“YOU MEAN YOU DON’T DO THIS IN POLITICS?”

In a consulting industry filled with hucksters, Alexander Gage was an awkward salesman. He fidgeted constantly and his sentences shared a tendency to trail off before reaching their destinations. But his discomfiting manner belied a career building political relationships. He spent much of his childhood near Washington, the son of a lobbyist representing E. & J. Gallo Wineries, but by Gage’s teenage years his father had moved the family to Detroit to oversee the vintner’s regional interests. At Christmas, Gage was given the job of driving cases of wine to the houses of Michigan legislators as holiday gifts. He went to a small college in northern Michigan in search of a hockey career; when the time came to give up on that, Gage transferred to the University of Michigan and studied politics. Desperate for work after graduation, he took a job cleaning rooms at the University Motel in Ann Arbor. One day, he read a newspaper reference to a poll conducted by a Detroit firm named Market Opinion Research. Gage looked up the company in the Yellow Pages, sent his résumé, and after months of his follow-up calls was hired as a part-time intern in the summer of 1974.

To the political world, Market Opinion Research was known as the home of Bob Teeter, a young Republican operative who had little training in statistics but had stumbled into polling when it was still, in his words, “a kind of black art.” Teeter was barely out of college when in 1964 he traveled to San Francisco with his father, a small-town Michigan mayor serving as a delegate for favorite-son presidential contender George Romney, and marveled at the spectacle of a national party convention. That year, while working as a teaching assistant and football coach at Adrian College, Teeter spent his summer vacation as an advance man on Romney’s reelection campaign. From there he was hired to manage a southern Michigan congressional campaign, and toyed with rudimentary precinct analysis so he could target prerecorded supportive phone calls from Romney to unaligned voters. In 1967, a twenty-seven-year-old Teeter joined Market Opinion Research and set out to recruit a roster of political clients; by 1970, his analysis had helped to elect a series of midwestern governors.

Polling was the fastest-growing consulting specialty of its day. The academic survey-taking on which the American National Election Studies had been based was slow, and typically occurred after the fact. To the extent that campaigns used polls through the 1960s it was for what pollsters called “benchmarking”: to take stock, usually at the beginning of an election season, of where voters stood—what issues they cared about, their attitudes toward the economy, what they thought of politicians in the news. By the end of the decade, polling grew more nimble, allowing candidates to use short-term, small-scale surveys to inform tactical decisions as a campaign proceeded.

By 1971, Teeter was so celebrated for his ability to translate survey data into practical strategic advice that Richard Nixon pushed aside his pollster and hired the young Michigander for his reelection campaign, which went on to spend more than $1 million on surveys. Teeter decided he had to get serious about statistical research, and to do that he hired Fred Steeper, then a twenty-nine-year-old graduate student at the University of Michigan who had worked at its Institute for Social Research during the 1968 election study. Steeper’s academic credentials impressed Teeter, who wanted to style Market Opinion Research as more analytically rigorous than its peers. Each poll the firm would return to a candidate would include a complete report, sometimes as long as one hundred pages, even when a candidate or campaign officials made clear they had neither the time nor interest in more than a five-page summary. Market Opinion Research’s name occasionally appeared on articles in political science journals challenging academics’ voting models with data from private surveys. Although Teeter and Steeper shared authorship of the journal articles, there was no question how the firm divided its labor. Steeper wrote the scholarly material while Teeter brought in the clients and translated the survey results into strategic advice, which usually required repositioning conservative candidates so they could appeal to centrist ticket-splitters. One client, Illinois governor James Thompson, called Teeter the “Midwestern barometer.”

Gage’s first job at Market Opinion Research mostly amounted to grunt work, but the ambitious young man—with intense blue eyes, broad cheeks, and the parted, slicked-back hair of an early Hitchcock protagonist—thrilled at how it put him at the center of the action. “I had the best of both worlds. I got to follow a brilliant tactician from the back of the room,” he said. “And I got to work with the best pure researcher in the business.” But less than a year after joining Market Opinion Research, Gage was laid off. It was a cyclical business, following the double-time rhythms of the political calendar. Encouraged by the academic orientation of the firm’s leaders, Gage went back to school for his master’s degree. Then, in the fall of 1975, he was summoned by Steeper to help with the polling for Gerald Ford’s presidential campaign. Gage was rehired as a twelve-thousand-dollar-per-year assistant analyst, effectively an apprenticeship to Steeper. Gage would gather the raw data collected by the firm’s Detroit phone room and wait for Steeper to come over to his desk to weight the samples, or watch as Steeper proofread his survey questions. Then Gage would run the numbers to the departments responsible for coding or entering the collected data onto mainframe computers.

This infrastructure necessary for what was known as “survey research” kept newcomers out of the business. Throughout the 1970s, there were three major Republican polling firms, who divided the country by region, an arrangement enforced by gentlemen’s agreements and basic economics. Lance Tarrance of Houston dominated the South; Richard Wirthlin, who had worked on Ronald Reagan’s campaigns for governor, owned California; and Teeter the Midwest and the Northeast. Even as the high cost of long-distance calling made it expensive for any of the three to expand outside their home regions, the lack of competition allowed them to take on ambitious projects.

Gage, who rose to be a vice president of Market Opinion Research, presided over a culture of innovation. In the mid-1970s, the firm perfected the method of rolling nightly samples that became known as “tracking polls,” which for the first time empowered a campaign to measure daily changes in a candidate’s standing and how fast it was rising or falling. Later Gage worked with an Oregon manufacturer to develop handheld “perception analyzers,” which allowed researchers to monitor a viewer’s instantaneous response to words and images by turning a dial to reflect support or disapproval. These so-called dial sessions became the dominant tool for campaigns trying to understand how particular phrases or ideas from speeches and debates resonated with voters. “They weren’t a polling firm,” says Mike Murphy, a Michigan media consultant who worked with Market Opinion Research. “They were a research firm that used polling.”

By the mid-1980s, as the personal computer supplanted the mainframe and the expense of phone calls dropped, polling became just another political consulting service. Falling costs grew the customer base to nearly everyone in politics: even state legislative candidates would routinely commission a poll to help them prepare campaign strategy. Amid the boom, academic training no longer amounted to much. Many of the top Republican polling firms were filled with former operatives who lacked statistical expertise but had an instinct for political strategy and a Rolodex full of potential clients. Polling was a growth industry, and it offered Beltway types an easier lifestyle. Unlike field work—which usually required settling down in a single district for weeks or months—a pollster could stay in his or her office in suburban Virginia, lining up dozens of campaigns each election season and corporate or lobbying clients to occupy the off years.

In 1986, Market Opinion Research did nearly $10 million in business and was one of the thirty largest research companies of any kind in the country. The firm was preparing to serve as lead polling consultant to the presidential campaign of George H. W. Bush, whom Teeter had befriended when Bush was chairman of the Republican National Committee. But even as Teeter’s firm sat atop the political world, it became harder to sustain the ambitious agenda or culture of scholarly inquiry and technical entrepreneurship that gave Market Opinion Research its early character. “Profit and loss in a polling firm is basically a function of volume—how many interviews you do,” says Will Feltus, who worked alongside Gage at the firm. “Research is not a profit center.”

Even Bush’s victory could not ensure Market Opinion Research’s future. Teeter had signed a contract to sell the company after the campaign and cash out of the business, and his employees scrambled to stick together after a new owner swept them out. In 1989, Gage, Steeper, and three other members of senior management formed their own firm, Market Strategies. Political clients may have given the firm its prestige—Steeper was the chief pollster for Bush’s reelection, with Teeter serving as the campaign’s chairman—but the business was increasingly coming from other sectors. For Market Strategies’ first two years, politics and policy delivered half of the firm’s revenue; a decade later, that figure had fallen to one-quarter. Polling in the political world had become a static enterprise; innovations in using data to measure and track personal opinions were all taking place in commercial research.

