New Political Science ISSN: 0739-3148 (Print) 1469-9931 (Online) Journal homepage: https://www.tandfonline.com/loi/cnps20 Manipulating Citizens: How Political Campaigns' Use of Behavioral Social Science Harms Democracy William A. Gorton To cite this article: William A. Gorton (2016) Manipulating Citizens: How Political Campaigns' Use of Behavioral Social Science Harms Democracy, New Political Science, 38:1, 61-80, DOI: 10.1080/07393148.2015.1125119 To link to this article: https://doi.Org/10.1080/07393148.2015.1125119 Published online: 05 Feb 2016. Submit your article to this journal C? IjJiL Article views: 2400 5>1 View related articles G? View Crossmark data G? Citing articles: 15 View citing articles B" Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journal lnformation?journalCode=cnps20 NEW POLITICAL SCIENCE, 2016 VOL.38, NO. 1,61-80 http://dx.doi.Org/10.1080/07393148.2015.1125119 Routledge Taylor & Francis Croup Manipulating Citizens: How Political Campaigns'Use of Behavioral Social Science Harms Democracy William A. Gorton Department of Political Science Alma, Alma College, Michigan, USA ABSTRACT Critical theorists in the mid-twentieth century argued that behavioral social science by its nature conceptualizes knowledge as the power to predict and control. As such, rather than serving as a source of enlightenment or emancipation, social science risked functioning as a tool for dominating and manipulating the public.The power of this criticism, however, was undermined by the behavioral revolution's initial failure to produce theories that offered much prognostic power. Recent methodological and technological developments in the social sciences have begun to generate an impressive ability to predict human behavior, especially when combined with new innovations in marketing and computer science. Disturbingly, political campaigns and interest groups, especially in the US, appropriate this new knowledge to try to alter the beliefs and behaviors of voters. This development bodes ill for US democracy and other liberal democracies where the use of these techniques is likely to increase. It turns citizens into objects of manipulation and undermines the public sphere by thwarting public deliberation, aggravating political polarization, and facilitating the spread of misinformation. "I was always amused by the virulence of Noam Chomsky's denunciation of the American social scientists who had helped the American administration wage war in Vietnam: He damns their political morality, but he cannot refrain, at the same time, from scorning their scientific pretentions. Not only had they prostituted themselves morally, but as scholars they were frauds, their verbiage and techniques had no cutting edge ... Possibly so; but does not the one charge cancel out the other? Would it have been better or worse, if their understanding of the situation had been genuine and their advice sound, and if that had made it possible to wage the war effectively?"—Ernest Gellner,"The Scientific Status of the Social Sciences (und leider auch Sociologies" 1985.1 Introduction An important criticism of social science, developed most forcefully by critical theorists in the mid-twentieth century, is that social inquiry modeled on the natural sciences can too easily serve as a tool to manipulate people. Indeed, critical theorists argued that social 1 Ernest Gellner, Relativism and the Social Sciences (Cambridge: Cambridge University Press, 1985), p. 101. © 2016 Caucus for a New Political Science 62 (Q) W.A.GORTON science so understood, variously described as "positivistic" or "behavioralist," by its very nature conceptualizes knowledge as the power to predict and thus to control. Rather than serving as a source of enlightenment or emancipation, positivistic social science then risks functioning as a handmaiden to an oppressive, undemocratic technocracy. The power of this criticism, however, was undermined by a fact of some embarrassment to social scien-tists:The twentieth-century behavioral revolution in social science never really made good on its promise of producing theories with genuine predictive power. Some forty years ago, this failure prompted Ernest Gellner to compare American behavioral social science to a cargo cult. Islanders in the South Seas once thought they could summon the technological power of the US Marines by creating crude facsimiles of tanks and airplanes out of bamboo. Similarly, Gellner said, American social scientists at times seemed to think that simply aping natural science methods would bring the"magic"of prediction and control."The empiricism of American psycho-social science is sometimes attributed to the pragmatic temper of the American nation," he wrote, "but surely the striking feature of some of those pseudo-scientific gadgets is their practical irrelevance, cumbersomeness, indeed obstructiveness. This is even noticed by their users, but they don't mind: the miraculous Cargo will not arrive at once, one must have a little patience ... The magic will work, but not just yet."2 Gellner's analogy, though perhaps grounded in crude and unflattering stereotypes of indigenous peoples, nonetheless provided a telling and stinging assessment of behavioral social science's value. However, Gellner was describing behavioral social science—particularly the variant practiced in the US, where it was most developed and influential—in its infancy. From the vantage of the twenty-first century, behavioral science's impotency is no longer evident. In fact, perhaps that magic of prediction and control has at long last arrived, at least in some measure and in certain domains, but whether this is a welcome development is by no means certain. The argument of this article is that methodological and technological developments in the behavioral social sciences and related fields, in conjunction with marketing techniques developed for selling consumer products, are beginning to produce the power to manipulate, in fairly precise and predictable ways, individuals'beliefs and behaviors. Some of these methods and techniques have been developed primarily by social scientists working in the academy while others have been created by computer scientists and artificial intelligence researchers working for private industry. Disturbingly, political consultants and operatives— and sometimes academic social scientists themselves—increasingly draw upon this knowledge to try to control the behavior of citizens in their most important democratic roles, including voting and forming opinions on public policy. These developments include the increasing use of field experiments, which are conducted on actual voters, usually without the voters'awareness, rather than on volunteers in a laboratory setting. Other potentially manipulative techniques include microtargeting individuals in combination with "big data" analysis, and the use of focus groups in conjunction with framing theory. Microtargeting involves creating finely honed messages targeted at narrow categories of voters based on sophisticated combinatorial analysis of data garnered from individuals'demographic characteristics and consumer and lifestyle habits. Framing theory posits that human thinking is guided by unconscious mental structures or "frames," which can be triggered by certain words or phrases. As discussed below, politically engaged proponents of framing theory, 2Emest Gellner, The Concept of Kinship (New York: Basil Blackwell Inc., 1987), p. 210. NEW POLITICAL SCIENCE (Q) 63 such as George Lakoff and Frank Luntz, in recent years have advised political campaigns on how to craft their messaging so as to trigger voter responses favorable to the campaigns' goals.This article argues that these developments are bad for democracy and social science, too.These techniques undermine the public's ability to act and deliberate freely and knowl-edgably in the political arena, and threaten to turn social science into a tool of technocratic manipulation. In the following sections, I examine how political campaigns, at times with the direct assistance of academic social scientists, have begun to draw upon field experiments, micro-targeting, and framing theory to facilitate the manipulation of citizens in their roles as voters. I argue that the use of these methods and theories enables manipulation in at least four ways. First, these approaches often produce much more precise predictive power when compared with previous methods employed by campaigns. When applied, they are often remarkably effective at altering citizens'behavior in predictable ways, thus helping to turn voters into potential objects of control rather than autonomous political actors. Second, in the case of field experiments, the experimentation itself alters the behavior of citizens in the act of voting, usually without citizens' awareness. Third, the use of these methods and theories, especially when employed in conjunction with each other, serves to undermine a healthy public sphere by individualizing, isolating, and distorting political information. And, finally, many of techniques that are used to try to alter citizens' behavior are grounded in models of unconscious processes of the human mind. Thus, when these theories are applied, they typically alter voting behavior and public opinion formation through processes that often completely elude the understanding of their intended