Home » Blog » ideas » Visualizing Twitter, understanding voluntary segregation

Visualizing Twitter, understanding voluntary segregation

I’m a total Martin Wattenberg and Fernanda Viégas fanboy. I loved their stuff at the Media Lab, at IBM and now at Flowing Media, their new visualization consultancy. Some of their work is “merely” beautiful:


Flickr photos of Boston Common, collected over the course of a year, abstracted to their color pallette.

And some is quite political and provocative. They show some new visualizations of Twitter at Personal Democracy Forum that focus on racial segregation on Twitter. (I’ve noted before that Twitter is a community with heavy African-American usage, a fact that’s invisible to many users.) They’ve carried out an experiment I’ve been desperate to try – they’ve selected a set of trending topics and hand-coded the participants in the threads to make guesses about the race of the authors. What they see are some tags – those related to breaking news, for instance – that don’t show a strong racial split. But others – including some that seem unlikely to be racially divided like “cookout” – are predominantly used by African American users. The amazing thing about this, Fernanda observes, is that the ability to interact with a different group of people is only a click away… and we rarely take advantage of it.

This sort of analysis and visualization requires a great deal of hand coding. But there’s lots of Twitter work that can be done by looking for the right hashtags. Martin describes “#hashtags as the bumper stickers of the 21st century”. By carefully selecting tags to study, they’re able to identify tea party conversations. Add some information to those feeds – the avatars of the people tweeting – and you’ll get a portrait of flags, eagles (and white faces.)

Interestingly, many tags aren’t an unambiguous signal for one side or the other. #p2 is a tag that’s common in progressive circles – it’s short for “Progressive 2.0″… and it appears surprisingly often in threads associated with the Tea Party. “What’s a nice, liberal tag doing in tea party clusters?” Well, a conservative Twitter user wrote an essay introducing hash tags to new users, offered a glossary of popular conservative tags and used #p2 as an example of tags liberals used. Now conservatives use #p2 to reach across the aisle to bug their listeners, and liberals use #teaparty to ensure conservatives see what they’re saying.

While the power of taunting may be one of the most powerful forces of the internet, it turns out that there is a pure, unpolluted liberal tag, one that is almost solely used by progressives. The tag? #npr

2 thoughts on “Visualizing Twitter, understanding voluntary segregation”

  1. You write that Fernanda notes, “the ability to interact with a different group of people is only a click away… and we rarely take advantage of it.” But isn’t this a post hoc ergo propter hoc argument? While numerous studies seem to indicate that we are very much more closely connected online than perhaps we are in, as we might consider it, the real world, our relationships are still typically informed by our immediate real world connections. Expecting that heterogeneity will spontaneously occur in the realm of, or even because of, social media seems fairly socially and scientifically optimistic to me. Groups form based around common interests, experiences, and beliefs; the Internet is just one more representation of this biological pattern.

    I would expect that the likelihood of increased cross-cultural exchange would be bolstered in online communities bereft of avatars or physical representations, but that may be a stretch, too. Our personas are not defined simply by what we look like, although that does inform personal experience as well as intellectual and affective representation. Humans are complicated, and personal understandings are contextualized in socially and culturally mediated ways; however, they are still experienced individually and subject to ongoing modification.

    In any case, the examination and subsequent visualization of these sorts of questions and hypotheses are worthwhile exercises, and I welcome them, even if their outcomes prove the simple idea that humans like to form groups and that both individually and collectively they are (somewhat) predictable in certain culturally-informed ways.

  2. Pingback: …My heart’s in Accra » Save the Galvao – the World Cup and good natured global taunting

Comments are closed.