Tuesday, February 19, 2008

Data Visualization Part 2: What's in a Name?


A couple of weeks ago, I wrote about the power of data visualization as an addition to the exhibit design toolkit. Paul Orselli made a thoughtful and challenging comment, saying:
...many data visualization art pieces, albeit elegant, seem to be inherently "push" technologies. That is to say, they parse selected bits of data for the viewer.

So how does finding patterns in streams of algorithmically-derived data move beyond the enjoyable exercise of discovering "shapes" in the clouds?

I couldn't come up with a satisfying response until a week later, when another colleague/reader (Matt DuPlessie) reminded me about one of the early, massively popular visualizations on the web: Name Voyager. Name Voyager touts itself as a "ba
by name wizard," allowing you to view the frequency of use of names for American babies per year, using data from the Social Security Administration. But it's not a list of names and numbers. Instead, it's a beautiful, quite intoxicating Java applet that shows you the relative frequency of names dynamically as you type--so that typing MO will show you how Mohammed matches up to Molly and Morgan, but when you get to MOR you just see the difference in frequency between Morgan (for boys) and Morgan (for girls).

Try it. It's another gorgeous time sink with data behind it.

And yet. The reason I bring up Name Voyager is because of a new element of their website: a link to a service called Nymbler, which invites you to type in names of interest so it can generate lists of related names you might like. It's sort of like Netflix for baby names: you rate names, it makes recommendations.

Checking it out last night, I was struck by the paradox that Nymbler gave me more useful information than Name Voyager, and yet I like Name Voyager more. I would use Name Voyager longer. Why would I prefer the less useful site?

Because I'm not having a baby. Name Voyager is a site that allows people to explore names through American history in an interesting way, whereas Nymbler provides an outcome-driven service. Perhaps if the Nymbler interface allowed me to see the algorithms behind their selections in a visually interesting way--as a shifting web of related names--I could get more deeply into exploration of what defines the set of names which appeal to me.

The difference between Name Voyager and Nymbler is instructive for exhibit designers. Too often, we go for the Nymbler model, both in terms of how we deal with content and the kind of interactions we provide. Content-wise, we are so interested in connecting the dots that we don't allow the kind of open-ended exploration that Name Voyager provides. Consider, for example, exhibits on global warming. Many such exhibits (and related websites) allow you to calculate your carbon footprint, step-by-step. Few allow you to do it in a way that dynamically reacts to each selection such that you can easily alter your choices to see the corresponding carbon drain. You select your vehicle, your eating habits, your power usage, linearly, and you have to go through the whole process again if you want to change a parameter. The design thinking behind this is that people are output-driven and want to see how their selections form a composite picture. But that precludes people from having a more flexible experience, one that is less focused on "my carbon footprint" and more on "contributors to carbon emissions." The general wisdom is that people will be more invested if it's "about me." But that's about me as an object of the exhibit, not I as the subject, I who am empowered to tinker with the parameters and my responses.

Bottom line, this comparison has made me realize that the thing that excites me about data visualizations is this empowerment. I am able, erroneously or not, to draw my own conclusions and perform my own simple experiments with the data. Intelligence officers get nervous about giving the president "the raw intelligence" for exactly that reason--it can be misinterpreted by non-professionals. But visitors aren't making national policy; they're learning. And at least in science centers, we profess to want to encourage visitors to think like scientists, like data-interpreters. With data visualizations, the visitors are no longer the object of the exercise. They are the subjects, and that's powerful, intoxicating, and hopefully, educational.

4 comments, add yours!:

Guy Davis said...

Interesting commentary. When I built this map of popular baby names, I wasn't really thinking about exploration versus goal-oriented design.

Hopefully, goal-oriented visitors will be able to quickly find the popularity of baby names on their short lists using the Search and Compare Names features.

For exploration of the dataset, I hope the use of Google Maps makes browsing popular names (both nearby and around the world) more interesting.

I'd love to hear your thoughts on how well the Baby Name Map accomplishes these two goals. Any suggestions you have for improvements would be appreciated as well.

Anonymous said...

From some of the people who brought you the Baby Name Voyager is the Sense.us project (http://vis.berkeley.edu/papers/sense.us/).
(The video is worth watching). I just wish this was live online, as the live demonstration was impressive.

The paper discusses the addition of annotation capabilities where people could argue about anomalies in the data.

Then of course there is Many Eyes (http://services.alphaworks.ibm.com/manyeyes/home) where I can not only play with other people's data, but download it, remix it, or add my own.

My question is: when can I play with data about museum collections this way? (http://xrl.us/bggmx)

POW! (Paul Orselli Workshop, Inc.) said...

Hi Nina,

Leaving aside whether somone was in the market for a baby name or not, Name Voyager is inherently more beautiful and has a more intuitive
interface.

I think Nymbler could be made to look and act more like Name Voyager, but the designers just want to make the look different for variety's sake?

Perhaps a different sort of lesson for (exhibit) designers there as well!

Ian Wilker said...

Found your post while rummaging about for data visualizations of carbon footprint, and I love this:

"Many such exhibits (and related websites) allow you to calculate your carbon footprint, step-by-step. Few allow you to do it in a way that dynamically reacts to each selection such that you can easily alter your choices to see the corresponding carbon drain. You select your vehicle, your eating habits, your power usage, linearly, and you have to go through the whole process again if you want to change a parameter. The design thinking behind this is that people are output-driven and want to see how their selections form a composite picture. But that precludes people from having a more flexible experience, one that is less focused on "my carbon footprint" and more on "contributors to carbon emissions." The general wisdom is that people will be more invested if it's "about me." But that's about me as an object of the exhibit, not I as the subject, I who am empowered to tinker with the parameters and my responses."

You say "few" allow users a dynamic experience of the data and let them play with it. I don't know of any — if you know any sites or have seen museum demos that do this, I'd be grateful for any referrals. Drop a line - iwilker at gmail.

I think you make a critical point here -- playing with the data allows one to begin to develop some perspective on how it all interrelates. I've been doing communications work around global warming for years, and the lack of good data visualizations is a huge impediment to mainstream understanding of the problem and its potential solutions, as well as to massive numbers of people getting into an action mode with regard to global warming. People just don't feel they understand all the moving parts -- they can't get a handle on it.