Gage still enjoyed the scrum of politics, keeping ties to Michigan’s Republican circles and dabbling in election season punditry for Detroit television stations. “He was always thinking about the practical problems facing a campaign,” says Steeper. But by the late 1990s, Gage’s portfolio had become stocked with corporate clients like the American Forest and Paper Association, and the work he most wanted to do looked like a distraction from Market Strategies’ core business. “Gage wanted to come up with these one-off projects that sometimes were profitable and sometimes were not at all profitable,” says Brent Seaborn, who started as an intern in the firm’s atrophying Washington office. “He’s always trying to think of the new idea, which for Market Strategies was a difficult position to be in, because Market Strategies wanted you to fill out a spec sheet, conduct a poll, present the results, and cash the check. They were a very big company and they had a very good routine.”

The engines of the firm’s growth were customer satisfaction and relationship management, interlocking concepts winning fans in the corporate world. Relationships between salespeople and customers were once considered an ineffable part of business, where commerce mingles with courtship. But in the 1980s, some companies started to track all of their interactions with individual customers—every purchase made, service call placed, rebate submitted, product returned—and farmed out the volumes of data to outside firms for analysis. Electrical utilities and health-care providers, in particular, would hire Market Strategies to question their customers about their views of the company (often monthly surveys of thousands of respondents) and use the results to determine managers’ pay. Along the way they marshaled as much information as they could about the people who used their services and how they responded to different stimuli. “The concept is ‘know your customer,’ ” says Gage. “When you touch him, know how you touched him. Did the touch cause him to do something you want him to do?”

This corporate data-hoarding was creating a major imbalance between the information available to commercial marketers and those in politics. A voter file might have a half-dozen categories of information for each person: gender, age, when they last moved, how often they vote. Consumer databases stockpiled hundreds of them, such as whether they had recently taken a cruise or had registered for a hunting license. When he had started in polling in the 1970s, Gage saw commercial marketers trying to learn from their political peers, who had after all pioneered the study of public opinion. Two decades later he was frustrated to find that the roles had clearly flipped. “Back in the seventies and eighties, there was this view that there were really smart new things in campaigns then being adopted by business,” says Gage. “Now all the smart stuff is happening over there.”

AS HE STOOD on the porch of Mackinac Island’s Grand Hotel in September 2001, Dowd recognized the approaching figure of Gage, who at fifty-one maintained a hockey player’s large build even as his white hair thinned into unruly curls sloping off the back of his head. The Michiganders did not know each other well, but they shared a common instinct that campaigns had a lot to learn from consumer research. Dowd’s father had worked in marketing for Dodge and had told his son about how automakers used data-mining services to sift through the hundreds of pieces of information on each possible customer. The data miners would profile individual buying practices and then clump people into clusters based on their tastes and habits. When Gage told Dowd he thought it would be possible to target voters the same way, it immediately clicked. Gage was proposing, in essence, the same technique Dowd’s father had used, but this time with Bush as the product instead of sedans.

Gage left Mackinac Island with Dowd’s enthusiasm ringing in his head. One thing that always depressed political creativity was the uncertainty that anyone would pay for a fresh way of doing things: campaign practices were so hardened, and campaign budgets so predictable, that consultants would rarely invest in new methods until they had seen them successfully demonstrated elsewhere. When he heard the 72-Hour Task Force declare that one of the major impediments to effectively turning out Republican voters was that the party could not locate them, Gage assumed it was a problem the commercial world had already solved. “They just had bad customer files,” Gage says of his party. “They didn’t know who their customers were.”

That year, the Michigan Republican Party had approached Gage and a longtime collaborator, media consultant Fred Wszolek, with a similar challenge. The 2002 midterm elections loomed, poised to feature a busy ballot with races for all statewide offices, including governor. Even though Michigan was a presidential battleground, Republicans felt they always started from way behind when they had to rely on mobilizing votes from friendly precincts. Voters did not register with a party, so every Michigander was effectively an “independent.” In precincts without a clear partisan character, the state party would hire phone banks to identify voters where it could afford to, and rely on coalition partners like the National Rifle Association and Christian Coalition to augment its roster of targets. But in working-class Michigan, this amounted to a fatal math for Republican candidates: there simply weren’t enough voters in non-Democratic precincts or on coalition membership rolls to meet a statewide vote goal. “There were 1.2 million people, and we needed two million votes to win,” says Michigan Republican Party executive director Michael Meyers. “So we all knew we had a problem of ‘how do we get to two million’ and the phones weren’t going to be able to fill in an eight-hundred-thousand gap.”

The flip side of that problem was that the party basically wrote off precincts that voted reliably for Democrats. In sections of populous Macomb County, a historic Democratic edge meant that the state Republican Party gave up directly contacting voters altogether. Gage imagined his job as developing “search-and-rescue” tools, using data to spot sympathetic bodies in a disaster zone and plucking them out to the polls. “We could say, over here on Elm Street in Sterling Heights, Michigan, this person needs a life jacket, and a helicopter comes in and pulls ’em out, takes ’em to the polls,” says Gage. “Otherwise he never would have been touched in any way.”

Such a tool would have immediate application in races around the country. With Dowd’s tentative support, and the RNC’s demonstrated seriousness about its 72-Hour initiative, Gage knew that any targeting breakthrough he could make in time for Bush’s reelection could prove extremely lucrative, and he started to think of the Michigan project as a “beta test” for the bigger contest two years later. He took the five-million-person state voter file and tried to match it to available consumer data from Acxiom, one of a few large commercial data vendors who maintained profiles of nearly every American. He found that 80 percent of Michigan voters matched up against an Acxiom record, which included hundreds of variables, from marital status to pet ownership. In May 2002, Gage commissioned a massive survey to ground those personal details in the political realities of a campaign year. He randomly selected a “test set” of five thousand households from the Acxiom records and called them with around twenty questions about the 2000 election, Michigan politicians (like outgoing governor John Engler and the various figures plotting to replace him), and issues that played a role in state politics (like abortion and the role of unions). Gage had algorithms find patterns linking personal characteristics with political beliefs so that he could use what he knew about his test set to predict how the other five million voters likely approached politics.

In August 2002, the month of Michigan’s primary elections, Gage got a call from Mike Murphy, who had left Michigan to become one of the country’s top ad makers. Murphy had been lead strategist on John McCain’s 2000 race and was now playing a similar role for Mitt Romney, a former venture capitalist and management consultant running to be governor of Massachusetts. Murphy was familiar with the work Gage was doing in Michigan and thought it could be especially useful for Romney, who could build a winning coalition in the liberal state only by sorting through independents and conservative Democrats to identify what he considered the “subtribes” open to crossing over to back a Republican. He summoned Gage to Romney’s headquarters in Cambridge, Massachusetts, nestled between the start-ups that had sprouted from nearby MIT, and suffused with a similar spirit of innovation.

Gage frequently grew restless before giving a presentation, his nervous energy leading him to endlessly tweak his slides—moving images slightly, adjusting the color of text—until the moment he had to show them. There was a particular reason for Gage’s nerves as he waited for his meeting with Romney’s brain trust to begin. The campaign’s ranks were filled with Harvard MBAs and former management consultants, perhaps the world’s foremost PowerPoint artisans, and Gage was pitching them a targeting technique yet to be fully implemented anywhere. Gage gave his primer on what he called “super-segmentation,” explaining how with the latest technology and data it should be possible to merge new consumer records with traditional political information to develop a rich profile of each individual, and then model them to look for once-hidden patterns that could help predict which voters would make the worthiest targets. Once Gage was done, he looked around the table for questions. Alex Dunn, a former high-tech venture capitalist who had left the business world to serve as Romney’s deputy campaign manager, raised his hand. “You mean,” Dunn asked Gage, “you don’t do this in politics?”