targets. I should note that my analysis focuses on how behavioral science is being used in the US to manipulate its democratic processes. This focus is not arbitrary. As I explain in the next section, the US was the primary locus of the twentieth-century behavioral revolution, and behavioral social science remains to this day more prevalent and highly developed in the US. In addition, political campaigns in the US have probably gone further than anywhere else in appropriating new behavioral theories and techniques to try to alter voter behavior. However, there is every reason to suppose that this knowledge and its uses will spread to other liberal democracies, where the incentives to use behavioral science are no doubt equally powerful. Knowledge is Power "From Science comes Prevision; from Prevision comes Action,"declared Auguste Comte.3 He called for a new science of society—sociology—that would seek to uncover general laws of society. Knowledge of such laws would provide the"prevision"to inform public policy, which would be a kind of social engineering. "[F]or it is only by knowing the laws of phenomena, and thus being able to foresee them, that we can, in active life, set them to modify one another for our advantage.'4 Armed with such knowledge technocratic planners could tackle society's most pressing problems: fostering stability, economic growth, and technological innovation; reducing poverty, violence, and disease. As society moved out of its metaphysical stage and into its final historical era—the positivist era—Comte predicted that a new epoch of "order and progress"would emerge. 3Auguste Comte, The Positive Philosophy of Auguste Comte (London: John Chapman, 1853), p. 20. 4lbid. 64 @ W. A. GORTON The idea of a science of society modeled on the natural sciences captivated other nineteenth-century founders of social science, including John Stuart Mill, Emile Durkheim, and, in the US, Lester Ward. Brian Fay succinctly describes the appeal of the idea: [J]ust as the natural sciences have provided men with a certain kind of knowledge by which they can control their natural environment, thereby making it more hospitable and productive, so also the knowledge gained from social science will enable men to control their social environment, thereby making it more harmonious and congruent with the needs and wants of its members. Natural science gives men an enormous power based on knowledge of the workings of the external world, and it is this power which sustains and supports the entire undertaking.5 This vision of social science underpinning a technocratic society was also, for many, troubling or even chilling. In particular, it came under sustained and particularly biting criticism from the original critical theorists of the Frankfurt School, most notably Max Horkheimer, Theodor W. Adorno, and Herbert Marcuse. A key criticism was that social inquiry modeled on the natural sciences tended to reify the social world, or treat it as a "second nature," in the words of Adorno.6'7.That is, such an approach, which Adorno and Horkheimer labeled broadly as "positivism," tends to reinforce the view that regularities uncovered by social science are natural and unalterable, as is the case with the laws of the natural world. Further, the critical theorists argued that inherent in the positivistic method was the tendency to view scientific knowledge in purely instrumental terms. Science becomes a tool for dominating both the natural and social worlds, and society and its inhabitants become objects to be manipulated.8 As David Held has summarized this view: "Matter is defined as a possible object of manipulation. People, embodying the natural, are also potentially controllable."9 For the critical theorists, knowledge so conceived produced at best a superficial understanding of society and, moreover, functioned as a handmaiden to domination by masking hidden power relations. Further, the envisioned technocratic society was, at its core, deeply undemocratic. Of necessity, it tended to cede control over policy-making to a technocratic elite, and it also served to transform questions of value judgment into instrumental problems of efficiency and utility. The scope of ordinary citizens'role in the political process, regarding both means and ends, accordingly shrank.Thus, while positivist social science claimed to be "value neutral," it in fact served the interests of power and fostered an approach to inquiry that has as its aim control and manipulation rather than understanding and emancipation. In short, positivism functioned as an ideology, in the Marxist sense—a system of ideas, practices, and norms that masks or legitimates domination. At the same time that the Frankfurt School was developing its critique of technocratic society, a growing number of American political scientists were eagerly embracing a broadly positivistic approach to political inquiry, in what became known as the "behavioral revolution/The roots of the revolution can be traced at least as far back as the 1920s, when Charles Merriam of the University of Chicago helped launch the social science research council (SSRC).10The aim of the SSRC was to bring a new spirit of "behavior research" to political 5Brian Fay, Social Theory and Political Practice (London: George Allen & Unwin, Ltd., 1975), p. 19. 6Herbert Marcuse, One-Dimensional Man (Boston: Beacon Press, 1964), p. 168. 7TheodorW. Adorno,"Sociology and Empirical Research," in Paul Connerton (ed.), Critical Sociology, p. 243. 8Herbert Marcuse, One Dimensional-Man (Boston: Beacon Press, 1964), p. 155. 9David Held, Introduction to Critical Theory: Horkheimer to Habermas (Berkeley: University of California Press, 1980), p. 167, emphasis in the original. 10Clyde Barrow, "Political Science,"in Darity, W. A. (ed.), International Encyclopedia of the Social Sciences, (2008), p. 312. NEW POLITICAL SCIENCE (Q) 65 science and, in particular, it emphasized experimental methods and the application of statistics to help uncover behavioral regularities in politics.11 By the 1950s, the behavioral movement was sweeping through American political science. Its most prominent advocates—amongst them were David Easton, Robert Dahl, V.O. Key, Eulau Heinz, and Harold Laswell—sought to create a social science grounded in controlled observation and quantification of political behavior. Their ultimate aim was production of empirically testable law-like generalizations of political behavior, focusing especially on such observable phenomena as voting, opinion formation, interest group and political party behavior, and legislative process. The behavioral revolution in political science thus promised to produce theories that mirrored those of natural science: broad in scope, precise, and law-like. The hope was to create a "pure science" of politics, which could then be converted into an applied science to help solve practical or urgent political problems and to improve institutional design generally.12 However, the pursuit of general laws of political science, on which the vision of the "pure science"of politics relied, was a bust.To be sure, a handful of purported general laws of politics were indeed proposed, amongst them "Duverger's Law," (which holds that simple-majority, single-ballot electoral systems produce two-party systems) and Roberto Michel's"iron law of oligarchy" (which posits the tendency of political parties over time to become oligarchical in organization). Upon closer inspection, such laws never included the features of genuine law-like regularities uncovered by natural science: They lacked clear boundary conditions, often could not specify the magnitude of the relationships between variables, and were generally imprecise and exception-ridden.13 In retrospect, it seems clear that the search for general laws of politics was doomed from the outset. It foundered on many obstacles, including the complexity and reflexivity of political phenomena, overly simplistic models of human rationality, and the general difficulty of conducting controlled experiments in the social world.This is not to say that the revolution was a complete failure. It produced much insight into political behavior—mechanisms were identified and clarified, correlations were uncovered, and a clearer factual picture of the political world was produced. But behavioralists' inability to uncover general laws of politics deprived would-be technocratic policy scientists of much of their potential ability to predict and control economic and social phenomena. In this sense, the promise—or the specter—of technocratic policy science never emerged. In the last third of the twentieth century, trends in social science, however, would emerge that would open the door, if only by a crack at first, to the kind of manipulation and control of which the original critical theorists warned. The most significant early adapters of this manipulative power would not be a class of technocratic policy scientists. Rather, they would be the managers and consultants working for modern political campaigns.The problematic nature of modern campaigning, and in particular campaigns'use of mass and new media, has certainly not escaped the notice of thinkers writing in the critical theory tradition or other progressive observers concerned about the health of contemporary democracies. However, most of the exploration of these issues has tended to focus on the various ways in which private or government power can damage democracy and distort the public sphere through, for instance, agenda setting or inhibiting the political participation of certain groups. This "The SSRC was amongst the first to employ the "field experiments" in voter turnout discussed below. l2lbid.,313. 