The Geeks Shall Inherit
The Victory Lab: The Secret Science of Winning Campaigns
By Sasha Issenberg
In a now much parodied scene from the 2012 presidential election, Republican strategist Karl Rove, appearing on Fox News shortly after the crucial state of Ohio has been called for Obama, refused to concede the obvious—that Mitt Romney had lost the election. Here was the boogeyman under the bed of liberals, with a long track record of handing it to Democrats, very publicly denying what the preponderance of polling and number crunching before and during election night had already signaled.
Meanwhile, over at Romney headquarters in Boston, similar denial most likely infected the building. After all, it has been reported that Romney only wrote one election night speech—a victory speech. It was the campaign’s lead pollster Neil Newhouse—the guy who should be the most in-touch with reality—who had been predicting a comfortable Romney win and whose polls convinced Romney that he was headed for victory. A few states west, at the Obama Chicago headquarters, a team of several dozen data scientists were busy aggregating data from external polls, internal polls, exit polls, early vote results, and millions of data points coming in from volunteer door knocks and calls. They didn’t need to make gut-based predictions about the state of the race because they trusted that data would give them unbiased projections. In Chicago the team knew within a small statistical margin of error that Rove was wrong. But how could these two campaigns, both flush with money, be so far apart in their level of sophistication?
In the post-mortem, it’s easy to blame the Romney campaign for its technical and strategic incompetence, but it had just found itself on the wrong end of a multi-year progressive tactical effort to regain a technological advantage after a decade of losses. In The Victory Lab, Sasha Issenberg describes the back-and-forth intellectual arms race between Democrats and Republicans, focusing primarily on the behavioral scientists who study why people vote and data scientists who have honed the tools to collect and analyze massive amounts of information to better target voters. At various times in history one party or the other has prioritized and supported analytical rigor more than the other and benefited electorally. But with the exception of George W. Bush’s second campaign and to an even greater degree both Obama campaigns, the intellectuals have been a sideshow to the consultants and communications gurus who are typically given deity-like status. That appears to be changing. The nerds are now winning.
The existence of this book is an indication of the growing influence of data in campaigns. While there are dedicated areas in the biography section of most bookstores for famous campaign consultants like Lee Atwater, James Carville, and Rove, Issenberg is the first author to systematically write about this topic. But now in an era of Money Ball, Nate Silver and evidence-based medicine (the movement to predict risk in health care through mathematical modeling) it was only a matter of time before an author wrote about political number crunchers. Issenberg, who has a foundation of campaign knowledge from his years covering Philadelphia politics, combines his strengths as a thoroughgoing analyst in his wonky field with an understanding of how data translates to on-the-ground advantages. He has interviewed a host of insiders in both behavioral science and data-focused campaign consulting and, while there are always more stories to be told (and I’m sure many he was unable to write about for reasons of confidentiality), Issenberg touches most of the major topics and developments up through the 2008 presidential election. Additionally, Issenberg captures an important dynamic in the growth of data-driven campaigns; much of the progress of the last decade on the progressive side has not come from inside the Democratic Party or from individual candidates but from liberal groups and dedicated analysts operating outside the party apparatus. This nuance shows that Issenberg did his homework.
However, unlike Moneyball, The Victory Lab often feels inaccessible, even to a political junkie who has worked in politics like myself. Because Issenberg writes about highly detailed strategy the book can read like a scholarly journal with no connecting narrative. But, if the reader can power through some of the more in-depth sections, they will finish the book with a better understanding of modern political tactics and be able to follow campaigns with sophistication consistently missed by superficial political reporting.
One of Issenberg’s strengths is his eye for historical perspective. Linking together every campaign effort throughout history is the fact that the only barometer of success is whether or not a candidate received at least one more vote than the other guy or gal (except, of course, presidential campaigns, which must take into account the electoral college). With that enduring mission connecting all campaigns, it’s important to understand how the strategy of winning 50% +1 votes has evolved. Issenberg takes us all the way back to 1840 when a young Illinois state legislator named Abraham Lincoln strategized about winning the presidential race on behalf of his party’s nominee William Henry Harrison:
Lincoln instructed each committee to divide its country into districts and appoint a subcommittee for each. “Make a perfect list of all the voters in their respective districts, and ascertain with certainty for whom they will vote. If they meet with men who are doubtful as to the man they will support, such voters should be designated in separate lines, with the name of the man they will probably support. It will be the duty of said subcommittee to keep a constant watch on the doubtful voters, and from time to time have them talked to by those in whom they have the most confidence, and also to place in their hands such documents as will enlighten and influence them.”
Issenberg doesn’t chronicle every election from that point on but fast-forwards to modern campaigns starting with the individuals who have been advancing the field since Richard Nixon’s first campaign. We come to understand the macro forces that led Republicans to be the first experts of micro-targeting, which is the art of reaching a specific voter or type of voter. For a long time Democrats had no need to target beyond party-affiliation. Democrats tend to be clustered in cities where campaign workers can indiscriminately knock on doors with relative confidence that they will encounter a supportive person. As Issenberg explains, Democratic campaigns would
Republicans, on the other hand, had to “employ vote-counting rigor” and, starting with Reagan, began to prioritize among voters who were scattered in hard to reach suburbs and rural areas. Computer generated models identified those voters who were likely to support Reagan, even if that individual’s party registration was Democrat. These voters were largely white, working class and sympathetic to Reagan’s message of self-reliance. Likely raised Democrat, many had decided that their party was no longer working in their interest and, while not willing to register Republican, were willing to vote that way. Communications strategists and political pollsters have gotten most of the credit for successfully targeting what continues today to be referred to as a “Reagan Democrat” (or in Nixon’s case, “the silent majority”). But, as Issenberg uncovers, behind the scenes it was the computer programmers who made this strategy possible.
take areas known to be more than 65% Democrat and flood them with manpower, hitting every voter without regard to party. In Philadelphia they dubbed it “knock and drag”; in the black counties of rural Virginia it was “hauling and calling.” Both represented a fast-food strategy, identifying reliable margins and counting on volume for victory.
Democrats, turning to academia, which tends to align more with progressive principles, were quicker to embrace political science and behavioral psychology theories. University of California, San Diego, political scientist Samuel Popkin, for example, helped conduct the post-mortem that aimed to understand why George McGovern lost so convincingly even though a majority of voters agreed with his views on the war. As Issenberg writes,
Popkin discovered that voters relied on shortcuts (to evaluate candidates). They interpreted symbols and looked for cues where they could find them, and then extrapolated. In one of Popkin’s favorite examples, when voters saw Gerald Ford fail to shuck a tamale before biting into it, they interpreted it as a sign that he did not understand issues facing Latinos.
Yes, voters agreed with McGovern’s views. But, in watching the way he conducted his campaign, they concluded he wasn’t competent enough to implement his policies. There are lessons here that liberals such as Thomas Frank, the author of What’s the Matter with Kansas, may want to keep in mind when claiming that some low-income Americans don’t vote their interest when supporting Republican candidates. If Popkin is to be believed, rural white voters may understand that Democratic ideas could potentially benefit them financially. Yet these voters either don’t believe that Democratic candidates are capable of passing their ideas or have come to distrust the government’s ability to manage social programs.
With academic research from Popkin and others showing that experiments can help inform smart campaign strategy, one would think that campaign managers across the country, who after all are evaluated on their ability to win, would have run out to the nearest campus quad and signed up their own university scholar. As AFL-CIO union Political Director Mike Podhorzer says in the book, “There isn’t an industry in America that would spend $32 million [the amount his organization had just invested on John Kerry’s behalf] without spending a significant piece of that budget trying to understand what works and what doesn’t.” But for many decades the campaign old guard resisted change or, more importantly, resisted evaluation. One of the book’s recurring figures is Hal Malchow, a Democratic direct mail consultant who comes to believe in controlled field experiments. Throughout his career he has faced resistance:
If you wanted to build a business designed to resist learning from itself, Malchow was discovering, it would look pretty much like the American electoral campaign. The paper trail that might illuminate what actually happened—the binders filled with polling data, the hard drives filled with databases accounting for every direct contact made with a voter—usually ends up in the nearest dumpster. Often no one even convenes a postmortem among the staff operatives, consultants, and candidate to talk about what went wrong and why.
Opposition to using research to inform strategy sometimes came from campaign leadership, who were unwilling to set aside a control group for fear that ignoring those voters would cost them their campaign. Others, like former DNC campaign director Tom Lindenfeld, just didn’t believe in new techniques:
Those smart guys speak that smart language. They collect smart theories to properly arrange their smart facts. Then they publish smart papers to make sure people know they are real smart…. The rest of us just know what works.
The story of The Victory Lab is about of the visionary campaign staffers like Podhorzer and Malchow whose persistence has now outlasted the outdated thinking of consultants like Lindenfeld. On the Republican side, one of the more colorful supporters of data analytics is Texas Governor Rick Perry’s lead consultant David Carney (a rival of Karl Rove). For Perry’s 2006 re-election campaign, Carney decided to use academic researchers to effectively audit the campaign’s budget. Even more audacious, he hired two Yale researchers Don Green and Alan Gerber, who also happened to be progressives. In exchange for public silence until after the election, Carney agreed to allow Green and Gerber to design tests to evaluate the strategies of Perry’s team of media consultants. In a predictably tense meeting, Carney tells his team that their work would now be audited by these two Democratic academics from Connecticut:
“One of you is right,” says Carney. “Either the eggheads are right or you’re right. We’re going to prove it out, and plan our campaign and allow these guys to develop experiments for everything we do.”
