Introduction: This Ain’t Your Father’s Data
Throughout history, in one field after another, science has made huge progress in precisely the areas where we can measure things—and lagged where we can’t.
Car insurance isn’t a terribly sexy or dynamic business. For decades, it has essentially remained unchanged. Nor is it an egalitarian enterprise: while a pauper and a millionaire pay the same price for a stamp ($0.45 in the United States as of this writing), the car insurance world works differently. Some people just pay higher rates than others, and those rates have at least initially very little to do with whether one is a “safe” driver, whatever that means. Historically, many if not most car insurance policies were written based on very few independent variables: age, gender, zip code, previous speeding tickets and traffic violations, documented accidents, and type of car. As I found out more than twenty years ago, a newly licensed, seventeen-year-old guy in New Jersey who drives a sports car has to pay a boatload in car insurance for the privilege—even if he rarely drives above the speed limit, always obeys traffic signals, and has nary an accident on his record. Like just about every kid my age, I wasn’t happy about my rates. After all, I was an “above average” driver, or at least I liked to think so. Why should I have to pay such exorbitant fees?
Of course, we all can’t be above average; it’s statically impossible. Truth be told, I’m sure that back then I ...