It turns out your parents were right all along: You really can be judged by the company you keep.
Using a software program they created called "Gaydar," Carter Jernigan and Behram Mistree (who have since graduated) analyzed the gender and sexuality of a person's friends to predict that person's sexual orientation.
They weren't able to verify all of the software's predictions, but based on what they knew about their classmates' offline lives, they found that the program seemed to accurately identify the sexual orientation of male users, in a sense indirectly "outing" them by analyzing the characteristics of their online "friends."
The findings have not been published but, in an e-mail, Mistree said the pair has a paper in submission to a journal. Aside from stating that "We thought that our work demonstrated a new threat to privacy that we wanted individuals to be aware of," he declined to comment.
Jernigan did not immediately respond to requests for comment from ABCNews.com, but he told the Boston Globe, "It's just one example of how information could be inadvertently shared. It does highlight risks out there."
Hal Abelson, the MIT computer science professor who taught the researchers' class, said that while the students couldn't execute the project up to rigorous scientific standards because of classroom limitations, the research still highlights the fact that social networking indirectly exposes a large amount of personal information.
"The whole notion that your information is just about you -- that isn't true anymore," Abelson said, adding that the project shows that that policy makers and companies need to adjust how they think about how individuals control privacy online.
"The point is when the information is so interconnected, information about me isn't just about me," he said.
Even if a user goes to great pains to withhold personal information (by changing default settings, refusing to post political or religious affiliations or sexual orientation, or refraining from posting photos), he said information on a friend's page -- or even just the friend list itself -- could lead others to make assumptions or draw inferences.
For their project, which they began in 2007, Carter and Jernigan accessed Facebook information for students in the MIT network and were in classes 2007-2011 or graduate students, according to the Boston Globe.
First, they analyzed the friend links of 1,544 men who identified as straight, 21 who said they were bisexual and 33 who said they were gay to determine correlations between a user's sexual orientation and that of his friends. They found that gay men had proportionally more gay friends, which gave the computer program a way to predict sexual orientation based on friends.