A computer can know you better than your friends can. You can also fool your friends more easily than you can fool a computer, a new study of Facebook data indicates.
The digital "footprints" you leave behind every time you go online -- most particularly Facebook "Likes" -- provide enough evidence for a computer to more accurately predict your personality than most of your family and friends can, U.S. and British researchers say.
In a study by scientists at Stanford University and the University of Cambridge, people offered judgments on others based on their familiarity with the individual being judged, while computer models used the specific digital signal of Facebook Likes.
A computer proved more accurate in predicting a subject's personality than a subject's own work colleague after analyzing just 10 Likes; better than a friend or a roommate with 70 and a family member such as a parent or sibling with 150.
Only someone's spouse could match the computer's accuracy in the assessment of personality, the researchers reported, and even that was reversed once the computer was given sufficient data for analysis: around 300 Likes.
The average Facebook user has provided around 227 likes, the researchers found, a number likely to grow steadily.
The pattern of people's Likes on Facebook is enough for a computer model to predict personal traits such as gender, sexuality, race and political leanings, they said.
Compared with people predicting their friends' personalities, the computer's predictions based on Facebook Likes were on average almost 15 percent more accurate, they noted.
This is in spite of experiments that have suggested, "people are very good at judging each other," Cambridge psychologist and study lead author Youyou Wu said.
The finding that computers can describe us better than we'd previously thought represents an "important milestone" on the way to increased social human-computer interaction, the researchers say.
"In the future, computers could be able to infer our psychological traits and react accordingly, leading to the emergence of emotionally intelligent and socially skilled machines," Wu said.
Given that such an outcome could raise privacy concerns, the researchers say they support the policy of users having complete control of their digital histories.
"We hope that consumers, technology developers, and policy makers will tackle those challenges by supporting privacy-protecting laws and technologies, and giving the users full control over their digital footprints," said study co-author Michal Kosinki from Stanford.