13James Farr, "Resituating Explanation," in Terrence Ball (ed.), Idioms of Inquiry: Critique and Renewal in Political Science (New York: SUNY Press, 1987), pp. 48-49. 66 @ W. A. GORTON study, however, focuses on the behavioralist mindset that informs certain new theories and techniques employed by modern campaigns, and how these theories and techniques not only conceptualize citizens as objects to be manipulated, they are actually disturbingly effective at manipulating citizens' beliefs and behaviors. This enhanced efficacy, this study contends, in effect constitutes a new form of social control. Manipulating Citizens Field Experiments on Citizens In 2000, two Yale University political scientists, Alan Gerber and Donald Green, published the results of an innovative study of voter turnout that they conducted in New Haven, Connecticut, in the weeks prior to the 1998 federal midterm elections. The study, entitled "The Effects of Canvassing, Telephone Calls, and Direct Mail on Voter Turnout" and which appeared in the American Political Science Review, was a "field experiment" in which New Haven residents were randomly selected to receive different kinds of nonpartisan messages encouraging them to vote. Some households in the treatment group were contacted in person by door-to-door canvassers, and others received postcards encouraging them to vote, while members of a third group of households were contacted by telephone. In turn, each of the three treatment groups was further divided into three subgroups that received different messages dubbed "civic duty,""close election," and "neighborhood solidarity." Recipients of the civic duty message were told that it was their"civic duty"to vote and that"[d]emocracy depends on participation of our country's citizens."The "solidarity" message informed individuals that"[p]oliticians sometimes ignore a neighborhood's problems if the people in that neighborhood don't vote," while the close election message stated that"[e]ach year some election [in the US] is decided by only a handful of votes."14 The study found that contacting potential voters face-to-face via canvassers was by far the most effective way to prompt citizens to vote. Individuals contacted by canvassers were ten to thirteen points more likely to vote than members of the control group, who were not contacted in person, by mail, or by phone. Contact via direct mail boosted turnout by only a few percentage points, while telephone contact produced no measurable increase in turnout at all. Further, within the subgroups, persons who received the"close election"message from door-to-door canvassers were three percentage points more likely to turnout for the election than those who received the "civic duty" message and seven percentage points more likely than those who got the "solidarity" message. Green and Gerber concluded that face-to-face canvassing was, by far, the most cost-effective way to improve turnout, at a cost of eight dollars per additional vote. In contrast, they estimated that each additional vote garnered via direct mail cost forty dollars. The New Haven voter turnout study was innovative and proved to be highly influential, inspiring hundreds of otherfield experiments of voter turnout in the years after its publication.15 Surveying the cumulative findings of these studies, Green and Gerber have offered precise calculations of the effectiveness of various get-out-the-vote tactics. For example, they estimate that door-to-door canvasing, the most effective approach, generates one vote per 14Alan S. Gerber and Donald P. Green, "The Effects of Canvasing, Telephone Calls, and Direct Mail on Voter Turnout: A Field Experiment," The American Political Science Review 94:3 (2000), p. 656. 15Donald P. Green and Alan S. Gerber, Get Out the Vote: How to Increase Voter Turnout (Washington, DC: The Brookings Institution, 2008), p. viii. NEW POLITICAL SCIENCE (Q) 67 fourteen voters contacted, whereas leafleting produces only one vote per 189 contacts.16 They also conclude that many of the get-out-the-vote techniques touted by consultants and campaign managers, such as robo-calls and email, are almost completely ineffective. In addition, voter-turnout field experiments collectively offer further evidence that the general decline in voter turnout since the 1960s likely stems in part from the decreasing level of face-to-face contact from canvassers (even as contact via mail and telephone has seen huge increases). The decline in face-to-face contact, in turn, probably helps explain the general decline in social capital, as documented by Robert Putnam.17 Clearly, field experiments in voter turnout add to our understanding of voting behavior. This inquiry, however, draws attention to aspects of these studies that should give political scientists pause. Voter-turnout field experiments include many of the attributes of purportedly "value free," positivistic social science that drew the Frankfurt School's scorn. Consider first the nature of field experiments themselves. Here are controlled experiments of social phenomenon that generate instrumental knowledge that can be used, and is used, to control the behavior of human subjects. Voter-turnout field experiments findings are not produced in laboratories using voluntary subjects in simulated elections, nor are they gleaned from fined-grained regression analyses of citizens'behavior after an election. Rather, such experiments attempt to alter and measure the behavior of actual citizens taking part in the definitive democratic act: voting in an election. Such experiments change the voters' behavior, and perhaps even the outcome of an election. Further, the studies can also be viewed as manipulative insofar as the canned messages used in such studies are not sincere attempts to use reason and evidence to persuade citizens to vote. Instead, the messages are conceptualized and employed as a kind of causal stimulus. The messages'value is purely a function of their power to prod people to the polls. The content of the argument or evidence in the message does not matter as such; all that matters is that the message "works." In fact, the most effective message in the original New Haven experiment, the "close election" script, relied upon voters' intuitive but inaccurate understanding of probability. The message sought to prompt individuals to vote by implying that their vote could quite possibly be the deciding vote, when of course the odds of such an outcome are vanishingly small. Finally, as noted above, findings from voter turnout field experiments are often converted into dollars-per-vote formulae, thus packaging them into a handy cost-benefit format intended for use by political campaigns and consultants.This provides a paradigmatic example of instrumental,"value free" knowledge: a tool ready for use, regardless of the aims or identity of its users. In fact, in 2005 Green and Gerber traveled to Texas at the invitation of Dave Carney, the chief campaign adviser to Rick Perry, who was then seeking election to the Texas governorship. There they met with George W. Bush's chief political adviser, Karl Rove, and two other academic political scientists to draw up plans to conduct an elaborate series of randomized field experiments with generous funding from the Perry campaign.The campaign organizers' intention was to draw upon state-of-the-art social science research techniques to learn how to use campaign dollars to their maximum advantage. As Sasha Issenberg noted in his account of the campaign:"Rove, who had an autodidact's fascination with American history, was eager to learn political science theories he could apply to the campaigns he understood only in anecdotal terms."18 Under the political scientists'guidance, 16lbi<±, 139. 17Robert Putnam, Bowling Alone: The Collapse and Revival of American Community (New York: Touchstone, 2000). 18Sasha Issenberg, Victory Lab: The Secret Science of Winning Campaigns (New York: Random House, Inc., 2012), p. 226. 