Tuesday, March 18, 2008

Six degrees of messaging

Even though Milgram's original paper has been largely discredited on methodological grounds, this item, which reports on an a large sample of IM traffic, seems to support its basic conclusion. Quoting the article:

Eric Horvitz, at Microsoft Research in Redmond, Washington, and Jure Leskovec, who was an intern at the time, crunched through masses of data, logging a month's worth of global 'instant messaging' conversations using Microsoft Messenger — software that facilitates chat, in a similar way to e-mail, but in a more instantaneous and less formal fashion. The researchers then counted how many messages were sent and from where: in total they tallied up a whopping 255 billion messages sent in the course of 30 billion conversations among 240 million people during June 2006.

No personal or identifiable data could be seen, and the researchers had no access to message content, although they could correlate messages with information about age and gender logged by users when they registered for the service. “We didn’t probe individuals,” says Horvitz, “we were looking at patterns.”

The resulting figures produced a neat map of communication hotspots across the world, and allowed Horvitz and Leskovec to trace the extent of separation between Microsoft Messenger users. They found that the average shortest number of jumps to get from one random user to another was 6.6; spookily close to the infamous six degrees of separation demonstrated practically in a group of 64 people by Stanley Milgram, at Harvard University, in the 1960s.

No comments: