Photograph: KTSDESIGN
Data Mining

Cruz-Connected Data Miner Aims to Get Inside U.S. Voters' Heads

Funded and promoted by secretive hedge-funder Robert Mercer, employed by Ted Cruz's campaign, Cambridge Analytica promises a transformative new approach to identifying voters. Does it promise too much?

I had spent about a half hour in the headquarters of Cambridge Analytica before I realized why my arrival at the tech firm—in a low-rise office building lodged inconspicuously on a dead-end service alley behind the Japanese Embassy, a block from Piccadilly—had been greeted so warmly. The company’s staff had been primed to attend to a psychological abnormality. “You’re in the top half percent of openness for everyone in Pennsylvania, which is quite stark,” marveled analytics director Alexander Tayler. “This is like a three-sigma event, to find someone that far out!”

I had traveled to London because the mysterious capability Tayler showcased with his welcome—to segment the American electorate not only by ideological predisposition but also by individual psychological characteristics—amounts to the most audacious new analytical innovation foisted on American politics this year. Of the thousands of companies making money off the presidential election, Cambridge Analytica also boasts the most compelling patrimony. Despite being a new arrival to American politics, it is an offshoot of an established British firm, SCL, which made its name advising governments and militaries on what it called “psy ops.” Even when it rebranded itself to compete for American political clients, as a specialist in profiling individuals’ personality characteristics in order to better tailor messages, SCL, curiously, chose not to obscure its foreign heritage. Instead, the company swathed itself in an exaggerated Anglophilia, calling itself Cambridge Analytica. The firm initially kept a low profile in the U.S., but was outed as a vendor to Ted Cruz’s campaign in July, when Politico reported that its investors included the family of Robert Mercer, a secretive hedge-fund investor who also happened to be one of the most generous donors to Cruz’s super-PAC. Mercer wasn’t talking publicly about his investment, but I could understand why someone with a background as a coder and a willingness to spend down his net worth, millions at a time, to support conservative causes was drawn to the company. Amid the appetite for high-tech innovations in Republican campaigns, Cambridge Analytica could claim it was not merely matching the Democrats—it was going where the opposition had not.

In my years of writing about the use of data in politics, other firms had developed a variety of profiles of me using public and private sources; each time I was told what campaigners who had never interacted with me assumed about my identity or attitudes. I had seen statistical models anticipating my likelihood of casting a ballot in various upcoming elections, of being married, of owning a gun. In an ideal world, campaigns would have access to this information from a less speculative source: my telling a canvasser how I intended to vote, a warranty form on which I had identified a spouse, or the inclusion of my name on a publicly available list of licensed hunters or private membership rolls shared by the National Rifle Association. But if those were not available, or out of date, statistical models could make inferences from facts that were available. Algorithms could trawl through as many of thousands of different variables—my past political behavior, consumer choices, the demographic composition of the Philadelphia neighborhood in which I was registered—and isolate the interaction of a few that would determine how much I statistically resembled people who were known to be married, or to own a gun. Campaigns could then communicate with me based on those calculated likelihoods. 

Cambridge Analytica’s assessment differed in one crucial way: The firm promised to tell me things I might not even know about myself. It claimed to predict where I would fall on the five-factor personality model, which won widespread adoption by psychologists starting in the 1980s as a standard inventory of universal traits known as “the Big Five.” According to Cambridge Analytica, I fell in the middle range for extroversion. When it came to neuroticism, I was in the seventieth percentile. I scored very low on conscientiousness and agreeableness, a combination which, when paired with my high openness, defined my individualism. “Fanciful/Imaginative Types are unconventional nonconformists who pride themselves on being different from others,” read a potted description attached to my numerical assessment. “In extreme cases they might be regarded as eccentric, but in most cases they are perceived by others as complex, well-read, imaginative and industrious.”

Of all the microtargeting profiles of myself I had seen, none had flattered my self-concept like this one. Its predictions already seemed more plausible than those of the Democratic data warehouse that had my religion pegged as Lutheran—a prediction likely tethered to the only slightly less dubious profiling of my ethnicity as German. (I presume that was the result of an algorithm which heavily weighted my surname’s Teutonic build and the preponderance of white people in my census tract who report German ancestry.)

“And so you can imagine with that information as well as all the other information about your political orientation it is possible to put you with the like-minded people to receive a very, very specific communication,” Cambridge CEO Alexander Nix told me.

The next day, I met with two of the employees Nix identified as the firm’s “message people” to understand what that communication might look like. Tim Glister is a former copywriter and one-time literary agent from Newcastle; Harris MacLeod a Nova Scotian who worked as a political journalist in Ottawa. Both spent much of 2014 working for Cambridge Analytica’s marquee American clients. Harris worked for John Bolton’s super-PAC, which was attempting to bring more attention to national-security issues in three select Senate races ahead of a prospective presidential campaign by the hawkish former UN ambassador. Glister was dispatched to North Carolina, where he was tasked with helping the state Republican party on behalf of Thom Tillis’s ultimately successful campaign to defeat Senator Kay Hagan. “I was English enough to be an entertaining curiosity,” he said.

I showed them my personality scores, and they nodded knowledgeably as I read through the description. Cambridge Analytica’s standard client pitch uses the example of gun rights to illustrate how a campaign could shape its messaging to voters based on assumptions about their individual personalities. “If you are extroverted and conscientious, you understand the argument about the Second Amendment in terms of the responsibilities that accompany it,” Nix says. For “closed and agreeable people, this is really about understanding these peoples’ need for tradition and value and community and family.” That made enough sense, but most campaigns were not trying to change voters’ views on fundamental issues like gun control, or their intensity of feeling about them. I had a harder time imagining how a campaign would tailor its messaging about a candidate based on the awareness that I was likely to be an “Individualistic and Imaginative Type.”

“If we were marketing Ted Cruz to you, we would emphasize the fact that he’s in a new generation of Republican leaders, that he’s not afraid to stand out from the crowd,” MacLeod said. “He’s got new ideas, fresh ideas, innovative ideas—you’ll be able to file your taxes on a postcard, whatever, abolish the IRS.”

“Pushing that even further, if you are low-conscientiousness,” Glister added, “it’s about being the Washington outsider, blowing through the received way of doing things.” 

“You’re not the type of person where we’d say, Well, we have a ten-point plan,” he went on. “That’s not going to resonate with you. It would be ‘everything is bigger in Texas’—it would be bold visuals, bold statements.” 

As he made his way through an office in which the median attire fell somewhere between ad agency and hackathon, Nix—a bespectacled, lanky man with the angular affect of the comedian Stephen Merchant—stood out among his own employees for carrying himself like a City banker. Nix had worked as a financial analyst before joining Cambridge Analytica’s parent company, SCL, a little more than a decade ago as a director helping to guide its global expansion. Since then, he has been the face of Cambridge Analytica to prospective clients in the U.S., and has a salesman’s slick patter even if he often seems to be speaking a foreign language. When he deployed phrases like “the guys who’ve got guns under their pillows because they don’t want to get burgled”—describing a potential Second Amendment constituency—I wondered how he would be received in often-insular campaign conference rooms, where I knew difference was not always prized. “He has no American political sense whatsoever,” an experienced Republican consultant who has met with Nix told me.

Indeed, the fact that this London-based firm was already working for a well-funded American presidential campaign represented a dramatic inversion of the natural order. Political consulting, after all, has been among of the most durable of the United States’ exports. The world’s political class often maintains a conflicted view of American professionals: envious of the sophistication of their extravagant electoral pageants; resentful of the idea that the mere fact of their passport gives them standing to direct strategy in countries they are often visiting for the first time. The idea of the dispassionate political legionnaire has been a trope familiar enough to Hollywood at least since 1986, when Sidney Lumet’s Power introduced its protagonist (played by Richard Gere) as a consultant who finds himself an unlikely arriviste in a Latin American political scrum. Little has changed: This fall, Sandra Bullock plays an operative who parachutes in to help a Bolivian presidential candidate in Our Brand Is Crisis, based on the 2005 documentary of the same name about the Bolivian adventures of James Carville, Stan Greenberg, and Bob Shrum. 

In each case, the Americans play the same role: jet-set hirelings, alternately ruthless and naïve, imposing cookie-cutter strategies on campaigns worldwide. “Politics in country after country has become as similar as Starbucks—and about as surprising. The assumption underpinning the international consultancy business is that the same principles apply everywhere, that a foreign country is just like another swing state, just like Ohio,” James Harding wrote in Alpha Dogshis book about American consultants who pioneered cross-border political consulting. “The battle is ever more for hearts, not minds: America’s winning and irresistible formula has been to repackage an intellectual argument inside an emotional appeal.”

When he launched Strategic Communication Laboratories in 1993, Nigel Oakes embraced the image of the worldly mercenary, albeit with an English accent. If the Americans abroad justified their emotional appeals as the result of instincts honed in the world’s most intense political environment, Oakes—a former Monte Carlo TV producer who had gained notoriety when he dated the daughter of the Queen’s cousin—marketed the rigor of Britain’s ancient universities. Oakes, who had also worked for the advertising agency Saatchi & Saatchi, argued that traditional advertising was incapable of effecting the type of mass opinion shifts necessary for social change. Instead, he marshaled scholarly research, much of it from psychologists and anthropologists filtered through the Behavioral Dynamics Institute, an affiliated non-profit that Oakes had established as “a research facility for understanding group behaviour.”

Over the course of the 1990s, Oakes applied upon this academic foundation a gloss of new technology. When he advised Indonesian President Abdurrahman Wahid, the country’s first credibly elected leader, Oakes set up a control room filled with computers, oversized monitors, and TV screens dubbed the Jakarta International Media Research Centre. A photograph of the high-tech installation still appears in the company’s pitch presentation without any description of what exactly Oakes’s team was monitoring on those screens. “It was fun and exciting, but also a bit dangerous because everything he did was so secretive,” an Indonesian who worked for Oakes told The Independent in 2000. “We didn't know the purpose of it all, we just did what he asked. We called him Mr. Bond because he is English, and because he is such a mystery.”

In its early years, Oakes’s company worked with the Conservative Party, according to an executive, but after the 1997 elections his firm pulled back from British politics. SCL tacticians had found that they were unable to maintain the same aloof sensibility that their London-based workforce brought to overseas politics. “It’s difficult to ask people in their own country to work on a campaign they don’t support,” says Nix, who joined fellow Etonian Oakes’s firm in 2003 as a director guiding its expansion. Over the next decade, SCL continued its international work, steering as many as ten campaigns for prime minister or president annually in countries as far-flung as South Africa, Argentina, Thailand, and Italy. “Literally everywhere around the world from first world to developing countries, from quite modern elections to really quiet complex rural communities that have very limited technologies and a lot more field work and grassroots consulting and so forth,” Nix says. “Everywhere but the United States.”

In 2010, Nix traveled to the United States to learn more about the country’s political sector but left discouraged by the insularity of the consulting industry. Instead, seeing military and security budgets boosted by post-September 11 spending, SCL expanded the range of services it could sell to governments. With former British Defence Minister Sir Geoffrey Pattie installed as the firm’s chairman, SCL in 2005 relaunched as a specialist in “psychological warfare,” designing communication that could be deployed to demotivate military opponents or influence civilian populations in conflict areas. A few years later, the firm launched SCL Social, a non-profit division that focused on changing behavior for humanitarian ends, still under the rubric of what its employees casually call “psy ops.” For both NGO and governmental clients, SCL analysts developed programs to increase condom use across the Caribbean, improve the effectiveness of tsunami warnings delivered by text message, and guide a Unicef project to discourage child marriage in South Sudan. “We used to be in the business of mindbending for political purposes, but now we are in the business of saving lives,” Oakes said. 

After the 2012 election, Nix found an American marketplace far more receptive to his entreaties. The overseas work in conflict zones amounted to a promising calling card, a new comparative advantage over entrenched American political firms.“This is really trying to use psychology to understand why hostile audiences do what they do, and to use this methodology to deconstruct that behavior and then use communication to try and change attitudes and ultimately behavior,” Nix says. “Persuading somebody to vote in a certain way,” he goes on, “is really very similar to persuading 14- to 25-year-old boys in Indonesia to not join Al Qaeda.” 

Among those apparently motivated by this pitch was the family of Robert Mercer, whom Forbes has repeatedly ranked among the country’s highest-earning hedge-fund managers. Mercer made his fortune as co-CEO of Renaissance Technologies, a Long Island fund unusually proud of its appreciation for academic credentials over investing experience. In that respect, Mercer was a typical Renaissance man, having worked at IBM’s Thomas J. Watson Center before entering finance, trying to teach computers to recognize human language. He remains uncommonly discreet: articles about him invariably note that few pictures of him exist and perhaps the only known detail about his personal life—that he is a toy-train hobbyist—emerged when he filed a $2 million suit against a model-railroad builder for overcharging him on a project. 

Mercer is the type of wealthy citizen most emboldened by the Citizens United decision, unlikely to engage in the greasy work of bundling others’ contributions but happy to directly sponsor his preferred causes. Along with his wife, Diana, Mercer gave just shy of $10 million to Republican campaigns and committees in 2014, making him the country’s fourth-largest individual donor, according to the Center for Responsive Politics. Yet his public appearances are few, and even then detached from his political objectives. "What I am is simply a computer programmer," Mercer said in a speech last year to the Association for Computational Linguistics, after receiving a lifetime achievement award. "I've taken great pleasure in programs that do remarkable things."

Mercer emerged as a major financial force in conservative politics just as it became an ideal sphere in which to indulge his pursuit for interesting technical programs. In 2013, SCL Elections spun off its American operations into a district entity, Cambridge Analytica, while also also removing the phrase “soccer mums” that appears in SCL Elections materials. “We want to look, feel like an American company,” says Nix.

Just as significantly, it then became a Republican company. The Mercers have repackaged Cambridge Analytica as an ingenious cog in the GOP party machinery that can crank out votes using methods unavailable to Democrats. Earlier this fall, the firm moved its Washington office to Alexandria’s Old Town, the seat of the Republican consulting sector. “It’s pretty clear that in America you’ve got to pick a side in this business,” says Nix. “It’s not really our decision. The market makes our decision.”

Nix crossed the Atlantic as though bearing a passel of Enlightenment-era insights about the ways the political profession had failed to appreciate the complexities of personal experience. “Your behavior is driven by your personality and actually the more you can understand about people’s personality as psychological drivers, the more you can actually start to really tap in to why and how they make their decisions,” says Nix. “We call this behavioral microtargeting and this is really our secret sauce, if you like. This is what we’re bringing to America.”

Cambridge Analytica’s trophy product is “psychographic profiles” of every potential voter in the U.S. interwoven with more conventional political data. The emphasis on psychology helps to differentiate the Brits from other companies that specialized in “microtargeting,” a catch-all term typically used to describe any analysis that uses statistical modeling to predict voter intent at the individual level. Such models predicting an individual’s attitudes or behavior are typically situational—many voters’ likelihood of casting a ballot dropped off significantly from 2012 to 2014, after all, and their odds of supporting a Republican might change if the choice shifted from Mitt Romney to Scott Brown. Nix offered to layer atop those predictions of political behavior an assessment of innate attributes like extroversion that were unlikely to change with the electoral calendar. 

Oakes may have cautioned against “focusing on individuals or audience segments,” but Nix was doing exactly that, albeit disingenuously characterizing how other firms sorted through the electorate in order to make his approach appear more humanistic. Conventional microtargeting, he argues, “of course has to be flawed because you’re making this presumption that all women think the same simply because they are women, or all African Americans likewise because of their skin color or whichever demographic you want to choose, which is archaic, really.”