68 @ W. A. GORTON the campaign created a series of field experiments designed to test the effectiveness of television advertisements and campaign stops on Perry's approval ratings. Media markets were randomly selected for advertisements and Perry appearances, and their impact on Perry's popularity was closely monitored. The political scientists involved in the experiment were required to pledge not to discuss the experiments until the campaign was over.19 Green and Gerber packaged the results of their now numerous field experiments in Get Out the Vote: How to Increase VoterTumout, which is intended to be"a practical guide to managing get-out-the-vote drives, while at the same time pulling together all of the scientific evidence about the cost-effectiveness of face-to-face canvassing, leafleting, direct mail, phone calls, and other tactics."20 The book includes dollar-cost-per-vote estimates for the different methods. Microtargeting and Big Data In 2012, The New York Times reported that both the Obama and Romney presidential election campaigns were using "microtargeting"to develop sophisticated voter profiles to help them identify friendly voters and nudge them to the polls.21 Microtargeting involves culling publically available data about individuals provided by the government (for example, voter registration, charitable giving, zip code, race, gender, age, family size) with consumer and lifestyle information (such as income, credit card spending habits, magazine subscriptions, number and kind of automobiles, church membership, stores visited, and the like). Such lifestyle information can be purchased from private firms such as Acxiom, Claritas, and Experian. These firms gather much of their information by monitoring the data trail individuals create through their Internet activity, such as website visits, online purchases, and interactions on social media sites, including status updates and "likes" on Facebook. The companies also provide their clients with "database marketing"analysis, which predicts what sort of products a consumer is likely interested in and what type of advertisements will most likely goad him or her into purchasing those products. Database marketing uses sophisticated combinatorial analysis enabled by "big data"to generate its models of consumer preferences and behavior. Acxiom, for example, draws upon some 1,500 data points to produce its predictive model of consumer behavior. Based on such data, Acxiom has created more than four hundred categories of consumers and given them whimsical names such as"McMansions and Minivans," "Flats & Convertibles," "Kid Country USA," "Young and Rustic," and "Bedrock America." The categories are used to predict buying habits, including not only what type of consumer products the different categories members are likely to seek out, but also when, where, and how they will purchase them. With its array of 23,000 servers, Acxiom can analyze some fifty million data transactions produced by five hundred million individuals each year.22 Political campaigns now use database-marketing techniques to generate similarly finegrained categories of voters that they use to predict political beliefs, values, and voting behavior. George W. Bush's 2004 US presidential campaign, for instance, used Acxiom data 19lbid.,229. 20Green and Gerber, Get Out the Vote, p. vii. 21Tanzina Vega, "Online Data Helping Campaigns Customize Ads," The New York Times, February 20, 2012, available online at: . 22Natasha Singer, "Mapping, and Sharing, the Consumer Genome," The New York Times, June 16, 2012, available online at: NEW POLITICAL SCIENCE (Q) 69 on 5.7 million Michigan consumers and wedded it to their own polling information to categorize Michigan voters into thirty-four"MicroTargeting Segments/'with labels such as "Archie in the Bunker,""Flag & Family Republicans,"and"WageableWeak Democrats."23 Similar efforts were launched in other battleground states that year. A key way in which microtargeting data is used is in the creation of scripted messages and advertisements targeted at the various narrow categories of voters. The 2004 Bush campaign generated numerous kinds of telephone and direct-mail messages for different categories of voters."Young Unreliable Pro-President Bush Independents"received mail emphasizing Bush's"No Child Left Behind" policy, while "Anti-Porn Women" learned about Bush's plan to restrict access to pornography in public libraries.24 More recently, campaigns have begun creating Internet advertisements tailored to the different categories of voters. Using such micro-categories, according to Zac Moffet, the digital director for Mitt Romney's 2012 US presidential campaign, "[t]wo people in the same house could get different messages. Not only will the message change, the type of content will change."25 Individuals identified by the Romney campaign's microtargeting analysis as committed Republican voters, for instance, received online advertisements that framed the 2012 presidential election as a battle "to save the soul of America." Persons identified as undecided voters saw advertisements that emphasized Romney's family life and character. In an example of even more precisely tailored microtargeting, in 2009 the New Jersey gubernatorial campaign for Chris Christie created an Internet advertisement aimed at Republican women who had searched online for information about breast cancer.The ad was hastily created in response to accusations by New Jersey Governor John Corzine, Christie's opponent in the election, that Christie supported reduced funding for mammograms.26 Civil libertarians have raised concerns about how the use of database marketing, whether to sell consumer products or political candidates, can invade individuals'privacy. As Philip Howard observes,"Whereas public opinion was once bluntly measured with polls, today it is modeled and predicted with surprising accuracy but not always with our informed consent. As we go about the business of our lives, we leave a data trail that is increasingly referenced by political actors, data from which our individual political preferences are extrapolated."27 The implications of database marketing for citizens'privacy are indeed troubling.This study, however, draws attention to other ways that using database marketing in conjunction with microtargeting, especially when used in combination with big data analysis described in greater detail below, may be damaging US democracy. First, microtargeting manipulates individuals'opinions by corroding and inhibiting public dialogue. That is, microtargeting undermines the public sphere. It does this by helping to shield potential voters from information or viewpoints that might challenge their beliefs and values. Microtargeting also makes it easier for campaigns to avoid contacting certain categories of citizens, thus helping to "redline" them out of political debate. In addition, microtargeting enables the spread of false or misleading information. Under the sway of microtargeting campaigns, citizens'opinions become increasingly manufactured and more tenuously anchored in reality. "Douglas B. Sosnik, Matthew J. Dowd and Ron Fournier, Applebee's America (New York: Simon & Schuster, Inc., 2006), p. 37. 24lbid., 40. 25Vega,"Online Data.". 26lbid. 27Philip N. Howard, New Media Campaigns and the Managed Citizen (Cambridge: Cambridge University Press, 2006), p. 187. 70 @ W. A. GORTON Yet the corrosion of public dialogue is not the only sense in which microtargeting fosters the manipulation of citizens. A second way in which the technique contributes to manipulation occurs when it is wedded to sophisticated big data analysis, which greatly enhances campaigns' power to predict potential voters' behavior, including whether they will vote, whom they will likely vote for, and what types of messages will likely resonate with them. With laser-like precision, campaigns can now locate friendly or potentially friendly voters and choose just the right message to nudge them to the voting booth. Critical theorists like Adorno and Horkheimer called our attention to how behavioral social science conceptualizes human beings as things to be manipulated. As I noted in the previous section, however, the generally feeble ability of actual behavioral social science to predict human behavior has taken some of the sting out of their critique: without prediction, no control. But with the advent of microtargeting, the ability to control voters' behavior appears much closer to reality. Disturbingly, as described below, this enhanced predictive power is often generated without any corresponding understanding of voter behavior. With great accuracy, big data analysis can tell campaigns which voters are likely to vote for which candidate or party, and which canned messages are likely to get allied voters to vote. But often nobody involved in the campaign can explain why a particular voter has been identified as a likely supporter or not, apart from the fact that the computer analysis says so. Indeed, even the mathematicians and computer programmers who design the algorithms that analyze the data cannot explain the particular predictions that their analysis generates.The combinatorial analysis, weighing thousands of disparate data points, is far too subtle and complicated to be grasped by the human mind.28 Undermining the Public Sphere Cass Sunstein has argued that in a democracy a "well-functioning system of free expression must meet two distinctive requirements": First, people should be exposed to material that they would not have chosen in advance. Unplanned, unanticipated encounters are central to democracy itself. Such encounters often involve topics and points of view that people have not sought out and perhaps find quite irritating.They are important partly to ensure against fragmentation and extremism, which are predictable outcomes of any situation in which like-minded people speak only with themselves ... [I]n a democracy deserving the name, lives should be structured so that people often come across views and topics that they have not specifically selected. Second, many or most citizens should have a range of common experiences. Without shared experiences, a heterogenous society will have a much more difficult time in addressing social problems. People may even find it hard to understand one another. Common experiences, emphatically including the common experiences made possible by the media, provide a form of social glue.29 Sunstein's concern is that the Internet enables behaviors that can undermine these two requirements of a healthy public sphere. He notes how the Internet allows users to filter out information that does not interest them as well as opinions with which they disagree. For instance, various Internet services, such as Rich Site Summary (RSS) and Google News, enable individuals to create news feeds that post only the type of articles that the user has 28FrankPasquale, The Black Box Society.The Secret AlgorithmsThat Control Money and Information (Cambridge: Harvard University Press, 2015). 