked this year’s conference quite a bit. After his scathing criticism last year (that got him quite a bit of attention in the InfoVis world), that is good to see. It just makes sense to have not just the academic InfoVis club at the conference, but people who do practical work like Stephen Few, and people who make software that is used by lots of people like Swivel (I also spoke to a lady from Microsoft’s Excel group, and I’m sure other companies were represented).

Quo Vadis, InfoVis?

I believe that we are seeing a paradigm shift in InfoVIs, away from the specialized user (who may not exist, after all) to a more general audience. This won’t be easy – paradigm shifts are always painful –, but it’s a good thing. It will make us more aware of what we are doing and why, and will greatly increase the impact of our field. There is certainly a danger of “dumbing things down” for untrained users, and we have to be careful about that. But most user studies are done with users who are not familiar with whatever tool they’re supposed to use, and often not even with the application area, so we may already be halfway there. And also for those applications that are specialized and that only apply to a certain field, it will be healthy and helpful to be aware of the context and the usage scenarios in a broader context. It’s all about the user, after all.

Leave a Reply