For a generation, pollsters have been attempting to move past simple demography by accounting for the psychological baggage that voters tote with them from one election to the next. As early as 1984, Ronald Reagan strategist Richard B. Wirthlin included batteries of abstract questions in his surveys, looking to identify relationships between what he called voters’ “values” and the messages that could shift their political preferences. For Bill Clinton’s 1996 reelection, Mark Penn conducted what he branded a “NeuroPersonality Poll.”  

What has changed is the range of data available about those voters whom a pollster didn’t reach, which in the 1980s and 1990s included little beyond what was on the public electoral rolls. Campaigns already used algorithms to infer political and demographic attributes about voters they couldn’t contact directly; why couldn’t those same statistical models predict innate psychological characteristics, as well? While Wirthlin and Penn could isolate archetypes—Penn christened “soccer moms” as part of such an exercise—they lacked the data about the electorate to identify the individuals who comprised each of the clusters so they could be contacted directly. By 2012, such data—and the statistical tools necessary to sift through 200 million potential American voters along its terms—were widespread.

SCL began hiring Ph.Ds, many of them from the University of Cambridge, from fields where manipulating large data sets is routine. As a result, it is probably the only political consulting firm whose employee bios delineate sub-disciplines within the hard sciences, separating those who studied astrophysics from theoretical physics, condensed matter physics, theoretical solid state physics. Very few of them have ever before stepped near a campaign office, and they demonstrate little familiarity with political life in the United States. In casual conversations about the subject, Cambridge Analytica employees speak of “the Tea Party” and “conservative Christians” as distant oddities, in a manner typical of educated Europeans.

“For most people who work here, it’s more about being professional and treating your clients as your clients,” says Tayler, whose Cambridge doctorate is in nuclear magnetic resonance, and who previously worked for the oil-services company Schlumberger. “They have their beliefs and you’re going to do your best for them, because that’s what a professional does.”

The associations with Cambridge serve to situate the company as heir to a specific scientific tradition. Starting in 1886, the University of Cambridge’s Cavendish Laboratory—where James Watson and Francis Crick would later map the structure of DNA—became the site of the first lab devoted to a field known, alternately, as psychometrics and psychophysics. What one researcher who had lectured at Cambridge described as “the quantitative estimate of some of the less commonly and less easily measured of the human faculties” was initially done with weights and forceps, hallmarks of the eugenics movement. 

In the century’s second half, researchers found less ominous ways to sort individuals into categories based on innate psychological attributes, leeching out insights through confessional interviews. In the years after World War II, Isabel Briggs Myers isolated 172 questions that would reveal the internal architecture of an individual’s mind. By the 1980s, the Myers-Briggs questionnaire was displaced by the personality battery that University of Oregon psychologist Lewis Goldberg dubbed “the Big Five.” Yet even among the extroverted, any new self-knowledge was usually well guarded. “It’s very difficult to get people to trust you to share psychological data with them,” says Vess Popov, development strategist of the University of Cambridge’s Psychometrics Centre.

A decade ago, the centre’s David J. Stilwell and Michal Kosinski uncovered a new way to get people to part with personal data: social-media quizzes. Since their MyPersonality app was launched in 2007, six million people have completed the questionnaire—nearly half of them allowing the Cambridge’s Psychometrics Centre to access their Facebook profiles as they did so. Once a user grants such access, algorithms trawl through likes and posts to train statistical models that use such “digital footprints” to predict personality types. Scholars are allowed to dip into that pool of anonymized data for worthy academic research, and the fruits of those models are promoted commercially as Apply Magic Sauce, a data stream that allows online marketers to adjust their appeals to potential consumers based on their likely attributes.

College ties may have shaped the company’s intellectual heritage, but Cambridge Analytica’s prospects are now determined by its links to American hedge-fund wealth. It is Robert Mercer’s daughter, Rebekah, otherwise responsible for the Mercer Family Foundation, who said to be the biggest booster of Cambridge Analytica’s methods. By all accounts, the company’s business has become deeply intertwined with the family’s political interests. In Manhattan, the company shares an address with the free-market advocacy group Reclaim New York, where Rebekah Mercer serves as treasurer. “To break into the U.S. political market you need more than just a good idea, you actually need to have people to believe in you,” says Nix, who refuses to specifically discuss any of the company’s investors.