29Cass Sunstein, Republic2.0. (Princeton: Princeton University Press, 2007), pp. 5-6. NEW POLITICAL SCIENCE (Q) 71 previously registered interest in, creating a "newspaper of me," in the words of George Bell, Chief Executive Officer (CEO) of the search engine Excite.30 Previous generations of media consumers, whether viewing television, listening to radio or reading newspapers and magazines, frequently encountered the same news stories and political advertisements. They also oftentimes came across news stories that they might not have chosen to encounter and viewpoints with which they disagreed. These unanticipated encounters enriched the public sphere by creating a kind of shared text of issues and information that in turn facilitated dialogue amongst citizens. As more and more Internet users channel themselves into informational cul-de-sacs reflecting their own interests and beliefs, Sunstein argues, the public sphere is gradually diminished. The emergence of microtargeting as a campaign tactic—a phenomenon that barely existed in 2001 when Sunstein first registered his concerns about the Internet—risks further erosion of the public sphere. As individuals get more and more of their news via their home computers, tablet personal computers, smartphones, and streaming television services such as Netflix and Hulu, they will increasingly encounter political messaging tailored specifically to them. (In fact, between 2000 and 2010, the percentage of Americans who reported getting political information from the Internet expanded from thirteen to seventy-three percent.31) In the early days of the Internet, individuals often sorted themselves into silos of political information. Now political campaigns, with the help of state-of-the-art data analysis and lifestyle information purchased from private firms, do the sorting, usually without the understanding or consent of the people being sorted. The result is interaction between the campaign and potential voter that is increasingly isolated and individualized. Consider Richard Semiatin's description of contemporary political campaigns: Campaigns are becoming more individualized and tailored to you, the voter. For the first 150 years, campaigns were largely the domain of party organizations. The birth of television and the advent of advertising spawned personality-driven campaigns. Today, we see the next revolution—that campaigns are attempting to reach each voter individually. The campaign of the future (and to some extent the future is now) can target each household. Campaigns used to be about parties and candidates. Increasingly, campaigns will become aboutyoiv, the customer.32 Howard describes political campaigns that use data mining and microtargeting as "hypermedia campaigns," as opposed to earlier campaigns, which relied heavily on mass media to engage voters:33 Traditionally, a large portion of the political information we had to digest each day was through random encounters with newspaper headlines and other opinions, but political hypermedia are designed to remove the risk of random exposure to political content from our lives ... Hypermedia campaigns are designed to present information in a largely unmediated form or in a form that is mediated by the citizen's own filtering preferences.34 Thus, in contrast to previous generations, which got their political information largely from mass media, citizens today are less and less likely to encounter advertisements and information from politicians, interest groups, and political parties whose views they oppose. Further, 30CassSuntein,"The Daily We: Is the Internet Really a Blessing for Democracy?" Boston Review, June 1,2001, availableonline at: < http://www.bostonreview.net>. 31Dick Simpson, "New Political Campaigns and Democracy," in Richard J. Semiatin (ed.), Campaigns on the Cutting Edge, (Los Angeles: CQ Press, 2012), p. 231. "Richard J. Semiatin/'lntroduction—Campaigns on the Cutting Edge,"in Richard J. Semiatin (ed.),Campa/gns on the Cutting Edge, (Los Angeles: CQ Press, 2012), p. 4. "Howard, New Media Campaigns, p. 197. 34lbid. 72 @ W. A. GORTON microtargeting facilitates what Howard has called "political redlining."35 As microtargeting empowers campaigns to become more effective in targeting voters, it also helps them to efficiently avoid expending resources on populations less likely to vote. Unlike in the previous era of mass media campaigns, significant swaths of the public can now be redlined out of political discussion altogether. As one anonymous political activist employed by a firm using microtargeting put it, "The data let you target. Who would want to target nonvoters, for example? Big waste of time.'86 Minorities and the poor are over-represented amongst nonvoters; thus, political redlining mirrors the practice of neighborhood redlining used by banks and realtors to deny housing loans and segregate communities. In addition to producing highly individualized messages, microtargeting produces messages that can more easily fly under the radar of the press and the broader public, markedly increasing their power to mislead and misinform viewers with impunity. Indeed, a campaign or a third party could potentially provide different categories of voters with plainly contradictory messages and elude detection. Reporters and independent fact-checking organizations simply lack the time and resources to monitor all the microtargeted messaging. As Kathleen Hall Jamieson observes, "Whereas ads carried in broadcast or cable channels can now be tracked by groups such as Kantar-CMAG, no comparable process exists to enable reporters and scholars to reliably intercept narrowcast information on the internet. As a result we do not know the characteristics of the targeted messages that the Interactive Advertising Bureau reported accounted for between $130 million and $200 million in ad spending during the 2012 presidential election."37 Enhanced Predictive Power without Understanding In some cases, campaign microtargeters draw upon academic social science research to refine their models of voter behavior. In an early and comparatively crude use of microtargeting, the campaign managers for Bill Clinton's 1996 presidential reelection bid broke down voters into nine categories, including "Social Liberals,""Crime Stoppers," and "Rich Conservatives." Mark Penn, the campaign's lead pollster, refined the voter categories by drawing upon a modified version of the Myers-Briggs personality test to help identify which kind of voter best responded to Clinton's communication style.38The core technical innovation underpinning state-of-the-art microtargeting, namely harnessing the power of big data, however, comes from the fields of computer science, network analysis, and artificial intelligence. Big data analysis depends upon the convergence of two relatively recent developments. The first is the massive processing and storage power of modern computing; the second is what Viktor Mayer-Schonbergerand Kenneth Cukier call"datafication,"which entails"taking information about all things under the sun—including ones we never used to think of as information at all, such as a person's location, the vibrations of an engine, or the stress of a bridge—and transforming it into a data format to make it quantified."39 Such information is analyzed using powerful algorithms designed to uncover complex hidden correlations amid the myriad data points. 35lbid.,131. 36lbid. "Kathleen Hall Jamieson,"Messages, Micro-targeting, and New Media Technologies," The Forum 11:3 (2013), pp. 434. 38Sosnik, Dowd, and Fournier, Applebee'sAmerica, p. 22. 