The iPhone, not the eye, is the window into the soul

 

It is great to be back at Prawfsblawg this year.  Thanks to Dan and the gang for having me back.  For my first post this month, I wanted to point everyone to the most important privacy research of 2012.  The same paper qualifies as the most ignored privacy research of 2012, at least within legal circles.  It is a short paper that everyone should read.

The paper in question,Mining Large Scale Smart-Phone Data for Personality Studies, is by Gokul Chittaranjan, Jan Blom, and Daniel Gatica-Perez. Chittaranjan and co-authors brilliantly show that it is straightforward to mine data from smart-phones in an automated way so as to identify particular "Five Factor" personality types in a large population of users.  They did so by administering personality tests to 117 smartphone users, and then following the smartphone activities of those users for seventeen months, identifying the patterns that emerged.  The result was that each of the "Big Five" personality dimensions was associated with particular patterns of phone usage.  For example, extraverts communicated with more people and spent more time on the phone, highly conscientious people sent more email messages from their smartphones, and users of non-standard ring-tones tended to be those who psychologists would categorize as open to new experiences.  

There is a voluminous psychology literature linking scores on particular Big Five factors to observed behavior in the real world, like voting, excelling in workplaces, and charitable giving.  Some of the literature is discussed in much more detail here.  But the Chittaranjan et al. study provides a powerful indication of precisely why data-mining can be so powerful.  Data mining concerning individuals' use of machines is picking up personality traits, and personality predicts future behavior.  

The regularities observed via the analysis of Big Data demonstrate that you can aggregate something seemingly banal like smartphone data to administer surreptitious personality tests to very large numbers of people.  Indeed, it is plausible that studying observed behavior from smartphones is a more reliable way of identifying particular personality traits than existing personality tests themselves.  After all, it is basically costless for an individual to give false answers to a personality questionnaire. It is costly for an extravert to stop calling friends.  

Privacy law has focused its attention on protecting the contents of communications or the identities of the people with whom an individual is communicating.  The new research suggests that -- to the extent that individuals have a privacy interest in the nature of their personalities -- an enormous gap exists in the present privacy framework, and cell phone providers and manufacturers are sitting on (or perhaps already using) an information gold mine.  

It's very unlikely that the phenomenon that Chittaranjan et al. identify is limited to phones.  I expect that similar patterns could be identified from analyzing peoples' use of their computers, their automobiles, and their television sets.  The Chittaranjan et al. study is a fascinating, tantalizing, and perhaps horrifying early peek at life in a Big Data world.