DATACLYSM

Are you a racist? Plainer-looking than you might wish? Inclined to vote left? Big data knows—and it’s talking.

Big data is more than numbers; it’s people. And it’s from the way that people describe themselves that the manipulators of big data know how to sell them stuff, which would seem to be the object of the exercise. If you visit a dating website such as OkCupid—which Rudder founded after receiving a math degree at Harvard—and say of yourself, “loves to be outside,” you’re statistically unlikely to be anything other than a white woman; add “country girl,” and the deal is sealed. The author looks at three big topics, often extrapolating from his own creation: “the data of people connecting,” “the data of division” and the data concerning “the individual alone.” What separates us is more interesting than what brings us together, and we’re incredibly inventive at finding ways to divide ourselves: sex and gender, age, appearance, cultural background, religion, musical likes, food preferences and, most of all, race. It is on that last, thorny subject that Rudder’s data becomes damningly meaningful: Americans are racist in ways that other nations are not, as measured simply by the exchange of flirtatious messages. The author is inclined to let the numbers speak for themselves without overlaying too much interpretation, though on race, he becomes impatient. We see things in aggregates and people as representatives of those aggregates, and “the patterns in the aggregate show that the dice, overall, are still loaded.” Although he hopes for a democratization of data that might further a more civil society, Rudder allows that it’s first-tier entities such as the National Security Agency and Facebook that are really in charge of the numbers, and it’s not comforting to know that “what’s being collected today is so deep it verges on bottomless.”

Demographers, entrepreneurs, students of history and sociology, and ordinary citizens alike will find plenty of provocations and, yes, much data in Rudder’s well-argued, revealing pages.