THE PUZZLE GAGE believed he had solved was one that had, largely unbeknownst to him, bedeviled analysts for nearly a half century. To social scientists and campaign operatives, breaking the electorate into clusters represented a holy grail; to outsiders, it was an alternately dazzling and dystopic symbol of modernity as democracy entered the computer era. For the close circle around John F. Kennedy, however, the quest to segment the country was a key part of the secret history—a project formally denied at the time and since forgotten to all but those who engineered it—of how they had won the White House.

In his 1964 novel, The 480, Eugene Burdick, a University of California, Berkeley, political scientist who had edited a collection of essays on voting behavior, unfurled a thriller about an unlikely presidential campaign of an American engineer who stumbles into a South Asian border conflict and emerges a national hero back home. Within weeks, John Thatch is being drafted to run for president by Madison Curver, a dashing young Ivy League–educated lawyer who believes he can use a novel statistical system—which splits the electorate into tiny demographic clusters—to shape the unformed Thatch into the perfect Republican candidate to defeat John F. Kennedy’s bid for a second term in 1964. “They went through every poll worth looking at and after a lot of work came up with four hundred and eighty groups which seem to react and vote in the same way,” Curver explains to a Republican power broker. “And now they know a lot about each of those groups, so much, in fact, that they can simulate how the group will act before the group has even heard of an issue.”

Two years earlier, during the Cuban Missile Crisis, Burdick cowrote Fail Safe, about an accidentally triggered nuclear conflict. Four years before that, he had cowritten The Ugly American, describing Western diplomats meddling in Southeast Asia. The 480, too, was grounded in real-life power games: during his brother’s 1960 campaign, which he managed, Robert F. Kennedy had ordered up top secret mathematical simulations of electoral outcomes run on a computer some ominously described as a “people machine.” In the discursive preface to his novel, Burdick described the Kennedy project as the catalyst for a new movement that would replace “the underworld of cigar-chewing pot-bellied officials who mysteriously run ‘the machine.’ ”

“The new underworld is made up of innocent and well-intentioned people who work with slide rules and calculating machines and computers which can retain an almost infinite number of bits of information as well as sort, categorize, and reproduce this information at the press of a button,” Burdick wrote. “This underworld, made up of psychologists, sociologists, pollsters, social survey experts and statisticians, cares little about issues. That is one reason the candidates keep them invisible.”

Even though the academics who had run Kennedy’s “people machine” shared some of the details of their operation with Burdick, he may have overstated the reach of their underworld. In 1959, Columbia sociologist William McPhee had begun using IBM 704 computers to predict how population changes would alter the electorate over a four-year presidential cycle. His tool was the simulation, measuring how strings of variables would interact under different conditions, an approach popular with military planners gaming out battles and engineers eager to see the potential effects of repeated stresses on a structure. When MIT political scientist Ithiel de Sola Pool heard of McPhee’s research, he thought it could be possible to rewrite the program so that it could simulate the potential impact of tactical decisions as candidates made them. McPhee in effect aimed to design a long-term climatological model; Pool wanted to predict the next day’s weather.

Pool, along with McPhee and psychologist Robert Abelson, spent a year compiling sixty-six pre-election polls from the previous decade, which included 130,000 respondents who could each be identified by their race, religion, gender, party, place of residence, and professional and socioeconomic status. With these categories, Pool’s team was able to divide the United States into 480 “voter-types.” In 1959, they conducted a poll to take the temperature of each voter-type on fifty-two “issue-clusters,” to determine if “Eastern, metropolitan, lower-income, white, Catholic, female Democrats” had different attitudes than “Southern, rural, upper-income, white, Protestant, male Independents” toward civil rights, Harry Truman, and which party was best in a crisis. With that framework, Pool believed, the computer could simulate the effect that a change in the issue terrain—a shift in a candidate’s position, or a reordering of voter priorities—would have on broader public opinion and even electoral-college math. “The Presidential election of 1960 was the first in which all the technological prerequisites for our project existed: survey archives, readily available tape-using large-memory computers, and previously developed theories of voter decision,” wrote Pool, Abelson, and research assistant Samuel Popkin.

McPhee brought the nascent technology to Edward Greenfield, a New York businessman and leading reform Democrat, who introduced the academics to officials of the Democratic National Committee and a liberal affiliate, the Democratic Advisory Council, which found a group of donors willing to invest thirty-five thousand dollars in the project. Because universities would not allow their professors to mingle scholarly business with political money, they organized as a private company that could make its research available for sale. Greenfield became the president of the Simulmatics Corporation, housed in a dowdy office in a converted town house near the New York Public Library. He enlisted many of the day’s most prominent political scientists in advisory roles, including Samuel Eldersveld, the Michigan professor who had run field experiments in his own campaigns for office in Ann Arbor, and Harold Lasswell, a protégé of Charles Merriam’s at Chicago.

One of Simulmatics’ academic advisers had already had a formative influence on the logic of unpacking the country into “voter-types.” Paul Lazarsfeld, a colleague of McPhee’s at Columbia, was a Vienna-born sociologist and prominent socialist who had stumbled into the study of elections from his interest in how people made consumer decisions. During the 1930s, as Austria became inhospitable to Jews, Lazarsfeld decided not to return home and resettled at Columbia University and launched an Office of Radio Research. A major project was panel studies that followed households to measure the effects that broadcast advertising had on their shopping habits. In 1940, Americans would be selecting not only dairy items but a president, and Lazarsfeld thought an election would offer a rich venue to study what he called “the psychology of choice.” With colleagues Bernard Berelson and Hazel Gaudet, Lazarsfeld went to Sandusky, Ohio, for seven months before the election, interviewing the same six hundred voters each month to track how their views changed with time.

Lazarsfeld and his colleagues had expected to find that voters chose between candidates the way they picked among brands, individually assessing products and their packaging before making a choice shaped by the influence of advertising propaganda. But in the book they published on the study, The People’s Choice: How the Voter Makes Up His Mind in a Presidential Campaign, Lazarsfeld and his colleagues were surprised that the centralized forces they expected to play a direct role in driving decisions—especially the party organizations, as mediated by television and radio—proved to have a meager impact. Voters arrived at an election season with existing “brand loyalties” to parties, derived from the religious and class context in which they lived and reinforced by the influence of peers. Likening voting to a “group experience,” Lazarsfeld was ready to altogether abandon the analogy to consumer choice that had first drawn him to study electoral politics. “For many voters political preferences may better be considered analogous to cultural tastes—in music, literature, recreational activities, dress, ethics, speech, social behavior,” he, Berelson, and McPhee later wrote. “Both are characterized more by faith than by conviction and by wishful expectation rather than careful prediction of consequences.”

In 1948, the three professors decamped from Columbia to Elmira, New York, with the goal of more properly studying those dynamics. Again they used the method of panel interviews—revisiting the same group of respondents in sequence. As the race between Harry Truman and Thomas Dewey narrowed, the Columbia scholars had Elmira residents explaining their decision making in real time. They saw Democrats who, despite worrying about Truman’s international leadership, returned to their home party as its nominee emphasized economic issues that were at the core of the party’s appeal. (This, as much as faulty polling, explained Truman’s comeback.) In their 1954 book Voting, they labeled this a “reactivation” of latent class loyalties, which had been temporarily weakened as Americans were distracted by the early rumbles of the Cold War.