Yet as Politico noted in its story about the Mercer investment, all of Cambridge Analytica’s clients in 2014 were also recipients of contributions from the financier. In 2013, the company’s only American client was the Middle Resolution PAC, a Virginia conservative group then working to elect Ken Cuccinelli as the state’s governor, an unpaid pilot project to show how SCL might work. 

“The traditional microtargeting vendors I’ve worked with, they were all waiting for me to call them, I guess,” says Paul Shumaker, a North Carolina general consultant who hired Cambridge Analytica to work for Thom Tillis’s successful campaign for the Senate last year. “We were one of the top US Senate races in the country and I never got approached by anybody else—which told me they were either too busy or took me for granted.”

Cambridge Analytica might not have had to work so hard for one of its more lucrative 2014 accounts. With a $1 million contribution, Robert Mercer was the largest single donor to the John Bolton super-PAC. The group had one objective—to convince voters to support Republican candidates based on national-security issues—and it served well to demonstrate a personality-driven theory of political persuasion. Bolton’s committee agreed to communicate over satellite-television systems like Dish and DirecTV, which, unlike broadcast and most cable systems, permitted ads to be assigned differently to specific subscribers, allowing Cambridge Analytica to fully exploit the benefit of its individual-level modeling.

The firm, which was paid $341,025 for its work, advised Bolton’s team on the design of six ads, thirty seconds each, with wildly different creative approaches. One ad, targeted at voters modeled to be conscientious and agreeable, was set to upbeat music and showed Bolton standing outdoors on a bright day, matter-of-factly addressing the need to “leave a stronger, safer America for our children.”

https://www.youtube.com/watch?v=r1U35rl17Zo

In another, aimed at neurotics, the diplomat was invisible—replaced by storm clouds, foreigners burning American flags, and an admonition to “vote like your life depends on it,” intoned by an disembodied narrator. “That’s obviously something that’s quite emotive,” says Nix, “as we’re really looking to drive an emotional reaction from an audience who would be inclined to give you one.” 

At the same time, Jobs, Growth & Freedom Fund, a leadership PAC launched by Ted Cruz in preparation for his presidential campaign, paid Cambridge Analytica just under $400,000, most of it to support digital advertising on behalf of other candidates. (Over the course of the fall of 2014, Mercer donated $1.75 million to another committee, the Ending Spending Action Fund, that has backed Cruz.) By the time the committee transitioned this spring into a full-fledged presidential campaign, Cambridge Analytica was fully integrated into the Texas senator’s political plans. Even before he formally announced his candidacy, opened his Houston office, or had a pollster in place, Cruz had a London-based firm on call to tell him which Iowans were introverted and which were neurotic.

Cruz for President has relied on Cambridge Analytica as a ready-made data-science department that spares the campaign the challenge of having to hire (and compensate) its members individually. This is already enough of a challenge for Republican campaigns, who have trouble identifying friendly quants from academia or the tech sector, even without sixteen different presidential campaigns all angling for the same talent. Finding astrophysics postdocs who will happily work for Ted Cruz may be easier in Cambridge, England, than Cambridge, Massachusetts. Rebekah Mercer is said to talk bullishly about the innovative potential of “psychographic” modeling,but her greatest gift to Republican analytics may be as an end run around a dispiritingly tight labor market: finding foreigners to do the analytics jobs that Americans just won’t do.

There is little evidence yet that Cruz’s campaign has determined if, or how, to use Cambridge Analytica’s “psychographics” at what the company depicts as their fullest potential. Chris Wilson, a pollster who had guided Cruz’s microtargeting efforts in his 2012 campaign, was unconvinced that predicting voters’ personalities was as universal a tool for unlocking their attitudes and behavior as Cambridge Analytica executives claimed. But if the method ever were to prove its utility, he thought it would happen in a fractured primary field, where—as opposed to a general election featuring stark choices between different parties—campaigns are constantly looking for ways to magnify ultimately minor differences among candidates into major distinctions in voters’ minds.  

