The Information Arms Race
A dozen graduate students visiting from New York are gathered in Alex Gage’s office, just outside Washington, D.C. They’ve come to learn about the business side of political campaigning. Gage has thinning, swept-back blond hair and the clear blue eyes of a Viking, and wears a collared shirt beneath a fleece vest unzipped to his belly. His voice is gravelly, his demeanor frank as he describes his work in the most commercial terms: A candidate is the “product,” a voter the “customer.” A young woman opens up her laptop, goes online and starts clicking away—a common-enough act, but one that Gage understands as something more. He jerks his finger at her. “You see what she’s doing? She’s giving off data exhaust.” The digitized record of a virtual life, data exhaust is composed of valuable bits of information about our habits and preferences. And it may help determine the country’s next president.
Gage is the leading Republican strategist in a field called predictive analytics, or microtargeting. He worked with George W. Bush’s winning campaign in 2004 and is now working with Mitt Romney, whom he hopes to fashion into the first presidential “data-driven candidate.” Gage praises the Democrats’ success in using microtargeting to build the Obama coalition in 2008. Ken Strasma, Obama’s national targeting director for that campaign and the founder of the consulting company Strategic Telemetry, says 2008 was the tipping point when microtargeting became a must-have tool for any political contender. Now, a data strategy is as necessary as a television strategy, and each campaign is “an information arms race,” explains Alex Lundry, who works for Gage’s company, TargetPoint.
Every time you “like” a friend’s Facebook status, sign up for a Nordstrom credit card, rate a Netflix movie, order a magazine subscription, or merely click on a website, you’re leaving behind a trail of data exhaust that up until recently has been like so many discarded Styrofoam cups lying along the information superhighway. The rise of microtargeting is a function of new logarithms—and computers fast enough to process them—that are able to capture all this trash and turn it into gold. Over the years, the data-mining industry has become adept at recycling information about the websites we visit and the products we buy. Rumor has it that some high-end companies, including Omaha Steaks, can now make more money by selling their customer pedigrees to data-mining firms than they can from selling their product.
Both parties use this data to craft hyperspecific messages to peel away voters from their opponents. With access to a database containing information on roughly 200 million individuals—not just email addresses, real people with names—Gage tries to predict which candidate they will vote for and what kind of message will help make up their minds.
“Let’s say I want to run for a Senate seat in Michigan,” he explains. “So I say, ‘How many customers are out there?’” Out of 7.5 million registered voters in the state, he decides that 4.2 million might cast a ballot. “So I need to figure out how to get on Election Day, say, 2.1 million voters to go and buy my product, to pull my lever. So you probably say who are these people? How do I get to them? That’s where we come in.” He’ll take a random sample of 200,000 voters from across the state and overlay census tract and consumer information for each person. Then they’ll conduct a microsurvey of about 10,000 people from the list to understand each voter’s attitudes. “Is he a Democrat or Republican? A protectionist or a free trader? How do they feel about guns? How do they feel about abortion? How do they feel about the role of government?”
By overlaying information, profiles emerge. Simply knowing someone’s age and gender doesn’t predict much. For any 59-year-old male there’s roughly a 54 percent chance he’s Republican (Gage didn’t want to reveal the exact stats, so he gave me “numbers rooted in truth”). But knowing that he’s a white businessman who lives in Louisiana, drives an SUV, and owns a Bible means there’s something like a 94 percent chance he’s Republican. In contrast, take a 59-year-old black businessman who lives in California, drives an SUV, and likes modern art. A man who fits this profile only has a 12 percent chance of being Republican.
The first profile is someone who might be good to reach out to for fundraising early on or with a reminder to vote just before Election Day. But he’s likely already in the Republican camp. Microtargeting can be used to rally members of a candidate’s base to ensure they turn up at the polls. But it’s also used to zero in on the issues that can persuade an individual swing voter. In 2004, Gage helped identify the middle-class, middle-age Hispanic women concerned about education—and therefore receptive to a message about Bush’s role in No Child Left Behind—whose votes helped Bush win New Mexico, and the Ohio “security moms” who were particularly worried about terrorism.
Gage has found that “anger points,” issues that provoke the most ire, are the most effective way to boost turnout. So his surveys are designed to find which issues—taxes, abortion, guns—fit each profile in order to get the most persuasive message to the right voter. He claims he can deliver a 3 percent to 4 percent bump for his clients. The margins may seem small, but the impact isn’t; presidential races boil down to close battles in a dozen or so states. Every vote counts.
Microtargeting helps both parties identify the issues they can use to peel away voters. Ken Strasma likes the SUV as an example, too. At first, the average SUV driver wouldn’t seem to care about gas prices or the environment. But find out she has four children and a two-hour commute and that changes: The kids explain the SUV; the commute tells you she’s susceptible to a political message about gas prices and that a radio ad might be better than direct mail or an internet approach.