Throughout the 1950s, the so-called Columbia Studies dueled with the Michigan Studies for primacy as a universal model of voter behavior. The midwesterners eventually won out, thanks to the weight of the regular election year polls backing their 1960 release of The American Voter. But the Columbia model proved psychologically potent, and Pool saw that the notion of different identities exerting competing forces on the human mind lent itself to the same kind of mathematical study as the engineering simulations that measured how a bridge held up against earthquakes or beneath the daily rumble of trucks. “The lifelong Democrat who is a rich, rural, Protestant is under cross pressure. So is the rich urban Catholic,” wrote Pool, Abelson, and Popkin. “The latter’s coreligionists in the city mostly press him toward what was the group’s traditional Democratic affiliation. His wealthy business colleagues press him to a Republican one.” Since their 480 voter-types included all these various identity permutations, the Simulmatics team believed they could measure how each of them would respond if the candidate’s choice of issues put new weight on one pressure, such as emphasizing foreign policy over economic concerns.

On August 11, 1960, nearly one month after the Democratic convention concluded in Los Angeles, Robert F. Kennedy ordered a series of reports from Simulmatics. The campaign’s leadership was most racked by the question of how to deal with his religion in a country that had never before elected a Catholic. Nixon led in summertime polls, and reports of anti-Catholic activity trickled in with growing frequency to John F. Kennedy’s Washington headquarters. Kennedy’s advisers knew they had two strong options: they could ignore religious questions altogether or they could address the subject directly and condemn as bigots those who would let Kennedy’s faith affect their vote. The latter approach, they knew, would take what had been a subterranean issue in the election—the stuff of flyers and local rumor, but ignored by the Kennedy and Nixon campaigns—and force it to the surface. When trying to assess what that would do to the race’s dynamics, none of the campaign strategists had much to go on but instinct.

So Pool ran the numbers for the Democratic National Committee. One of the Simulmatics issue-clusters had compiled respondents’ views when asked how they felt about having a Catholic president. (The typical question: “If your party nominated an otherwise well-qualified man for President, but he happened to be a Catholic, would you vote for him?”) At this point, he had to rely on a little guesswork of his own. Among his 480 voter-types, Pool identified nine significant subsets he thought worth measuring on the faith question, mixing and matching between three party categories (Republicans, Democrats, and Independents) and three religious ones (Protestant, Catholic, and Other). Two groups, Protestant Democrats and Catholic Republicans, he felt, would face the greatest cross-pressure on the question of a Catholic in the White House. Raising the salience of religion in the election would likely push each group in a different direction. (Other groups would also be affected but to a smaller degree, Pool predicted, such as Catholic Democrats, who would be less likely to defect if Kennedy’s faith became an issue, and “Negro and Jewish Republicans,” who could be pushed to vote Democratic if the election became a referendum on bigotry.)

After two weeks, Simulmatics returned its report on “the consequences of embitterment of the religious issue” to Robert F. Kennedy, each copy numbered to guard against leaks of a document considered highly sensitive. Inside was a ranking of thirty-two non-southern states in their likelihood of going for Kennedy if he directly addressed the Catholic question. Eleven states, they projected, would move away from Kennedy on religion alone, totaling 122 electoral votes. But the issue could pull six states, including three out of the four largest in the country, into the Democratic column. Together they were worth 132 electoral votes. “Kennedy today has lost the bulk of the votes he would lose if the election campaign were to be embittered by the issue of anti-Catholicism,” the Simulmatics report stated. “The simulation shows that there has already been a serious defection from Kennedy by Protestant voters. Under these circumstances, it makes no sense to brush the religious issue under the rug. Kennedy has already suffered the disadvantages of the issue even though it is not embittered now—and without receiving compensating advantages inherent in it.” Less than three weeks after his brother received that advice, the Democratic nominee traveled to Houston to talk about his faith before a gathering of ministers. “I believe in an America where religious intolerance will someday end,” Kennedy said, “where there is no Catholic vote, no anti-Catholic vote, no bloc voting of any kind.”

It was an ironic aspiration for Kennedy, since his campaign’s embrace of Simulmatics reflected by far the most serious effort ever to develop a science of bloc voting. Pool did not know if their math had changed the campaign’s calculus around the Houston speech, or if any of the three reports that Simulmatics delivered (others concerned Kennedy’s image, Nixon’s image, and the role of foreign policy as an issue) informed strategic decisions. “They were seen during the campaign by perhaps a dozen to fifteen key decision makers, but they were read intelligently by these talented and literate men,” Pool and his colleagues wrote. Even if their analysis had not shaped campaign plans, the returns in November offered some validation for their technique. Pool calculated that Simulmatics’ state rankings had an 82 percent correlation with the actual vote.

When the existence of the 480 voter-types was reported after the election, the Simulmatics project was heralded as ushering in a new era of space-age politics, as Harper’s described it in a story headlined “The People-Machine.” “This is the A-bomb of the social sciences,” Lasswell said, likening it to the first self-sustaining nuclear chain reaction, conducted by wartime Manhattan Project researchers at an abandoned University of Chicago football stadium. “The breakthrough here is comparable to what happened at Stagg Field.” Newspapers and wire services covered the Harper’s report as news, even as Kennedy’s spokesman denied the existence of the Simulmatics reports altogether. “We did not use the machine,” press secretary Pierre Salinger lied to UPI.

Even though Burdick wrote affectionately about the scholars looking to bring new rigor to smoke-filled political backrooms, The 480 was often read as a cautionary tale about the ability of campaigns to be cynically mechanized at the expense of real people. Pool may have been able to disassemble the electorate into microscopic pieces, but the tools for speaking to voters—national advertising, broadcast television, and speeches covered by metropolitan and regional papers—still existed only on a macro scale. Simulmatics had a good idea of knowing what a small-city, Catholic, Democratic, lower-income woman was likely to think of tax policy, but it offered no guidance to a campaign that wanted to locate members of that category and speak to them directly.

New sources of granular data would make that easier. In 1962, the U.S. Postal Service rolled out its Zone Improvement Plan, which split the country into thirty-six thousand zones and assigned each a five-digit code to help post offices automate their procedures. Soon businesses began using these ZIP codes to presort their catalogs and magazines, and as the direct-mail business boomed these numerical anchors became a useful way to root consumer data in place: they were more compact than counties or towns, the closest thing to quantifying neighborhoods. In 1974, computer scientist Jonathan Robbin used research from customer surveys and block-level Census data to compile 535 demographic variables that could be attached to each ZIP code. Those boundaries had been drawn to aid mail delivery, but Robbin’s Claritas Cluster System used the arbitrary lines to fence people in by their common lifestyle traits. Robbin’s computers assigned each ZIP code to one of forty different clusters, which he gave colorful names, from Furs & Station Wagons (“new money in metropolitan bedroom suburbs”) to Norma Rae–Ville (“lower middle-class milltowns and industrial suburbs, primarily in the South”). The profiles became popular with marketers, from Colgate-Palmolive to Time, who relied on Robbin’s data and vivid social portraiture as a new lens onto America: a way of visualizing their consumers, and then knowing where they lived. Robbin soon became known as the “King of the Zip Codes.”

In 1978, he persuaded Matt Reese, the already legendary Democratic voter contact consultant, that his clustering system could be used for politics. That year, Reese was working on behalf of the United Labor Committee of Missouri to beat back a proposed right-to-work referendum in Missouri, an issue that did not fall neatly along partisan lines. Reese hired Democratic pollster Bill Hamilton to survey 1,367 Missouri voters about their political views, and then used the respondents’ addresses to identify them with one of Claritas’s clusters. Hamilton’s polls found eighteen clusters rich with persuadable voters, building a list of 595,000 targets, and identified the most promising arguments for each. Those who lived in areas Claritas had designated as Grain Belt clusters received Reese’s “pocketbook argument” (which emphasized that right-to-work laws would also hurt those farmers whose customers belonged to unions) while those in Coalburg & Coaltown clusters saw mail with a “status quo” message (pointing out that neighboring right-to-work states had impoverished economies). “The campaign was so carefully targeted,” Hamilton said, “that one resident of Springfield, Missouri, might think that defeating the initiative was the most important thing since the invention of sliced bread. Meanwhile, a few blocks away, someone might not even know the initiative was on the ballot.”

