Jennifer Golbeck is the director of the Human-Computer Interaction Lab at the University of Maryland. Her work focuses on how to enhance and improve the way that people interact with their own information online.
About two year ago, a story about Target spread through the news. The retail giant’s mailing system had sent pregnancy related discounts to a 15 year old girl and received some backlash from her father. It turned out that girl was in fact pregnant, but hadn’t even told her parents yet. By looking at little patterns of behavior, like buying extra vitamins or bigger handbags, Target was is able to make very accurate predictions about its customers.
Jennifer’s lab does similar work and can predict things like political preference, intelligence, age, just by using Facebook. One study was done just using people’s Facebook likes (looking at which pages they liked). A list of the top 5 likes indicative of high intelligence showed that one of those likes was actually for the “curly fries” page. Why? Because the action of liking reflects back to the attributes of the other people who liked it. If someone intelligent created the page, than their friends who like the page are probably smart as well.
The problem is that people don’t really have any power over how this data is used. Jennifer says that if she wanted to, she could quit her job and start a new company, selling reports to H.R. companies that predict how well you work in teams or if you use drugs or not. This is certainly something you would want control over. A solution she proposes is to develop mechanisms that tell users how risky certain online actions are. “By sharing this piece of personal information, you’ve improved my ability to predict if you use drugs”.
Although the work she does depends on using that very information, she would rather see a user base that is educated and informed. Her goal is not to infer information about users, it’s to improve the way people interact online.