39Viktor Mayer-Schonberger and Kenneth Cukier, Big Data (Boston: Houghton Mifflin Harcourt, 2013), p. 15. NEW POLITICAL SCIENCE (Q) 73 Big data's ability to comb through enormous amounts of data—in some cases all the apparently potentially relevant data—offers the promise of much greater predictive power compared with the random sampling methods of traditional statistical analysis. The social sciences "have lost their monopoly on making sense of empirical social data, as big-data analysis replaces the highly skilled survey specialists of the past."40 Importantly, such analysis can be conducted absent any theory or hypothesis linking causes and effects. All it requires is a sophisticated algorithm designed to tease out hidden correlations in the sea of data, with no need for instructions on where to look for them. Big data effectively turn the social science sin of "data mining" into a virtue. As such, big data offers—indeed, boasts of—the power to predict without understanding. Heralds of the big data revolution have declared that the need to uncover, or even posit, causal mechanisms is becoming obsolete."The era of big data challenges the way we live and interact with the world. Most strikingly, society will need to shed some of its obsessions for causality in exchange for simple correlations: not knowing why but only what'.'4'1 In fact, for Mayer-Schonberger and Cukier, "[t]he ideal of identifying causal mechanisms is a self-congratulatory illusion; big data overturns this."42 In this sense, big data represents positivism in its most purified form. Explanation simply is identification of constant conjunction. The idea of causation is unnecessary, a relic of the metaphysical period that preceded the positive era in Comte's historical account. Here perhaps we witness the apotheosis of the kind of "unmetaphysical positivism" that Adorno attacked, the"affective realization of an instrumental thought alienated from its object [that] is mediated through its technification."43 In the hands of modern campaign managers, the ultimate aim of microtargeting is prediction and control of voters' behavior, not understanding their views, values, needs, or desires. Understanding is only useful insofar as it helps prediction, and big data has indeed greatly enhanced the predictive power of campaigns'voter models without producing a concomitant increase in understanding. A key benefit that big data offers to campaigns is the ability to find "stranded" partisan or persuadable voters in enemy territory. The campaigns' older, cruder models of voter behavior were based on party identification, voting history, and a handful of sociological and economic factors (annual income, education, gender, and so forth). Big data's super-refined categories, forged out of thousands of lifestyle data points, help identify patterns that predict counterintuitive voting tendencies in particular individuals. Consider the case of Debbie Palos, a middle-class nurse residing in the Detroit, Michigan suburbs, one of the 5.7 million Michigan residents whose demographic and lifestyle data were analyzed by the 2004 Bush campaign's LifeTargeting program. Palos was pro-choice, the daughter of a Teamster, opposed privatization of Social Security, identified as a Democrat, and voted for Bill Clinton in 1992 and 1996.44The older voter models would have predicted that her vote was a lock for John Kerry, Bush's Democratic opponent in the 2004 presidential election. The Bush campaign's big data analysis knew better, predicting that there was a ninety percent probability that Palos would vote for Bush. Based on its analysis of Palos' 40lbi<±, 30. 41lbid.,7. 42lbid.,18. 43Quoted in EmetGellner, The Concept of Kinship (New York: Basil Blackwell Inc., 1987), p. 6. 44Sosnik, Dowd,and Fournier, Applebee'sAmerica, p. 15. 74 @ W. A. GORTON consumer and lifestyle habits, the campaign's LifeTargeting system categorized Palos and Lynn Jensen, a woman with similar lifestyle patterns, as "Terrorism Moderates.""The Bush team ... knew the size of their mortgages, their favorite vacation spots, magazine, music, sporting activities, and virtually every other lifestyle checkpoint that money can buy. Without ever talking to Palos or Jensen, the Bush team knew how they had voted in the past elections and could predict with ninety percent certainty how they would vote in 2004."45 Based on its LifeTargeting modeling, the Bush campaign felt confident enough that Palos would vote for Bush that it directed get-out-the-vote efforts toward her, as well as microtargeted messages designed to resonate with her and other Terrorism Moderates. Previous Republican campaigns, lacking the benefit of big data analysis, likely would not have wasted resources on a voter with Palos' profile. But big data analysis enabled the campaign to locate her and other predicted "stranded partisans." Palos did, in fact, vote for Bush in 2004, but when interviewed a year later, could not say why. "Palos struggled to answer the question 'Why did I vote for Bush? I don't know'.'46 In a sense, the Bush campaign did not know why either, and did not need to. All they needed to know was that, according to the algorithm, the summation of her lifestyle data points more resembled a Republican voter than a Democratic one, surface appearances to the contrary notwithstanding.The campaign could predict her behavior—both how she would vote and what sort of messages would most likely prompt her to vote—without any need to understand what really motivated her voting. Owing solely to their analysis of her lifestyle data trail, the campaign knew that Palos likely shared the political attitudes of other "Terrorism Moderates," but they did not know, and did not need to know, why her data trail indicated that she probably shared the same attitudes as the other 101,200 Michigan voters placed in theTerrorism Moderates category. Based on this analysis, the Bush campaign targeted Palos, Jensen, and other potential voters with similar data profiles with its get-out-the-vote drive and microtargeted messaging. Since the 2004 US presidential election, the sophistication of big data analysis has become vastly more sophisticated, so that campaigns'voter behavior models can in effect dispense with demographic categories altogether. Sophisticated combinatorial analysis now enables campaigns to assign each individual voter a numerical score that predicts the likelihood that he or she can be persuaded to favor a particular candidate and then prompted to vote for him or her or to donate money to the campaign. Bruce Bimber has described how microtargeting increased in sophistication between Barack Obama's 2008 and 2012 presidential campaigns: The practices of 2012 ... represented a leap beyond previous "micro" messaging toward the modeling of multiple behaviors of citizens using dozens of predictor variables at a new scale. Data analytics went far beyond classifying people into demographic subgroups such as"soccer moms"or"NASCAR dads."lt permitted modeling why individual soccer moms and NASCAR dads behave as they do.47 Ken Strasma of the Democratically aligned voter-analysis firm StrategicTelemtry calls the rich data underpinning these scores the voters"'demographic DNA":"The actual combinatorial analyses that we come up with aren't really anything that you could put on a bumper sticker. You know, soccer moms or office park dads. Sometimes people will ask to see the formula, and it comes out to ten thousand pages long."48 45lbi<±, 34, my emphasis. 46lbid.,57. 47Bruce Bimber, "Digital Media in the Obama Campaigns of 2008 and 2012: Adaptation to the Personalized Political Communication Environment," Journal of Information Technology and Politics 11 (2014), pp. 130-150. 48Ryan Lizza,"The Relaunch: Can Barack Obama catch Hillary Clinton?" The New Yorker, November 26,2007. NEW POLITICAL SCIENCE (Q) 75 Framing Theory, Focus Groups, and Finding "Words That Work" A third way that political campaigns use scientific research to manipulate citizens involves the conjunction of framing theory and focus group research. In recent years, campaign consultants have drawn upon framing theory and focus group research to develop phrases, slogans, and a general rhetorical style designed to draw voters to their candidate's or party's side and to get those voters to the polls. Campaigns use focus groups to help them uncover words that activate certain cognitive frames in voters' minds, especially frames that guide their moral thinking. Phrases that have become common currency in American political rhetoric—"climate change,""the death tax,""tax relief,""opportunity society,""common sense reform"—are the product of this type of research. This approach to voter persuasion is potentially manipulative insofar as it seeks to alter voters' beliefs and behaviors by intentionally and precisely targeting their unconscious cognitive processes. When successful, application of framing theory alters a person's beliefs and attitudes without her knowing why or how her beliefs and attitudes have changed. Of course, it hardly needs to be noted that techniques designed to use words to manipulate people predate focus groups and framing theory. Sophists in ancient Greece were accused of teaching their students rhetorical tricks that "made the worse argument appear the stronger." In the Gorgias, Plato's Socrates concludes that rhetoric, as opposed to philosophy, is simply a set of skills for exercising power over an audience and, unlike true philosophy, is unconcerned with truth or justice. Honing rhetorical techniques via modern scientific theory and experimentation holds out the possibility of raising rhetorical manipulation to a level of sophistication and effectiveness that surpasses anything that the ancient Greeks could have imagined. George Lakoff, a distinguished linguist, has written extensively on how verbal framing affects political debates. Lakoff defines frames as "mental structures that shape the way we see the world."49 They are part of the'"cognitive unconscious'—structures in our brains that we cannot consciously access, but know by their consequences: the way we reason and what counts as common sense. We also know frames through language. All words are defined relative to conceptual frames."50 In ordinary life, people usually do not think and reason using deductive logic in terms of the precise taxonomies of modern science. Rather the frames that people use in everyday thought are largely metaphorical and replete with fuzzy, unstable categories. Successful political rhetoric, according to