Upon defeating the Missouri referendum, Reese credited the clustering method (which had cost three hundred thousand dollars, one-fifth of the labor committee’s budget) that he called “the new magic.” Reese and Eddie Mahe Jr., a former RNC deputy chairman and leading Republican consultant, joined forces to become bipartisan evangelists for clusters in Washington. The clusters seemed ready-made for a new decade that would, at least in the popular imagination, be remembered for its consumerism. Claritas was selling a map key for decoding the politics of a mobile, postindustrial America where even the middle classes had the means to self-segregate according to their tastes and interests, and people were more likely to identify themselves as consumers than as workers. In director Sidney Lumet’s 1986 film Power, a mercenary political consultant played by Richard Gere—always shuttling by private jet to prop up Latin American strongmen with media manipulation or lifeless domestic campaigns with disingenuously action-packed TV spots—introduces clusters to his clients. “We’ve concluded a high favorable is Pools & Patios: suburban, white-collar, married twenty-five- to forty-nine-year-olds,” Gere’s character tells a New Mexico gubernatorial candidate. “We tailor the mail and phone pitches based on what we know is already bothering them. But the really exciting stuff comes when I work out a simulation model. That’s when you tell me what you’re thinking of saying and I tell you how they’re going to react.”

At the same time, the Times Mirror Company was deep in a three-year statistical project to split the electorate into eleven clusters it called “typology groups.” While the party identity of the 4,244 Americans whom Gallup surveyed played an important role in defining the clusters, the report issued in September 1987—as both parties wrestled with open primaries to find nominees to succeed Ronald Reagan—radically avoided using the language of an ideological continuum to define any of them. “In 1987, the conventional labels of ‘liberal’ and ‘conservative’ are about as relevant as the words ‘Whig’ and ‘Federalist,’ ” the report’s authors declared. “We will divide that electorate into distinct, new constituencies and identify the fundamental outlooks on life and major institutions that animate virtually all American political behavior.” Two of the clusters were distinctly Republican (Enterprisers and Moralists), four Democratic (New Dealers, Sixties Democrats, the Partisan Poor, and the Passive Poor), and two leaning in each direction (Upbeats and Disaffecteds towards the Republicans, Seculars and Followers towards the Democrats). Eleven percent of American adults were found to be fully, and seemingly permanently, detached from politics; Times Mirror called them Bystanders.

And yet even as clusters infiltrated pop culture and public opinion research, they never found a steady place in the political arsenal. Reese and Mahe, who had negotiated an exclusive franchise to market Claritas to political customers, struggled to translate the clusters so they would be easier for Washington hands to grasp. “The names of the groups didn’t resonate politically,” says Mahe. “The descriptions were written with marketing in mind.” He and Reese rewrote profiles so they invoked political behaviors, with terms like conservative and swing voter, and emphasized that the clusters would nonetheless be most useful to campaigns not for visualizing types of voters but for locating them geographically. Still they were unable to sell clustering to any presidential candidates in 1980 or 1984. It was an expensive proposition for a campaign, and only a handful of congressional candidates ever bought it. By the end of the 1980s, Reese and Mahe had let their Claritas franchise expire.

Those who looked closely at the categories of data being used to shape the clusters were shocked to see that they didn’t include markers for race or ethnicity as a demographic variable. A two-page summary of Claritas’s “Downtown Dixie-Style” cluster recorded that its residents, in working-class neighborhoods of cities such as Fayetteville, North Carolina, and Selma, Alabama, were disproportionately devoted consumers of soul records, malt liquor, and Jet (and infrequent overnight campers, Ms. readers, or frozen-yogurt eaters), but never emphasized the one attribute they shared above all else: they were overwhelmingly African-American. “I’d have to sit through all these pitches where they’d say ‘We know this stuff better than you guys, what we do applies to your work, we know how to do it,’ ” says Tom Bonier, who as one of the lead analysts for the National Committee for an Effective Congress was invited to observe marketers’ presentations to the DNC and the party’s campaign committees. “But the big thing they didn’t use, which is sensitive in the corporate world but not in the political world: they don’t use race in their clusters, because to them it’s distasteful whereas in politics it’s an accepted fact.”

When pollsters tested the clusters, by looking at how members of different groups answered political questions, they realized that they didn’t add much to the basic mix of precinct-based targeting and polling to find basic demographic splits. The arbitrary lines around ZIP codes had forced Claritas to effectively compromise when it tried to summarize an area’s character, defining it only by its dominant lifestyle traits. What about all the people who didn’t match the prevailing consumer sensibility? (One political data expert stumbled upon a glaring example of the risk of treating everyone in a cluster as the same. The areas that Claritas called Pools & Patios included significant populations that looked nothing like the people with whom they shared an address: those who lived in the ZIP code in order to clean their neighbors’ pools and patios.) It may be a sensible business decision to put a GapKids in a New Homesteaders neighborhood that is 55 percent childless households, but it would be malpractice for a Republican campaign manager to run a GOTV operation in that same area if it is 55 percent Democratic. “While they may have worked very well for selling Sonys or Toyotas or Mercedeses or Fords, they didn’t work very well for the politics. They didn’t discriminate that well,” says pollster Mark Mellman, who began experimenting with clusters in 1996. “You’re stuck with how they divide people for marketing purposes, and those marketing purposes might not overlay with political purposes. The truth is the political purposes from one election to another may vary.”

YET WHEN THE MICHIGAN REPUBLICAN PARTY approached Gage in 2001 about moving beyond precinct targeting, his first instinct, too, was Claritas. He ordered up the profiles and commissioned a poll to survey two hundred people in each of its clusters. Initially, Gage encountered many of the same shortcomings others had faced before him. When he looked more closely at the components of the clusters, however, Gage saw that all the ingredients should be available to him in raw form. Gage could reassemble that raw data into his own clusters, purpose-built for nothing other than politics. Then, he concluded, he could unshackle the clusters from the arbitrary geography of ZIP codes. Gage could put people into the clusters that really suited them—based on their political views and behavior—regardless of where they lived.

Gage and Seaborn traveled to Little Rock to meet with officials at Acxiom, one of only a few data vendors that claimed to have built a file covering the whole American population. In 1969, an Arkansas school bus manufacturer named Charles Ward had decided to start the company as a way to help the Democratic National Committee use computers to manage its fund-raising lists. As Ward’s staff got better at gathering personal information, their manila punch cards became valuable to customers outside of politics, such as the American Bible Society, which Ward was pleased to learn was a more reliable client than parties and campaigns. Over the years, the company, which took the Acxiom name in 1988, acquired smaller list vendors and cut deals to bring in data from a variety of businesses about their customers: magazine publishers like Rodale, mail-order retailers like Lands’ End, financial institutions like Charles Schwab, pharmaceutical manufacturers like Pfizer, plus hundreds of boutique lists that compiled things as arcane as the types of motors purchased by boat owners. Between 1983 and 2004, with advances in computing, the amount of data Axciom was able to store increased a millionfold, and the company used every byte to fill out one of its personal portraits of an American with a new brushstroke of data. While civil libertarians chafed at companies buying and selling personal consumer information, some of the most valuable details came from government itself. Bureaucratic applications, for gun licenses and construction permits, could say a lot about how much money someone had and how he or she spent it.