"In a primary composed of Republicans, there's not a big gap of difference between where each of the candidates are on a single issue, so your ability to connect with a voter on an issue is very important,” says Wilson, who now serves as the campaign’s director of research and analytics. “You're appreciating the reason they care about the issue."

Indeed, few of Cambridge Analytica’s clients appear to be taking full advantage of what employees describes as its “bespoke” service, the tailoring of messages and targeting tactics to align with its personality profiles. “I’m not convinced it would make any difference whatsoever,” says one veteran consultant who has worked closely with the Republican party’s data infrastructure and has met with Nix. “Even if what they are pitching is tangible and legitimate and can really give you voters to talk to in a different tone no campaign has the bandwidth…There’s not the targeting ability to do it.”

In most cases, Cambridge Analytica appears to be functioning as the type of straightforward microtargeter that Nix considered insufficiently alert to the vagaries of human psychology. In North Carolina, where the company was paid $150,000 by the state party and $30,000 by Tillis’s campaign, Cambridge Analytica developed models to predict individual support, turnout likelihoods, and issues of concern that would recalibrate continuously based on interactions with voters. Shumaker says that dynamic process allowed Tillis’s campaign to identify a sizable cluster of North Carolinians who prioritized foreign affairs—which encouraged Tillis to shift the conversation from state-level debates over education policy to charges that incumbent Kay Hagan had failed to take ISIS’s rise seriously. “It gives you an edge in increasing the probability that voters would pay attention to your message,” says Shumaker.

In conference calls and pitch meetings, Cambridge executives and analysts have betrayed confusion, if not outright ignorance, about some basics of American campaigns—from the definition of precincts (the smallest unit at which voter data is collected) to the difference between turnout patterns in primaries and caucuses. When Nix and Tayler showed me my record, they seemed to suggest that a prediction that I had an infinitesimal chance of voting in the Republican primary reflected my ideology, rather than Pennsylvania laws that exclude those registered as independents (as I am) from participating in any primaries. When it comes to describing the process behind the psychological predictions, they seem even more deliberately obtuse. “Our conversations with the Cambridge guys left us befuddled,” says a Republican consultant who has been on the receiving end of their proposal. “Their team is enamored by the promises of psychometrics, but they were surprisingly vague about its specifics and too quick to dismissively cry 'Analytics!’ in the hopes that would be enough to sell us. Their technical ambition is obvious, but they've got a ways to go before they impact the U.S. electoral space."  

A few weeks after I visited their London office, I went to the University of Cambridge Psychometrics Centre’s website to see what I could learn about myself from a psychological assessment detached from political considerations. The centre’s site shows users who sign in with their Facebook accounts the personality profile that Apply Magic Sauce generates from their digital footprints, but I hadn't liked or posted enough to adequately feed its algorithm. Instead I was directed to a familiar-looking series of agree-or-disagree questions that promised to plot me on both the Myers-Briggs and “Big Five” matrices. 

It took nearly half an hour to complete the 100-item questionnaire before receiving a diagnosis. The modeled assessment I had been shown by the Cambridge Analytica team had been accurate in placing me in the middle range for extraversion and low on agreeableness. But on other personality traits, the company’s prediction had been wildly off. I fell in the 16th percentile for neuroticism, while Cambridge Analytica had placed me in the 70th. The inverse was true on conscientiousness: the online test said I was in the 76th percentile, while Cambridge’s model put me in the 11th. While I was more open than closed, I was not much of an outlier—in the 60th percentile—and far from the psychological curiosity that Cambridge Analytica's test had revealed.  

The summary generated by these scores conjured an entirely different person. “You seem to describe yourself as someone who avoids foreseeable trouble through purposefully planning, and achieves success through persistence,” the site told me. “From your responses it appears that you are reliable and prepared for life's challenges.” This was far from the creative free spirit than I had been revealed to be in London, but it, too, managed to flatter me. How I would decide my vote for president remained a mystery to us all.

(Correction: A previous version of this story incorrectly described Mercer's involvement with Cruz's leadership PAC in 2014.)