Lakoff, is largely an exercise in using metaphors that will "activate" the frames that generate moral reasoning favorable to one's political aims. Lakoff believes that conservatives in the US have enjoyed electoral success in recent decades owing in part to their superior framing of political issues. Conservatives intuitively understand that winning rhetorical contests entails triggering certain moral emotions by framing political disputes in terms that favor their agenda. In particular, Lakoff contends that conservatives tend to describe political disputes using metaphors that activate voters' desire for order, discipline, and obedience. Liberals (or, rather, "progressives," a term that Lakoff says is less likely to trigger negative frames), in contrast, have been hampered by an outdated view of reason, stemming from what he calls the "Old Enlightenment" view of human thought.This view sees thinking as"conscious, literal, logical, universal, unemotional, 49George Lakoff, Don't Thinkofan Elephant! Know Your Values and Frame the Debate: The Essential Guide for Progressives (White River Junction, VT: Chelsea Green Publishing, 2004), p. xv. 50lbid. 76 @ W. A. GORTON disembodied, and serves self-interest."51 Attempts to persuade the public through direct, conscious appeals to logic and evidence are bound to fail, Lakoff says,"not only because the public's mind is mostly unconscious, metaphorical, and physically affected by stress, [but also] because its brain has been neutrally shaped by past conservative framing."52 What is needed is a "New Enlightenment" understanding of human reason, one that recognizes that the human mind is "largely unconscious, embodied, emotional, empathetic, metaphorical, and only partly universal."53 Informed by this new understanding, progressives would be able to identify and deploy metaphors and frames that will activate moral thinking that favors their core values: empathy, protection, nurturance, and equality. Lakoff believes that he has identified the master metaphor informing both the contemporary American conservative and progressive worldviews, namely the metaphor of the "Nation-as-Family."54 Conservatives and progressives alike tend to think of the nation as a big family, and the way that they conceptualize families in turn has implications for their views on a slew of social and political questions. According to Lakoff, conservatives generally adopt a "Strict Father"model for thinking about politics, whereas progressives tend to think in terms of a"Nurturant Parenf'model.The Strict Father model "posits a traditional nuclear family, with the father having primary responsibility for supporting and protecting the family as well as the authority to set overall policy, to set strict rules for behavior of children, and to enforce the rules."55 In the Nurturant Parent model, "[l]ove, empathy and nurturance are primary, and children become responsible, self-disciplined and self-reliant through being cared for, respected, and caring for others, both in their family and in their community."56 These two models in large measure explain conservatives'and liberals'differing views on authority, law and order, redistribution of wealth, moral desert, and the proper roles of men and women. While the Strict Father model predominates amongst conservatives and the Nurturant Parent model mostly governs progressives'thinking, both models can be detected at work in most individuals, Lakoff says. In fact, persons found nearerto the middle of the American political spectrum are likely to draw upon both models in different contexts. He calls such persons'^iconceptuals."57 Winning theirallegiance requires"activating your model in [such] people."58 This is where science meets political campaigns. Lakoff is an open advocate of using the science of framing to help liberals craft language that will push centrist voters into the liberal camp while also energizing the already progressive electorate. He has published a how-to guide {Don't Think of an Elephant! [2004]) and also helped create the Rockridge Institute, a Washington-based research think-tank, to help them do so.59 A key impetus behind the project is to counter what Lakoff believes are conservatives' more skilled and focused attempts to trigger frames that favor conservative policies and values.Their superior ability to control public debate, he says, relies upon a closely integrated network of well-funded right-wing think tanks, political consultants, ideologically aligned media outlets, and politicians. Progressives can counter this advantage because "we have the advantage 51George Lakoff, The Political Mind: A Cognitive Scientist's Guide to Your Brain and Its Politics (New York: Penguin Books, 2009), p. 2. "Ibid., 42. "Ibid., 13. 54George Lakoff, Moral Politics: How Liberals and Conservatives Think (Chicago:The University of Chicago Press, 2002), p. 13. 55lbid.,33. 56Lakoff, Don't Think of an Elephant!, pp. 33-34. "Lakoff, The Political Mind, pp. 69-73. "Lakoff, Don't Think of an Elephant!, p. 21. 59The institute closed in 2008, citing lack of funding. NEW POLITICAL SCIENCE (Q) 77 of having science on our side.Through cognitive science and through linguistics, we know how [conservatives] did it. And we know how we can do the equivalent for progressives in much shorter time and with many fewer resources."60 Understanding the science can help progressives identify precisely what types of words will trigger progressive-friendly frames. In recent years, a proliferation of private firms have offered to sell their knowledge of cognitive science and framing theory to political campaigns and interest groups. Consider RKM Research and Communications. The firm, staffed by experts with backgrounds in scientific polling, cognitive scientists, and communication theory, boasts that its"lnfluence™"package "gives campaign managers precise insight into the messages that are most compelling to a target audience, as well as insight into the precise vernacular and rhetorical framing that are most compelling (cognitively convincing) and motivational (emotionally appealing)."61 RKM uses "experimental manipulations" involving priming and framing to test the effectiveness of different campaign messages. According to RKM's white paper entitled "Influence™: How to Move Public Opinion to Win a Campaign," their approach "offers a fully representative understanding of the cognitive appeal and emotional power of the message under investigation, and a complete understanding of how the message should be stated, justified and placed within a consistent metaphorical frame for each audience under investigation."62 Interestingly, RKM is aware of the potentially ethically problematic nature of its work: "We recognize the power that Influence™ represents for marketers, politicians and others interested in manipulating public opinion." But the firm concludes:"Our goal is not manipulation, but communication. Understanding how people think and feel about an issue, we believe, will empower campaigns with the insight they need to talk to—not at—the audiences they wish to engage."63 Lakoff, too, is aware that his call to apply science to campaign rhetoric may draw charges of manipulation. His response to the accusation is threefold. First, he views science itself as value neutral. "This kind of language use is a science. Like any science it can be used honestly or harmfully."64 But"[c]ognitive science is, in itself, apolitical."65 Second, he says that that liberals and progressives tend to fall prey to the "Enlightenment myth," the view that "[t]he truth will set us free. If we just tell people the facts, since people are basically rational beings, they'll all reach the right conclusions.'^The myth traces its roots to "an Enlightenment tradition of supposedly literal, rational, issue-oriented discourse, a tradition of debate using 'neutral'conceptual resources."67 But in the real world of public debate, as opposed the ideal speech situation of a philosophy seminar, effective persuasion always entails triggering the unconscious frames that favor one's position. "If the facts do not fit a frame, the frame stays and the facts bounce off."68 Finally, Lakoff denies that intentionally attempting to activate certain frames while suppressing others is intrinsically manipulative. Doing so is not manipulative, he says, provided that the frames accurately express the moral and factual beliefs of the speaker. The frames serve to reinforce what the speaker already believes to be 60Lakoff, Don't Think of an Elephant!, p. 106. 61RKM Research &Communications,"lnfluence™: How to Move Public Opinion to Win a Campaign," p. 7, available online at: "Ibid., 8. 63lbid., 9. 64Lakoff, Don't Think of an Elephant!, p. 23. 65Lakoff, Moral Politics, p. 17. 66Lakoff, Don't Think of an Elephant!, p. 17, emphasis in the original. 67Lakoff, Moral Politics, p. 387. 