Sign In

Communications of the ACM

ACM TechNews

Project 'gaydar' Computes Orientation

rainbow handprints


Two students at the Massachusetts Institute of Technology (MIT) sought to determine whether online social network users were unknowingly revealing sensitive details through their virtual interaction with others, and they discovered through analysis of Facebook data that they could predict a person's sexual orientation just by looking at that person's online friends. MIT students Carter Jernigan and Behram Mistree used a program that studied the gender and sexuality of a person's friends and predicted homosexuality or heterosexuality through statistical analysis. "That pulls the rug out from a whole policy and technology perspective that the point is to give you control over your information — because you don't have control over your information," says MIT professor Hal Abelson.

Project Gaydar, as Jernigan and Mistree call it, is an example of the rapidly accelerating field of social network analysis, in which the connections between people are examined to see what information can be extrapolated. The project taps the principle of homophily, which posits that similar people tend to group together. The method has demonstrated practicality in the identification of gay men, as opposed to gay women or bisexuals.

The Electronic Frontier Foundation's Kevin Bankston says this discovery shows the risk that people run in participating in social networks. "Even if you don't affirmatively post revealing information, simply publishing your friends' list may reveal sensitive information about you, or it may lead people to make assumptions about you that are incorrect," he warns.

From The Boston Globe
View Full Article


Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA


No entries found