It’s doubtful that anyone really thinks of Twitter as a good example of democratic discourse. Sure, there’s plenty of good, interesting, even important things to read, but the Internet in general isn't known as being a safe place for everyone to express their opinions. Still, in some respects, Twitter might be a touch more democratic than you might think.
Democracy means many things to many people, but for the purposes of their study, Zhe Liu and Ingmar Weber turned to Jürgen Habermas and his idea of the public sphere, which placed the highest value on interpersonal equality, inclusiveness, and discussion that focused on common concerns rather than those of one social class. In theory, at least, Twitter could be such an institution, albeit in 140-character form.
The top 50 words in inter-ideology chats included "kill," "murder," and "hate," compared with words such as "love," "thank," and "great" for discussions among allies.
Before they could figure out whether Twitter met Habermas’ conditions, of course, they had to make those conditions concrete and define groups that might engage in some sort of political discourse. On the latter question, they decided to focus on tweeted conversations involving Democrats and Republicans, the Hamas-associated military group Izz ad-Din al-Qassam Brigades (whose original account Twitter was suspended, though others have since popped up) and the Israeli Defense Forces, and—why not?— Real Madrid and FC Barcelona. To address Habermas' criteria, Liu and Weber sampled a total of 226,239 Twitter accounts that had re-tweeted one of the group's tweets and examined how they interacted with allies and foes alike.
The pair’s analysis showed that Twitter isn’t an idealized Enlightenment salon, but it might not be quite so bad as we all thought. First, the bad news: Across social groups, defined by how many followers users had, tweeters were more likely to engage in conversation with those on their own side, just as political scientists have found in other contexts. Meanwhile, cross-ideology conversations were shorter than others and more often than not initiated by mentioning high-status users. They weren’t the most pleasant chats, either. The top 50 words in inter-ideology chats included “kill,” “murder,” and “hate,” compared with words such as “love,” “thank,” and “great” for discussions among allies.
There is a bright side, though. Users generally disregarded their Twitter social status and participated in cross-ideology conversation at rates largely independent of status. And, despite less-than-pleasant word use, a qualitative analysis of conversations between foes suggested tweeters avoided insults and made generally logical arguments, although they rarely cited any references.
Liu and Weber will present their results at the Sixth International Conference on Social Informatics in Barcelona this November.