68lbid. 78 @ W. A. GORTON true.69 He distinguishes his recommendations on how to reframe political discussion from propaganda and "spin.""Propaganda is an attempt to get the public to adopt a frame that is not true and is known not to be true, for the purpose of gaining or maintaining political control"and spin is manipulative because it attempts to re-describe something embarrassing using an "innocent frame."70 By these standards, Lakoff says, Frank Luntz—"the right's language man"—plainly uses framing to manipulate people.71 In the past decade, political campaigns, interest groups, and private corporations have employed Luntz and other similar wordsmiths to try to sell their candidates and favored policies to the public. Luntz's approach combines focus group research with framing theory to identify just what words will woo voters to the Republican side. Sociologist Robert K. Merton developed focus group research in the 1940s while serving as associate director of the Bureau of Applied Social Research at Columbia University.72 His initial focus group studies were done at the behest of the federal government, which was trying to gauge the appeal of different pre-war morale-boosting messages on its radio programs.The approach requires researchers to conduct guided, in-depth discussions with a small group—usually between a half dozen to two dozen individuals—on a chosen topic. A key aim of focus group studies is to tease out information or relationships on the topic that may remain hidden from prepackaged, highly structured opinion surveys. Luntz has used focus groups to try to uncover the precise words or phrases that trigger frames favorable to his Republican candidates. The remarkable uniformity and consistency of Republican politicians' rhetoric is evidence of his influence. The fruits of his research include GOP politicians substituting "climate change" for "global warming,""energy exploration" for "drilling for oil," and "death tax" for "estate tax," which, he boasts, "turn[ed] a relatively arcane issue into a national hot button."73 He is also credited with naming the "Clear Skies Act," a Bush Administration law that weakened air pollution standards.74 Luntz touts the effectiveness of his approach, but, like Lakoff, denies that his work is manipulative. His defense is curious. He approvingly cites Hollywood screenwriter Aaron Sorkin:"There is no difference" between manipulative and non-manipulative language."lt's only when manipulation is obvious, then it's bad manipulation."75 Luntz fails to note the seemingly important fact that Sorkin was referring to language use in writing fiction. Whether Luntz's or Lakoff's work is manipulative by their own criteria, it is clear that the aim of their research is to trigger the "right" response in their audience, rather than uncovering the truth regarding matters of public policy or engaging in genuine dialogue. (Tellingly, the title of Luntz's most recent book is Words that Work, not words that inform, persuade, or enlighten.) In any event, their work may be said to be manipulative in at least this sense: As with the field experiments and microtargeting described above, it has the potential to enable its practitioners to alter voter's beliefs and attitudes unconsciously in fairly precise and predictable ways, for there is no doubt that framing works. Political scientists and communications scholars have studied the effects of frames on political attitudes and behavior exhaustively over the past two decades, using a variety of approaches, including 69Lakoff, Don't Think of an Elephant!, pp. 100-101. 70lbi<±, 100. 71lbid.,22. "Robert K. Merton, The Focused Interview (New York: The Free Press, 1990). "Frank Luntz, Words That Work (New York: Hyperion, 2007), p. xxii. 74Lakoff, Don't Think of an Elephant!, p. 23. "Luntz, Words That Work, p. xix. NEW POLITICAL SCIENCE (Q) 79 content analysis, laboratory experiments, and survey experiments. A consensus finding of such studies is that the frames political elites (politicians, interest groups, the media) use have a powerful effect on the public's views and attitudes with respect to public policy.76 Attitudes on a wide range of public policy issues, including affirmative action, Social Security, the Patriot Act, free speech, and aid to the poor have been shown to be highly susceptible to framing. Moreover,"studies suggest only limited inoculation against framing effects in all but the most knowledgeable members of the public."77 Controlled experiments also show that one-sided frames can have powerful effects on individuals'attitudes with respect to an issue even when individuals are familiar with opposing viewpoints on the matter. "[F] amiliarity with [an] issue prior to the experiment does not appear to have inoculated them against manipulation."78 Conclusion I have argued that political campaigns in recent decades have begun to use findings from the scientific study of human behavior to try to manipulate voters and that these efforts are often highly effective but troubling for democracy. Of course, any social analysis that aims at prediction of human behavior can enable control and manipulation of citizens. So it could be said that the analysis presented here merely shines light on a few particular ways in which behavioral social science facilitates such manipulation, as critical theorists long ago warned it could. However, I contend that the type of social science enabled control of individuals discussed in this article should be seen as particularly troubling, especially to readers sympathetic to the progressive values and aims of this journal. The campaign tactics described above target citizens qua citizens in their formation of political opinions and in their voting behavior, and they often operate by identifying or triggering unconscious processes. Campaign persuaders armed with knowledge gleaned from focus group discussions informed by framing theory know just what rhetorical buttons to push to trigger the desired response from voters. They also know from their randomized field experiments what type of contact and what type of message will most likely prompt people to vote. And, thanks to microtargeting and big data analysis, they know with near pinpoint precision how to identify those voters whom they wish to get to the polls and what sort of messages will resonate with them. Considered singly, perhaps none of these campaign techniques present particular cause for concern. However, used in concert, as they are, they present an unmistakable image of effective manipulation. Voters are largely oblivious to the scientific research that leads election canvassers to their doorsteps, the pop-up political advertisements on their computers, and the political rhetoric that they hear on television and radio. Nor are they likely aware of how the canvasser, advertisement, and rhetoric affect their beliefs and behaviors. Progressive political scientists would do well to reflect upon their implication in these developments. It might be argued that progressives should fight fire with fire—that the techniques described above need not be the sole province of powerful groups that are indifferent or even hostile to genuine democracy. Could not field experiments, microtargeting, big data analysis, and framing theory be used to mobilize and energize groups and individuals 76For a comprehensive review of the literature on the effects of framing on public opinion, see Dennis Chong and Jamie Druckman, "Framing Theory," Annual Review of Political Science 10:1 (2007). 77Chong and Druckman,"Framing Theory,"p. 119. 78lbid. 80 @ W. A. GORTON who seek progressive aims? This is indeed the view endorsed by George Lakoff, at least with respect to the use of framing theory, as was noted in the previous section. However, the reader should recall that Lakoff explicitly rejects what he calls "Old Enlightenment" ideas, which he says include the view that public opinion can be produced by cognition that is "logical,""conscious,"and"universal."While Lakoff is certainly right to draw attention to naive or overly idealistic views of public opinion formation and voting, it does not follow that opinion formation and voter behavior can never be anything more than manufactured. This view is cynical and too pessimistic. Recently, for instance, Jurgen Habermas and Helene Landemore have provided strong arguments that genuinely inclusive and noncoercive deliberation in liberal democracies can be attained in some degree and produce public opinion that is informed and rationally grounded.79,80 Progressives must not abandon the goal of a truly democratic policy, one grounded in core ideas shared by critical theorists from Marx to Habermas: a society governed by reason rather than coercion or deception, whose operations are transparent rather than opaque, and that fosters community and solidarity rather than atomistic individualism. True, these are regulative ideals, never fully attainable in the real world. We should nonetheless remain leery of embracing ideas and techniques that will move us further away from them. Disclosure statement No potential conflict of interest was reported by the author. Notes on contributor William A. Gorton is an associate professor of political science at Alma College in Alma, Michigan. His research interests include critical theory, democratic theory, voting behavior, and the philosophy of social science. He is the author of Karl Popper and the Social Sciences (SUNY Press, 2006), and his research has appeared recently in Public Understanding of Science and the Journal of Political Science Education. 79Jurgen Habermas, "Political Communication in Media Society: Doe Democracy Still Enjoy an Epistemic Dimension? The Impact of Normative Theory on Empirical Research "Commun/caf/on Theory 16 (2006), pp. 411-426. 80Helene Landemore, Democratic Reason (Princeton: Princeton University Press, 2013).