The devil is in the aggregation of the details. “There started to emerge a few years ago,” says Strasma, “a subset among Christian evangelical voters who viewed environmentalism as an important issue. They saw it as a biblical imperative.” So, rather than banging their heads against the wall about trying to fight the abortion fight, Strasma says the Democrats could target those voters with the message about a pro-environment stance.
The problem with microtargeting, says Robert Shapiro, a political scientist at Columbia University who specializes in the increasing polarization of American political life, is that it’s an intensification of politics as usual. He says microtargeting strategies, along with innovations such as cognitive scientists monitoring a person’s brain waves to find the most effective talking points, increase the influence of messaging. “That is,” he says, “they’re basically trying to create and mobilize more of an ideological base in both parties.” The effect could increase the bitter partisanship that already exists, making compromise—on how to tackle the federal deficit, for example—difficult. Shapiro argues that partisanship has changed the way people process facts, and points to the growing number of people who believe that weapons of mass destruction were actually found in Iraq.
Ken Strasma admits that microtargeting does probably amplify the ideologically extreme cases in both parties. “On the other hand,” he says, “the issues and messages can also mean someone who is frustrated by partisan gridlock, frustrated by extremism, and who wants someone who’s going to campaign and govern in a new way. And that was definitely one of the successful messages in the Obama campaign.” In other words, “angry” and “partisan” are not synonymous. “Change” was a successfully targeted message.
Gage disagrees that microtargeting has altered the nature of politics. “It’s basic electoral politics as it always has been but with new technology,” he says. “You find ’em, you vote ’em, and you count ’em. Abraham Lincoln did that. Everyone does that.”
That’s true to an extent, according to Lawrence Jacobs, a University of Minnesota political scientist working on a book about how presidents since Kennedy have used polling to shape messages that would appeal to critical subgroups of voters, shifting the debate from policy to personality—a cruder form of microtargeting. “The effort to selectively mobilize supporters is as old as politics.” The founding fathers did it in the debate over ratifying the Constitution. Political candidates did it in ancient Athens. But the “potency and the consequence of targeting” has since changed for two reasons. First, more sophisticated technology makes it much more precise. The advent of computer-based analysis in the 1960s was a breakthrough that allowed politicians to break down regions by precinct. In the 1990s, neighborhoods could be analyzed block by block and by individual. “Now you’re able to kind of tailor a message to a person, knowing what that person’s interests and predispositions are. That’s kind of a revolutionary step,” he says.
“The second part,” Jacobs continues, “is a political system in which the incentives profoundly changed over the last four decades.” Until the 1970s, the two major parties were dominated by strong leaders, and the focus was on winning elections, which meant building the largest possible coalition. But reforms to limit the power of party bosses over the nomination process shifted influence to those who showed up at caucuses and primaries—often single-issue activists and the more ideologically extreme members of each party.
“It was a good intention, which was to democratize American politics and move away from smoke-filled rooms and all that. But the actual effect was that the nomination process became controlled by these very small, very ideological groups in both parties.” Rather than voters shaping the candidates’ platforms, the incentives push campaigns to seek out peripheral issues, building a narrow coalition of voters without compro- mising the predetermined policy agenda of its backers.
In theory, anything that gets more people to the polls should be good for a democracy. But microtargeting could also be used to suppress turnout; Republicans sending messages about Democratic support for gay marriage or gun control to anti-gay-rights or gun-owning Democrats, for example, could diminish their interest in voting. And because microtargeting selects for the most persuadable voter, which often means the least-attentive voter, Jacobs believes it can be used to distract and mislead. “Microtargeting is not simply about turning out voters anymore,” he says. “But actually manipulating and confusing voters to the extent of propagating misinformation.” As an example he cites the persistent belief that Saddam Hussein helped plan the 9/11 attacks. “About 31 percent of Americans, almost entirely Republican, believe that,” he says. That was part of a deliberate messaging strategy.
In 2012, both parties will be looking to persuade voters who went with Obama in 2008 but are now disaffected. And both parties will likely use microtargeted messages to frighten them, Jacobs says.
Strasma is more optimistic. Microtargeting can help a candidate craft his message and speak to each voter on the issues he or she most cares about, he says.
But it is important, Strasma cautions, for voters to know that microtargeting is being used. He compares the situation to retailers using store discount cards to track customers’ shopping habits so they can send them special offers to maximize profits. “Being an informed consumer depends on having some understanding of that. The same is true for politics. You get a piece of mail and think, ‘Wow, this candidate is focusing on my top-three issues and really seems to care about the same things I care about.’ You should realize that it’s quite possible that your neighbor across the street got an entirely different piece of mail.”