After the meeting with his consultants Carney says Green and Gerber’s presentation was like, “going into the Catholic Church telling everyone that Mary wasn’t a virgin and Jesus really wasn’t her son.” But, in the end, Green and Gerber’s research dispelled several campaign myths using experiments and data. For example, they proved that the affect of television advertising is measurable but only has short-term impact and that you can quantify the positive influence of a candidate’s appearance in a local media market. These insights were turned into campaign strategy and Rick Perry has yet to lose a race in Texas.
Most importantly, these mathematical models helped maximize the efforts of the volunteers who donated their time to knock on doors and make phone calls. When a volunteer wastes even a few minutes calling the wrong voter, a campaign does a disservice to their efforts. Advances over the past year have helped provide these volunteers drastically better-targeted list of voters to contact. I can attest to the challenges Democratic campaigns had prior to 2008. Before campaigns prioritized cleaning up databases and before Democrats started hiring data analysts to refine target voters, our data was fairly inaccurate. I would walk around campaign offices and hear volunteer after volunteer frustrated because they were dialing wrong numbers, talking to staunch Republicans, already-decided voters, or even asking to speak to long deceased people. When you take these unproductive efforts and multiply them across a vast campaign, the waste can be staggering. Voter list management was clearly an area that needed improvement. This year, the Obama campaign took unprecedented steps to optimize their targets through modeling and list management techniques. According to a memo put out by 2012 Obama field leaders Mitch Stewart, Jeremy Bird and Marlon Marshall, last year the campaign organized volunteers to make 125,646,479 personal phone calls or door knocks. If the Obama campaign’s targeting and data cleaning improved their lists by say a mere 5% (I’m guessing here and I think this is conservative number), that effort potentially saved over 12.5 million minutes of volunteers’ time.
In addition to detailing Republican and Democratic efforts to beat each other with new strategy, Issenberg writes about another cyclical back and forth. In 2002, a Republican political strategist named Alexander Gage was summoned to the headquarters of a Massachusetts gubernatorial candidate, Mitt Romney. Gage had developed what he thought was a novel strategy for clustering voters based on consumer information and was asked to present his PowerPoint to the campaign:
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,” explained 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 passage illustrates just how far, in 2002, the political world lagged behind the corporate world when it came to the use of technology and data analytics. Contrast that with today. Last year I enrolled in the introduction to marketing class at UC Berkeley’s Haas School of Business, where I’m currently finishing my MBA. I opened our textbook to hopefully learn the latest in corporate marketing. On page one of the book, instead of a profile of the marketing strategies of Amazon or Coke there was a case study on the Obama campaign. Additionally, of all places, Google Chairman Eric Schmidt spent election night last year at Obama headquarters and reportedly Facebook has extended job offers to the entire Obama data team. Now, it’s the corporate world who look to political campaigns for inspiration. You can almost imagine Obama staff in Chicago telling Google, “you mean, you don’t do this in the consumer product world?”
And the new relationship between tech and political campaigns is now spilling into public policy. The White House recently announced the creation of The Presidential Innovation Fellows, which aims to bring developers and entrepreneurs into government for 6 to 12 months to solve tricky IT and innovation challenges. Additionally, the newly formed Code for America is linking together the tech community and local and state governments by placing developers into yearlong public service-focused fellowships. While this development is not covered in The Victory Lab it is the natural extension of the history Issenberg describes.
Because The Victory Lab only includes campaigns through 2008, the reader is left to wonder what has transpired since then. Many of the characters profiled in the book, such as Obama Chief Analytics Officer Dan Wagner, went on to develop even more sophisticated campaign innovations not mentioned in the book. And a little disappointingly, Issenberg ventures no prediction about the future of campaigns. I’m sure Issenberg has thoughts about how Republicans will respond to their current situation based on his knowledge of how Democrats reacted when they were in a similar situation several years ago. My first campaign was John Kerry’s 2004 failed presidential bid. After that loss, Democrats finally started to face the reality that their campaign infrastructure had fallen far behind the Republicans’. Issenberg describes how, after the defeat, Democratic consultant Laura Quinn’s “desk was covered with newspaper and magazine clips about how Bush had won, many of them lionizing Karl Rove, whom Bush had described the morning after his victory as ‘the Architect.’” For four years, progressives pieced together what Rove and his team had built and then set out to surpass their efforts. They created new permanent institutions such as the Analyst Institute, which researches best practices for progressive campaigns, built a nationwide database of voters, and invested heavily in year-round data and targeting staff. Democrats were fixated on never losing to Karl Rove again.
For many gleeful Democrats, seeing Rove self-destruct on Fox News this last Election Day was proof that they had finally won. And for those who have struggled to put data at the forefront of campaign strategy, it also signaled a victory of the nerds over the traditional strategist. But, The Victory Lab tells us, no advantage lasts forever. Somewhere there is a Republican studying Obama’s efforts and developing the next revolutionary technology. We will have to wait for a follow up to The Victory Lab to find out how this story continues.
Gabe Cohen was the 2008 Obama Pennsylvania General Election Director. He will be graduating in May from UC Berkeley’s Haas School of Business.