With Acxiom’s individual dossiers, Gage realized, he could finally design clusters that were unmoored from geography. The Republican National Committee had assembled the country’s first national voter file in 1990, and had compiled IDs conducted by local parties and campaigns, information shared by coalition partners, and some so-called enhancements from commercial vendors, such as individual phone numbers. (Because that data would come from a magazine subscription record or completed rebate form, it included many numbers unlisted in the phone book.) No one had ever proposed that the RNC buy a list of pool owners, because it was extremely expensive and no one knew how it could possibly be useful in a political campaign.

If Gage merged Acxiom’s personal dossiers with the RNC voter file, he could use that as the base for polling calls, picking names off the file instead of dialing random digits. Then he wouldn’t have to waste polling time asking people how much money they made or what job they had—Axciom already had categories that knew the answers to those questions, or at least predicted them based on information it did have. The polling questions could stick generally to political matters, taking in respondents’ views of issues and personalities in the news. Gage could come away with nearly one thousand variables for each of his respondents, most of them Acxiom’s consumer categories. Gage knew that a lot of those variables would never have any bearing on politics, but it wasn’t possible to tell for sure in advance which ones he would need. He would let the computers find out.

In Little Rock, Gage and Seaborn realized that the logic of political targeting was foreign to Acxiom officials. They were used to dealing with corporate clients who knew their intended audience, and so Acxiom could offer them things like a roster of truck owners, or of Kansas City truck owners who also buy religious reading materials. Bewilderingly, though, Acxiom didn’t have a way to order a list of truck owners who did not read the Bible—or any other combination based on subtraction as opposed to just addition. Perhaps more important, the company’s pricing was structured for commercial customers who wanted to string together a few variables on a limited group of people. Gage would need to cover a whole state’s voting-age population, and demanded many more pieces of data on each of them; he needed all the variables so he could isolate the ones that were helpful in predicting political behavior. Gage and Seaborn negotiated a deal to get access to all the variables Acxiom had available at a flat rate, and left Little Rock with hundreds of them for every Michigan adult so they could start hunting for patterns that would pull them together into targetable groups.

Because his groupings would not be defined by geography, Gage preferred to call them “segments” instead of clusters. He viewed their creation as something of a passive process. It wouldn’t be for political consultants to divine that segments should be formed around socioeconomic status or religious views or participation in party primaries. Algorithms could find the variables that were pulling people together in ways that informed their likelihood of backing Bush’s reelection, whatever they may be. The segments would be only as large as they needed to be to ensure that everybody within one of them belonged there equally. And when the campaign wanted to speak to a segment of voters, there would be no doubt on how to find them. Gage could print out a list for a mail vendor or canvasser, with enough information on every single member of a segment to ensure that he or she was only a phone call, postcard, or door knock away.

IN EARLY 2003, Gage returned to Dowd, this time with PowerPoint slides that referred to the method as “MicroTargeting.” Gage thought this was an improvement over “super-segmentation” (even if a search of news archives revealed that the new term was used elsewhere to describe a medical technique for removing cancerous tumors). Gage had emerged from the 2002 elections with a mixed record in gubernatorial campaigns, but he thought he had a good story to tell—especially as Dowd began plotting an electoral-college strategy for Bush that emphasized some traditionally Democratic states, like West Virginia, Oregon, and Wisconsin. “We’ve got to find out who is more likely to be a Republican,” says Gage. “We know they’re in there, somewhere.”

The previous fall, Michigan Republicans lost the governor’s mansion but won the attorney general’s office for the first time in four decades, while expanding their footprint in both houses of the state legislature. “The down-ticket performance of the party that year was incredible,” says Gage, who credits it to an advanced ability to rouse Republican voters living in Democratic strongholds. In Massachusetts, Romney had entered the last week of his campaign for governor lagging Democrat Shannon O’Brien by five points, according to his internal polls. He ultimately beat her by that margin, helped by major gains among the independents and conservative Democrats who had been Gage’s targets. Gage had not been given a lot of time to develop a new data-driven strategy for Romney before the September primary. He merely wanted to rank-order the state’s nearly two million independents based on their openness to Romney’s appeals, so that the campaign could devote its resources to speaking to its friendliest targets first. Gage invented an index he called “consideration,” a ten-point scale predicting how likely a voter would be to “consider voting for Mitt Romney.” At the same time, Gage modeled those voters’ issue priorities to see which ones should be approached with mail and phone calls emphasizing Romney’s tax plan and which ones should get an education pitch.

As he ran his numbers, one variable popped out. Those who ranked highly on Gage’s consideration index were very likely to be premium cable-TV subscribers. Gage suspected HBO subscriptions were a proxy for other variables—something that neatly packaged the well-to-do and highly educated suburban independents who would warm to Romney’s technocratic approach—but the reason didn’t matter as much as the result. Now instead of trying to pay or recruit the manpower to canvass more than one million potential targets by phone and to judge whether they should receive Romney’s mail, Gage could just send brochures to everyone shown in Acxiom files to be a premium-cable subscriber. Meanwhile, the campaign used the targeting to bring new value to its volunteer operations. Romney, struggling to overcome the almost complete lack of Republican organization in Massachusetts, was able to assign the campaign’s five thousand volunteers to speak to those persuasion targets who lived within their own neighborhoods. “We felt like it was a pretty powerful way to do outreach in these communities, because you had someone who was calling from two blocks away and could talk about the local school,” says Dunn.

Dowd found the results from the Michigan and Massachusetts efforts to be promising enough that he persuaded Bush’s White House advisers to back a trial run of Gage’s approach in Pennsylvania, which had a series of judicial elections in 2003 and would be a key presidential battleground a year later. Gage again attached Acxiom consumer data to the RNC’s voter records, and commissioned a survey of five thousand voters. Then he used algorithms to divide the state’s electorate into more than twenty-five segments based on concentric patterns in voters’ lifestyles and beliefs. Sometimes the Acxiom variables that formed Gage’s segments were apparent: many of the 446,698 “Bible Believers” had shopped at a Christian bookstore, had told a consumer survey that they had religious material in their home, or otherwise resembled those who did. Regardless of how they got into the group, Gage wrote, “despite their higher than average scores on other conservative indicators, social conservative messaging is a must to maximize the vote in this segment.” Other segments—like Pennsylvania’s 243,517 “Dining Room Debaters,” or the 139,586 members of the “Republican Intelligentsia”—were a little less intuitive or self-explanatory. “That’s where we saw the data dance a little for us,” says Adrian Gray, who became the campaign’s voter contact director.

Still, it wasn’t clear that Gage’s computer models were any more effective than traditional targeting methods. After Pennsylvania’s election, the RNC hired Dave Sackett, a pollster for the Tarrance Group, to sample voters and see how accurately the microtargeting had predicted voter behavior. Sackett’s memo argued that Gage’s microtargeting had actually been less efficient than traditional methods at predicting turnout, but had succeeded at finding friendly Democrats and independents and determining which were likely to be pro-life. “That was the piece that was most radical—with microtargeting we’re talking about things that are not necessarily absolutes. You have to trust the statistical inferences in the data set, and that’s a bigger leap of faith,” says Terry Nelson, the campaign’s political director. Because Republicans were used to mailing issue-specific messages to those whose names had been gathered by coalition partners, there was some certainty about why people were on the lists. This was particularly important on cultural and social themes where there was a constant fear of backlash; Gage’s segments appeared, in some ways, to be too refined for the Manichean moral conflict waged on the glossy surface of direct mail. “With a segment said to be eighty percent likely to be pro-life, was it really eighty percent likely?” Nelson asks. “If it was sixty percent likely to be pro-life we’d probably target it differently—because you might not want to mail them with these messages because it would turn them off.”

