Facebook knows you better than you think
When we think about social networks, we often think about them in relations to the members of your network. For Facebook, a typical network would consist of people that are your friends (nodes), as well as friendship (edges). But a lot less often do we consider the consequences of our actions of networks in terms of personality. According to the Wall Street Journal, a lot of what we do on Facebook can reveal more than just who we are friends with, but what we like.
WSJ reports on a recent study by Michal Kosinski who capitalized on a large amount of like data from Facebook. Turns out, computers are vastly better at judging key personality traits such as introversion and extroversion. The vast number of data that can be obtained from a person’s likes is incredibly representative of personality traits.
With a very small number of like data, a computer can get to know you better than some of your closest relationships:
10 likes: computer knows you better than a coworker 70 likes: computer knows you better than your friends 250 likes: computer knows you better than your spouse
In relation to INFO 2040, Kosinski’s algorithm is probably using the property of Strong Triadic closure to make some predictions. For example, if the person has a lot of strong ties to activities like reading, knitting, or time alone, then it is more likely to form connections between those strong ties and ties to personality traits, such as introversion.
Capitalizing on this information raises some privacy and ethical concerns, and Kosinski brings up the point of advertising and marketing. If a computer is able to examine the things that you have liked and make inferences about what your personality is, it is also more than likely able to form a network between your personality traits and the products you are more likely to buy. Kosinski believes that empowering marketing companies with this info could lead to more effective ads.