barking up the wrong (digital) tree

just saw this and I couldn't resist commenting.

apparently, DARPA is soliciting proposals for research that will help security and intelligence officials use social media (read: Facebook and Twitter) identify rebellious intent, predict mass public action, etc. OK, not a particularly new story, even if they are sinking $42 million into it.

what's interesting here is the implication, present in so much research on social media, that social media = reality. if it's on the internet, it must be true. for example, consider the 2004 Howard Dean primary campaign. Dean had a huge following on the internet, and, going into the Iowa caucus, many predicted him as the frontrunner. however, he ended up finishing a distant third and soon thereafter dropping out of the primaries (no doubt in part due to the media reaction to the Dean scream). what happened? the exuberance Dean's supporters showed online was not representative of the general voters in the Democratic party, or at least not those at the Iowa caucus. since these other voters were not part of the online political scene, it would have been hard for analysis of online content to determine that the Deaniacs were likely a minority.

granted, this is a bit of a simplification, and I'm choosing here to focus on a particular, limited portion of the account. the point, though, is that online furor and enthusiasm does not necessarily translate to impact and action. yes, there are certainly numerous interesting examples of being able to predict everything from election results to movie ratings based on social media content and activity, and the Pentagon seems to hope to expand this predictive power to issues of national security. however, it seems equally if not more important to attend to when, why, and how such predictions prove inaccurate, especially when we're talking about suspicions of political unrest or even terrorism.