Furthermore, the price tag for expanding the technique used in Pennsylvania to the entire battleground nationwide was staggering. Gage estimated it could total $3 million, which would cover the cost of acquiring consumer data from Acxiom and other vendors, the lengthy large-scale surveys to benchmark the electorate, and the statistical analysis that would bring them together. In 2000, the RNC and Bush campaign had spent little of their election-year budgets on data, relying on the party’s permanent voter file and the generosity of coalition groups.

Even for a campaign many expected to be the best-funded in history, $3 million was a lot of money, the equivalent of two weeks of very heavy advertising across Pennsylvania. Dowd set out to do something he knew was among the hardest tasks in politics—rewriting a campaign budget to include a line item from a category no one had ever seen before. Dowd had already successfully persuaded Bush’s top political advisers, particularly Karl Rove, to invest in a year of voter contact research. Gage’s technique, as he saw it, was a natural extension of the 72-Hour Task Force’s findings. Now, as Dowd and White House political director Ken Mehlman argued to Rove, the only way to reap the benefits of more efficient field contact was to ruthlessly segment the electorate. “Without the ability to find three people out of ten on a block, we wouldn’t have had the resources,” says Dowd. “We would have had to knock on all ten doors.” Microtargeting, he and Melhman explained, could effectively automate the sort necessary at the beginning of any voter persuasion operation: separating those already on board from those who will never be, and then sifting through the remainder to identify the best candidates to receive mail and phone calls making the case for Bush. This process, Dowd hoped, would help pay for itself.

In many ways, Rove should have been an ideal consumer for microtargeting. He was an unabashed data nerd who had run nonrandomized experiments in 1994 to measure the impact of his mail and phones in Bush’s first gubernatorial campaign. Previous political strategists in the Oval Office had been pollsters, media consultants, or campaign managers; he was the first presidential consigliere to have a background working in voter contact. But Rove’s experience in direct mail also made it hard for him to imagine a contact universe built from anything other than the mix of precinct targeting, coalition rolls, and paid IDs that had proved so effective at illuminating a latent conservative coalition in Democratic Texas. Those lists had been built through manual assembly—a few hundred names from here, a couple thousand from there—whereas Gage conjured his through the alchemy of computer algorithms. “Karl was against this originally, because it was new and different,” says Dowd. “He’s a direct-mail guy and so he thinks he knows that.”

Dowd and Mehlman finally won him over. Someone would now have to contend with consultants certain to feel threatened by Gage’s technique, especially the phone vendors who each year performed millions of dollars’ worth of ID calls that would now be considered superfluous. Coddy Johnson, a Mehlman deputy who had become the campaign’s field director, traveled the country to meet with senior state party officials to tell them that their targeting methods, which usually focused on their strongest precincts, were no longer going to be supported by the RNC. “We’re not mailing them, we’re not going to call ’em, and we’re not buying radio in their neighborhoods,” Johnson would explain. “Here’s how we’re going to do it.” He brought with him PowerPoint slides of the Pennsylvania microtargeting project, and would flip to a Stay-at-Home Independents segment: families in suburban areas that might not be loyal Republicans or even regular voters but that the algorithms showed would be ripe targets for Bush. “That’s who we are targeting,” said Johnson. “We’re not going after the fifty-year-old man who’s voted in every primary and caucus in the last twenty years.”

Dowd, the former Democrat, didn’t care that his decisions would antagonize local Republican officials and the party’s consulting class. “Everybody knew I was going to be bailing quickly, and that I wasn’t going to be part of any new regime,” says Dowd. “After election day I wasn’t hanging around to do a bunch more campaigns.” That message would become clear to anyone who walked by Dowd’s office at Bush’s suburban Virginia headquarters and saw the handwritten sign he affixed to it. “GTT,” it read, the same abbreviation for “Gone to Texas” that Tennesseeans—including many who ended up dying at the Alamo—had scrawled on their homes before fleeing the state to escape their debts after the Panic of 1819. When Dowd envisioned the conflict he was triggering with his new political methods, another nineteenth-century analogy came to mind. “It’s a business,” he says. “You show up with an automobile that runs on gasoline and the horse-and-buggy people go crazy.”

IT WAS CLEAR to anyone closely watching ground-level politics in a battleground state that the Republicans were doing something different in 2004. By the last weekend of the campaign, there were six thousand field-workers walking streets with clipboards (or in some cases primitive handheld digital devices) or manning phone banks. Many of them were in places that had never before seen Republicans hunt for votes, let alone in such a disciplined fashion. When Ryan Johnson, a Bush field organizer responsible for five suburban counties ringing Minneapolis, led canvassers into blue-collar, union-heavy neighborhoods of South St. Paul, he had to reassure them they hadn’t made a wrong turn. “It was like, ‘Why are we going there?’ ” he recalls. The same question was asked on one of Bush’s visits to the state, when he stopped in Duluth, a Democratic stronghold that rarely made Republican target maps let alone earned a few hours of the president’s time. Now, Bush’s strategists could count the number of voters they were trying to reach in Duluth, even if they were a minority, in the hopes of tipping the whole state—“real people who support you behind enemy lines,” as Nelson put it.

In Washington, deputy party chairwoman Maria Cino had converted the fourth floor of RNC headquarters into a command center for 72-Hour operations, filling conference rooms with staffers responsible for booking flights, hotels, and rental cars for ground troops nationwide. Cino had, in effect, created a travel agency, with five people handling arrangements for Ohio alone. When Democrats talked about enlisting volunteers for field, it often involved union members and college kids. The RNC found its volunteer ranks thick with congressional staffers and lobbyists. “This is not a volunteer effort where you can have everybody staying in people’s homes,” she says.

They arrived at their destinations to find clear instructions waiting for them. Bush’s state directors were judged by the weekly spreadsheets they sent back to headquarters listing the number of phone calls and house visits their volunteers had made, and perhaps more important, how many new ones they had recruited. During the summer, headquarters demanded that each state administer a “Test Drive for W.” operation, devoting a day to a full-scale deployment of turnout resources, giving Hazelwood’s team a new array of data on which to judge their state-level personnel. Afterward, she spent weeks leaning on state party officials to enlist county leaders, with local field organizers walking the new recruits through PowerPoint presentations titled “You will be the Margin of Victory,” which outlined the 72-Hour Task Force’s new set of highly regimented best practices. Those identified as turnout targets would typically get three rounds of contact by phone or door knock, the first two in September and October to encourage them to vote early or by mail. Field organizers were able to show the results of experiments demonstrating that, in states permitting it, getting voters to cast a ballot early was more efficient and cost-effective at delivering votes. “They ate it up, and it made them true believers,” says KC Jones, the campaign’s deputy executive director in Minnesota. “They worked harder. A lot of volunteers, if they feel they’re just leaving voice mail after voice mail, they wonder: Am I making any sort of effect on this? Now they knew they were.”

But even as those experimental findings were projected onto the crudely spackled walls of field offices across the country, the microtargeting information was closely held. The term itself was rarely spoken beyond the upper echelons of the campaign, or even outside the team assembled to match the new data with the traditional voter contact program of mail and phones. The Michigan gang relocated to Washington to run the numbers operation, with Seaborn and Meyers building the data models. From his new home in South Carolina, Wszolek wrote the pithy segment names and descriptions that Gage felt helped him explain the data to political operatives in a way they could visualize it.

