Levy: Campaigns Get Personal

Well before the party conventions, it's fair to say that technology has made its mark on the 2008 election cycle. Bloggers are at the center of the political conversation, and the Pandora's box of YouTube is an endless source of macacas, ObamaGirls and incendiary pulpit sermons. But within the campaigns themselves, the talk is of a technological advance based on number-crunching, databases, selective surveying and laser-focused voter contact. To political technocrats, 2008 marks the maturation of "microtargeting"—a technique that, if things are as close in November as expected, may well affect who takes the White House.

Microtargeting, as its name implies, is a way to identify small but crucial groups of voters who might be won over to a given side, and which messages would do the trick. It's also a bit scary, because instead of trying to figure out how to direct media and mailings to a fuzzy cohort like "soccer moms," microtargeters know who you are, and try to push your personal hot button so that you'll choose their candidate.

I got an explanation of how this works from Colin Shearer, a former artificial intelligence scientist who is now a VP of SPSS, named after a software project called the Statistical Package for the Social Sciences. His company provides the key tools of what it calls "predictive analytics" (the secret sauce of microtargeting) to commercial concerns and, increasingly, political customers. SPSS's customers include companies you may not have heard of—Strategic Telemetry, Catalist, TargetPoint and Penn, Schoen & Berland—but they are the key consultants for microtargeting efforts of both parties. (Strategic Telemetry is currently working for Obama, Penn for Hillary, Catalist works with the DNC and TargetPoint worked for Mitt Romney but is widely expected to be onboard for McCain in the general election.)

The process involves gathering elaborate information on voters (Alex Lundry, research director of TargetPoint, talks of a "data DNA profile") that can include public items like party affiliation, ZIP code-based assumptions on income level and housing, and fairly detailed consumer preferences such as which car you drive, where you vacation and which entertainment you prefer. (Yes, that stuff is for sale, enabled greatly by a widespread neglect of those "opt out" checkboxes in the fine print of privacy policies.) That information is augmented by surveys that link those traits and behaviors to attitudes on political and social issues.

Careful analysis can yield counterintuitive opportunities to win votes. In 2004, Republican microtargeting in New Mexico found a strain of education-obsessed Hispanic moms who responded positively to mailings and phone calls touting George W. Bush's No Child Left Behind law. Democrat microtargeters discovered what they called Christian Conservative Environmentalists. Find such people (by data-mining the information), craft a message that resonates with their particular bugaboos, contact them directly, and you may get votes that otherwise would never have found their way into your tally.

In 2008, microtargeting is coming into its own. Not only are the techniques more refined, but the software has been improved and more powerful hardware allows standard laptops to quickly churn out results that previously could be done only with bulkier and slower computers. Databases can be mined more efficiently. As a result, says Marc Rotenberg of the Electronic Privacy Information Center, "this is the first time when campaigns can go beyond directing a mailing to the Hispanic or Jewish community, and instead can figure out what John Smith at 286 Main Street is thinking."

Rotenberg worries that voters, who have no idea of what campaigns know about them, can be unfairly manipulated. Consider this example from TargetPoint's work for Romney this winter. According to Lundry, Romney voters were strongly represented among "country-club Republicans"—well-off folks who care deeply about financial issues that favor their portfolios. But TargetPoint identified another group, one not quite sold on Romney, but susceptible to a pitch on his economic policies. These were people who didn't make as much money as the country-clubbers, but displayed consumer habits similar to the snob set—drove SUVs, went to the theater, bought natural foods. What's more, says Lundry, "they identified with the politics of those they are emulating." Calls to those strivers with messages about Romney's tax policies got results.

On the one hand, that's simply efficient communication, matching voters with a candidate who shares a point of view. But seen another way, the Romney camp has sorted out individuals whose striving makes them vulnerable to a pitch that, at least with their current financial status, is at odds with their economic interests. Would they have been as susceptible if the caller informed them why they were the ones receiving the call? According to our data, you're living beyond your means. Wouldn't you like to vote that way, too?

That's troubling, but the real problem with microtargeting isn't its practice, but the way it embodies how overpolled and overcalculated our elections have become. How can a candidate be great if his or her targets are micro?