'Project Gaydar' started as a project for an MIT ethics and law class to find out how much information could be inferred by examining a person's network of friends. A software algorithm applied statistical analysis to the gender and sexuality of a person's friends network, according to a report in the Boston Globe.
The program analysed the friends links of 1598 men, using the 'interested in' element of their profile to deduce whether they were straight, bisexual or gay. They ran the same test on 947 men with an unknown sexuality, and used personal knowledge of 10 people in a network who were gay but were not publicly 'out' on facebook. The analysis worked best with gay men, said the Globe.
Nathan Eagle, a researcher at MIT, worked on a separate project to infer peoples' friendships using data gathered from their mobile phones. Ninety four students and staff at the Institute were monitored using Bluetooth technology and phone mast information. Both location and phone calls were monitored, and used to build up a profile of that person's movements. This was then tracked against information recalled by the subjects.
Eagle's team could predict who was friends with whom with 95% accurately, simply by examining proximity to each other during key social periods such as Saturday nights.