These issue profiles were conceived to make it easy for operatives to intuitively match messages to specific groups of voters. Minnesotans who received federal farm subsidies were almost certain to get a piece of mail arguing that Bush’s free-trade position would not damage the state’s sugar beet economy as badly as many farmers believed. Moderate Republicans in the Philadelphia suburbs learned about Bush’s support for the Clean Skies Initiative, which the campaign presented as a policy of pragmatic environmentalism. “The universes shrank but they did many more pieces,” says Kevin Shuvalov, who worked on Bush’s mail team. “Once you took all these little clusters and put them together you were basically having an ongoing conversation with the entire universe you have in a state.”

There were so many specialized pieces that Ted Jarrett, who coordinated the mail operation at Bush’s headquarters, stopped looking at the individual orders he sent to vendors. One day, Jarrett got a call from one of the firms producing a mail piece. Did they really want to print only three hundred copies? It was comically microscopic: city council candidates rarely put in mail orders that small. “If there’s one thing I think people don’t get about the Bush election in ’04, it’s this idea that it was a base election and all they were concerned about was the base,” says Meyers. “That was in a sense true but they treated the base as anyone who agreed with them strongly on an issue.”

Dowd had already made a priority of knowing how to rile up a voter who stood with Bush on only a single issue. As he watched 72-Hour Task Force refine the party’s procedures and protocols for reaching their supporters, Dowd had worried that Republicans wouldn’t know what to say to them once they had. So he asked Fred Steeper for what the survey taker called a “mobilization poll,” focused only on one piece of the electorate. What issues or themes could Bush use to push loyal Republicans to the ballot box on his behalf?

A decade earlier, the RNC had ordered up from Steeper another poll of Republicans, this one to explain why they seemed to have deserted George H. W. Bush in his loss to Bill Clinton. Traditionally, polls asked people to process politics analytically, but from what Steeper had witnessed on campaigns it seemed that the issues that really drove elections were the ones that pushed voters emotionally. Steeper decided he would just ask them directly “how pissed off they were,” as he put it—a hunt for what he thought of as their “anger points.” Instead of prompting people to place abortion’s importance as an issue on a five- or seven-point scale, or asking whether Bush’s position had changed their likelihood of voting for him, Steeper’s survey asked “how angry” they were made by the number of abortions that took place annually in the United States.

Steeper’s polls convinced him that one popular take on the elder Bush’s electoral failure—that Republicans had been fractured over social issues—didn’t ring true. After the election, few respondents recalled having been energized by Pat Buchanan’s convention speech declaring a “culture war.” Yet all of Steeper’s questions about unemployment and the economy elicited a strong reaction. Anger points did not need to be only a retrospective tool, he realized; they should help media consultants isolate issues and craft messages around them while a race was still to be won or lost. “The practitioners are always looking for hot buttons!” Steeper says.

When Dowd ordered up the mobilization poll, Steeper thought it was time to hunt for hot buttons. He gathered a sample of people who had identified themselves as Republicans in other surveys done for Bush by Jan van Lohuizen, and felt around for their anger points. Did estate taxes make them angry? What about activist judges, late-term abortions, or trial lawyer fees? Steeper’s polls showed that while September 11 had had a temporary effect on Bush’s broad popularity—the president’s national approval jumped to 86 percent immediately after the attacks—it had an enduring influence on Bush’s base. They were emotionally invested in the “war on terror” that Bush had declared: angry about efforts to repeal the PATRIOT Act, pleased about Saddam Hussein’s removal. As he drafted questionnaires for the benchmarking surveys on which his microtargeting segments would be built, Gage decided to add a battery of questions, inspired by his old mentor Steeper, that would probe for “anger points.” In one of the campaign’s endless sequence of conference calls, another Bush adviser asked why the poll didn’t investigate voters’ emotional responses to the administration’s successes. Gage called these “pleasure points,” a term that reliably made him snicker even as he was pressured to include such questions in his polls. “How pleased and happy are you that Bush has reformed the education system?” he says, derisively. “It was easier to write anger points.”

When funneled into Gage’s microtargeting algorithms, anger and pleasure points helped to turn message development on its head, with acceptable language trickling up from voter contact needs instead of sent down from media consultants trying to translate advertising themes for smaller audiences. The pleasure points questions yielded one unlikely pocket of targets for Bush—his No Child Left Behind school reforms had left a mark on Hispanic women in New Mexico—and helped identify issues for others, such as the environmentally minded Pennsylvania moderates. Weak anger points scores also helped exclude voters from contact on sensitive issues. “We realized some people were pro-life but that talking about it put their religion or their morals on their sleeve and were uncomfortable,” says Todd Olsen, who inherited Rove’s Austin-based firm and worked on the campaign’s direct-mail team. Most important, the new measure of intensity allowed those writing direct-mail pieces to calibrate the emotional potency of their language and imagery. “It helped with the body language of message, the nuance,” says Chris Mottola, a media consultant on Bush’s ad team. “It helped you know how far you can go in terms of rhetoric.”

The last weekend of the campaign, a four-page brochure started arriving in mailboxes across the country. By most aesthetic and moral standards of the time, this mailer went too far. The front flap featured a collage of September 12, 2001, front pages of the Des Moines Register and Orlando Sentinel, newspapers intentionally chosen to represent battleground states, with an image of the World Trade Center in the midst of the previous day’s plane attacks. “How can John Kerry lead America in a time of war?” it asked. On the back, in hazy chiaroscuro, was Osama bin Laden, making eye contact with every reader. The head shot may have been the single most familiar facial image in news coverage from 2001 through 2004, but it had never entered the official visual language of the campaign. An informal prohibition on explicit depiction of the September 11 events had been accepted by both sides, and studiously enforced by elite opinion. When Bush had run an ad in March showing the World Trade Center wreckage and a firefighter carrying a body from the site, victims’ families hosted press conferences to protest. A draft of the proposed mail piece sat around untouched for months amid a heated debate over its propriety. Many of Bush’s advisers argued that such a graphic appeal could force swing voters to recoil from Bush, goaded by a media eager to claim that the White House was preying on a climate of fear it had helped to nurture.

But the microtargeting scores that Gage had built suggested that swing voters were precisely the right audience for the most visceral appeal of Bush’s candidacy. The targets would be ones Gage thought of as a persuasion reach: they were very likely to vote, but according to Gage’s numbers there was a less than 50 percent chance they would pick Bush. The anger points questions revealed this group of voters to be emotionally sensitive to the war on terror. Bush’s mail team knew they were setting off a bomb, but they had placed it so precisely that they could be fairly confident of its blast radius; they had drawn the circle so tightly there would be no downside. “If we lost them it wouldn’t be a big deal,” says Jarrett, the national director of voter contact. “This was to pull votes from Kerry.”

After election day, little was said about the bin Laden mailer or the complex research that had given Bush’s mail team the confidence to take on such a touchy subject before an unlikely audience. The popular storyline among Democrats looking to explain Bush’s victory credited the campaign’s use of quiet communication to rile its base over issues like gay rights. Yet even though Bush allies pushed to get anti-gay-marriage initiatives on the ballot in key states, the campaign rarely found those fruitful topics for direct contact. “That’s been the big myth in that campaign: that we drove on the social issues which we really didn’t,” says Shuvalov. “Because Kerry was not going to challenge us on them.”

Gage, whose name had barely appeared in newspapers during the campaign, was profiled by the Washington Post in the weeks after the election. The story noted his breakthrough as being able to calculate “Coors beer and bourbon drinkers skewing Republican, brandy and cognac drinkers tilting Democratic.” Those who worked on the microtargeting project knew that this detail, even if in parts technically accurate, completely misrepresented their work. They laughed at the fiction that the race had been won and lost through mastery of liquor store transactions. “The Bush campaign people were paranoid. As we talked about this publicly we had to make stuff up,” says Wszolek. “We had to give examples that were completely phony. It would drive the Democrats completely loony. We just wanted to keep them from knowing how accessible it is.”