Information visualization conversation with Fernanda Viegas and Mike Migurski (to be continued…)
Fernanda and Mike will be talking about the latest actvities in information visualization at IDEA, on October 24.. I’ve goaded them into engaging in an email conversation, the first part of which I’m reprinting here.
Fernanda: At the conference, I’m hoping to talk about how to “democratize” visualization use (following successful deployments such as the NameVoyager vis) and I’ll demo the new public visualization site we’re building in our group. Michal got a peek of it this weekend [At FooCamp].
Even though we got a conversation started, I don’t think we got to any conclusions about how to do our presentations.
Michal, what are your thoughts on this?
Michal: I like what you’re saying about democratization, though strangely enough I’ve always used the term “downmarketing” in the same context, as in making the techniques we use understandable and desirable to a wider range of people. The NameVoyager project is a perfect example of this. Is there a reason you’re using a political word while I use an economic one? Is that the difference between coming out of academia and research environments vs. having a service-oriented design firm?
That may be a good starting point.
Fernanda: You ask an interesting question. I’m not sure what the difference is between my “democratization” and your “downmarketing” but I can tell you a bit more of where we’re coming from.
Screenshot from Themail, described below
Martin and I have been building visualizations for a while now and it’s recently become clear to us that whenever we make a visualization public, it takes on a life of its own and becomes a much more powerful artifact. To give you an example, when I was finishing my thesis at MIT, I was working on visualizing people’s email inboxes. I was super cautious about privacy and made a point of always explaining to my users (the owners of the email archives being visualized) that they would be the only ones looking at those visualizations. I explained to them that I would never show those images to anyone else without their consent. Well, as soon as people started playing with the visualizations, they wanted to share the images with others!! They started sending screen shots to friends and family and they would call others to sit with them and look at the images together. Users were using the images as social artifacts for reminiscing and storytelling. Martin’s experience with the baby name visualization was very much along the same lines in the sense that it allowed hundreds of conversations to get started because of the visualization.
Meanwhile, if you look at the academic information visualization community, researchers aren’t focusing on the social side of their applications. Infovis folks love to explore techniques that allow them to scale the data they are showing. But what happens when you scale the audience that’s looking at a visualization? This is the question we are currently exploring.
So I guess our “democratization” efforts don’t necessarily attempt to make visualization techniques simpler as much as they try to support social activity around visualization viewing (things such as: how can you support conversation around visualization?).
I hope this helps. I’d love to hear your thoughts on this.
Peter: I love your insight, Fernanda. I’ve long been frustrated with the state of information visualization, because it seemed to offer such promise, and yet languish for so long. And I wonder if you’ve hit on a key — visualizations are taking off, of late, because they serve a primarily social role; they enable sharing between people.
It reminds me of one of the simple, but genius, moves of Flickr. Whereas other photo services defaulted to “private” sharing, Flickr defaulted to “public”. This allowed for the rapid development of community.
Information and data visualizations seem to be taking off as artifacts suitable for sharing. I’m reminded of the buzz a couple years ago around extisp.icio.us, which visualized your del.icio.us tags (and this before the prominence of tag clouds). This allowed you to create a kind of visualized avatar of yourself. (Which in turn reminds me of “Personal Dictionaries”, an art project from 1995, where people’s additions to their word processor’s dictionary were overlaid on their photograph.)
Anyway, back to your point, I think a key success of the NameVoyager is to keep the data being visualized *super simple*. It’s almost like there’s an inverse relationship between the complexity of the data, and the complexity of the conversation around the data. And I suspect Mike saw some of this around Mappr…
Michal: Fernanda, the sharing aspect of the forthcoming piece you mentioned at Foo was, I think, a total masterstroke. Peter, some background: similarly to how you describe people communicating insights (e.g. “note how Adolf drops off in the 1940’s”) via the NameVoyager, Fernanda & Martin’s new piece provides a “comment on this” feature, which saves the viewer’s state along with a thumbnail screenshot. This way, the comments section attached to each dataset is augmented with tiny screenshots that show what the commenter was looking at when they decided to respond. It’s a super elegant solution to the problem of synchronizing views of changeable data.
I agree with Peter’s opinion about the relationship between complexity and popularity. It’s a lot like pop music in that way - provide a hook that can be hummed, and then backfill the complexity into the production and subtext for longevity. The KLF’s book “The Manual” articulates this more effectively than anything else I’ve seen, and has served as a foundation for most of my interests for the past 4 or 5 years. An effective visual interpretation of information hides the same kind of subtlety in a simple presentation.
Probably the point at which our focus differs most from IBM’s is that we’re currently fixated on liveness - data that changes as you view it, typically because it’s being generated at the same time. The recent Digg Labs work is the most high-profile version of this we’ve got, but we’ve attempted to approach it with other pieces that change from day to day or hour to hour. It’s related to Peter’s mention of sharing, since a lot of interesting or constantly-shifting data is from social sites like Flickr or Digg. Those are the places where change happens fastest and matters most to people. I like the idea of visualization as an alert system for social information rather than a contemplative one, and we’ve made efforts to spur projects with this characteristic.
A side effect is that visualization of live information turns out to be a lot more TV-like than NameVoyager or similar work. Extispicious and Personal Dictionaries both remind me of the quiz craze on LJ (e.g. “what kind of Hostess baked good are you?”), whose output was very badge- or avatar-like. I’m not sure if the same impulse can be drawn out of something live, or whether perhaps viewing live-but-narrow data is actually quite a different experience from manipulating static-but-deep data. Fun fact: the original drafts of our Digg work were called “The Ultimate Stoner Tool”, because author & blogger Om Malik said during some panel that his favorite activity was smoking dope and watching Digg Spy scroll by.
[[to be continued…]]