Alessandro Acquisti and Ralph Gross, researchers at Carnegie Mellon University, published a report with serious ramifications for United States law enforcement. This report, entitled "Predicting Social Security Numbers from Public Data" demonstrated that, if given a person's State and Date of Birth, they could often correctly guess that person's Social Security Number.

In their paper, Acquisti and Gross showed that their computer could properly guess 8.5% of social security numbers. And, their computer program guessed the first five digits of a social security number a surprising 44% of the time.  (Social Security Numbers have nine digits)

So, how did they guess so many numbers?

To start with, they analyzed publically available records from the Social Security Administration's Death Master File, which is a list of Social Security Numbers for people who have died. They found patterns between Date of Birth, State and Social Security Number and used these patterns to start predicting Social Security Numbers.

This study has major ramifications for everyone interested in combatting identity theft in the United States. Social Security Numbers have long been used as a secure, personal identifier.  If the system that generates these numbers is not secure, it is difficult to claim that Social Security Numbers actually add security to a transaction.

Thankfully, the researchers had some recommendations on how to combat this problem.  Acquisti and Gross recommended making the number assignment fully random and abandoning the matching of area numbers to states.