How Twitter Proves That Place Matters
Twitter is a fascinating place to explore not just the connectedness of people but of places. In a previous post, I mapped the locations of the 500 leading "Twitterati." When it comes to celebrities, the Twitterverse is still overwhelmingly American: almost three quarters of them are located in the United States. Los Angeles, with its large celebrity contingent, took the top spot among metros, followed by New York, San Francisco, Washington, D.C., and Atlanta.
A new study, "The Geography of Twitter Networks," by my University of Toronto colleagues Yuri Takhteyev, Barry Wellman and Anatoliy Gruzd from Dalhousie University takes a far more detailed look at the geography of Twitter and what it can tell us about the nature of interaction and proximity in the Internet age. Many predicted the rise of the Internet and of social media would annihilate distance and overcome the constraints of place by allowing people to communicate and build virtual communities. But the fact of the matter is Twitter actually works with and reinforces the power of place.
To better understand the connections between Twitter and location, the authors worked with a sample of 500,000 tweets. From these, they painstakingly identified roughly 2,000 "dyadic pairs" of tweeters, or Twitter users who often interact. Checking by hand, they were able to pinpoint more than 1,200 of these pairs to precise locations the size of a large metropolitan, spanning 386 distinct locations or "regional clusters."
What they found is that four out of every ten pairs of connected Twitter users fall in the same regional cluster—indeed, at distances of less than 10 kilometers, they are within easy driving distance, according to the analysis. Ties of less than 1,000 kilometers are more common than expected (if such ties were random) and ties of greater than 5,000 kilometers are much less so. While Twitter ties are also closer among those in the same country who share the same language, those two factors alone cannot explain away the effects of proximity. The frequency of airline connections between places is the best single predictor of long distance or "non-local" Twitter ties. This, the authors suggest, is a likely reflection of longer standing physical ties between places.
Their findings indicate that place and proximity continue to matter even in social media. Twitter doesn’t replace the networks that exist in the real world—it reinforces them and makes them stronger. Rather than freeing us from place, this study suggests, the Internet appears to enhance and even expand its role.