Want to know who your friends are? Check out your cellphone. No, not your list of contacts. Social scientists, after a five-year-long study, have developed a more full-proof method to sniff out your 'real' friends.
The study has opened new possibilities for social scientists, epidemiologists, and other researchers to understand how people connect and interact socially.
Nathan Eagle of the Santa Fe Institute in New Mexico and his colleagues Sandy Pentland of the Massachusetts Institute of Technology and David Lazer of Northeastern University in Boston handed out cellphones to 94 volunteers at MIT.
The phones were modified with software that logged the volunteers' calls, and used Bluetooth to detect when another of the phones was close by.
They looked for simple patterns in the logs of calls and times when phones were close together, and found that it was possible to predict who the volunteers would identify as their friends with 95 per cent accuracy.
For example, being nearby on campus during work hours meant little, but if two phones were close together for several hours on a Saturday evening their owners were likely to be friends.
"You can think of it as a behavioural signature," New Scientist quoted Eagle as saying.
The scientists could also link the phone data to the volunteers' satisfaction at work.
They found that people who reported themselves to be less satisfied were less likely to have friends in close proximity, and more likely to call friends during work hours.
The phones proved more accurate than the volunteers themselves at measuring how much time they spent physically near to others.
It was found that people typically overestimated how much time they spent close to friends, and underestimated how much time they spent with more casual contacts.
Although some of these findings may sound obvious, the study has offered an important proof of principle - the gadgets we carry day-to-day can accurately record the nuances of our relationships.
Using cellphones for social science research could replace interviews, which are laborious and sometimes unreliable, to find out about people's lives.
The cellphone approach may also have immediately practical applications such as helping epidemiologists predict how swine flu will spread from person to person.
The study has been published in the Proceedings of the National Academy of Sciences.