InfoVis impressions, part 2: "Infovis for the Masses"

This is part 2 of my impressions of InfoVis 2007. Click here for part 1.

On Sunday afternoon was the panel I was most excited about coming in to the conference, “Infovis for the Masses.” It featured Fernanda Viegas, Martin Wattenberg, and Frank van Ham from the IBM Visual Communication Lab presenting on Many Eyes, Wesley Willet representing a group from UC Berkeley presenting their paper “Scented Widgets: Improving Navigational Cues with Embedded Visualizations,” Jock Mackinlay from Tableau Software presenting “Show Me: Automatic Presentation for Visual Analysis,” and Zach Pousman representing a group from Georgia Tech presenting a paper called “Casual Information Visualization: Depictions of Data in Everyday Life” (I blogged about this one a couple of weeks ago). Ben Schneiderman chaired the panel.

However, while each of the panelists made interesting presentations, the direction of the panel as a whole came off as largely incoherent. It was clear that we were looking at several different conceptions of “infovis for the masses,” none of which were really able to interface with one another. To me, the presentations on Many Eyes and “casual visualization” were most related to the conception of popular visualization (“for the masses”) that I write about on this blog. They were also the most controversial, based on audience response. Many Eyes is about democratizing and socializing visualization, empowering anyone to become a producer and consumer of visualization, and encouraging the benefits that come from social interaction around the visualization of data. The paper on “casual visualization” suggests that there are valuable ways to use visualization that don’t have to do with solving specific problems in an analytic way, and that traditional design principles don’t necessarily support this type of interaction.

The other two presentations seemed less “masses”-focused to me. The “scented widgets” presentation was certainly interesting, but maybe too technique-specific for the scope of the discussion (or what I hoped the discussion would be). Mackinlay’s presentation was a demo of the charting features in Tableau. To him, “the masses” referred to corporate data analysts needing to produce charts and graphs with software like his. While I think Tableau is an impressive piece of work, its cost alone guarantees that it isn’t “for the masses.”

Finally, following the panelist presentations, Ben Schneiderman, one of the forefathers of information visualization field, closed the discussion with a few statements related to this slide (this was taken from a presentation of his in 1998, but the slide he showed was the same) that hammered home the incoherence of the panel. He argued the need for “scientific studies” and “whole product solutions” to bridge the gap between “visionaries” and “pragmatists, conservatives, and skeptics.” I may not have motivated it well enough here, but this, in my mind, had almost nothing to do with what was discussed in the previous two hours. Perhaps it spoke to what Mackinlay was talking about, but it utterly ignored the presentations about Many Eyes and “casual infovis.” Discussion on “infovis for the masses” should be about how non-experts consume and produce information visualization, not about how brilliant “visionaries” can sell product to them.

The questions from the audience at this panel mostly revolved around Many Eyes, but in my opinion also missed the point. There was a lot of concern over the validity of the data being uploaded to the Many Eyes site, and the possibility of “poor” visualizations being produced by people who don’t understand how to “best” use the visualization templates offered. These questions certainly reflected the characterization of the community described in part 1 of my impressions: overly focused on the scientific validity and usefulness of the content on Many Eyes. Such a focus ignores what Many Eyes is actually about; what’s interesting about the site is not really the data being uploaded, but the type of use and interaction that is happening around the visualization of that data.