How Facebook was used to push voters’ buttons
A common response to the suggestion that a shadowy data firm called Cambridge Analytica harnessed the power of social media to win the Brexit vote for Leave and the White House for Donald Trump has been a determinedly skeptical eye-roll. Everyone has heard dodgy marketers using catchy buzzwords — neuro, nano, quantum — to give some mundane strategy the air of magic. The “psychographic” ploy of using Facebook personality tests to relentlessly target swing voters with propaganda seemed no different.
Critics struck an exasperated, seen-it-all tone to call out “breathless claims about dark arts” and mind control. One technology writer compared Cambridge Analytica’s political clients to stupidly credulous “climate deniers, anti-vaxxers, young Earth Creationists, eugenicists, Islamic extremists and readers of Gwyneth Paltrow’s Goop.” Several skeptics reached all the way back to the aphorism of Gilded Age retailing pioneer John Wanamaker that half the money he spends on advertising is wasted; the trouble is he does not know which half. The implication was clear. Psychographic vote hacking is a scam, and anyone who paid for it is not an evil genius, but a dupe. But the point of Wanamaker’s quip is not that advertising does not work. The point is that a system that only works some of the time can still have an overall effect, even if the reasons are unclear in any specific instance.
“The beginning of wisdom in the study of voting behaviour is to separate what’s going on at the individual level from what it takes to produce a winning outcome, the collective result,” said Harold Clarke, the Canadian editor of Electoral Studies and professor of political economy at the University of Texas at Dallas, who researches polls, elections, referendums and the political climate that moves them.
People generally do not know why they do anything, let alone why they vote. This is one of the foundational insights of psychology. They can rationalize retrospectively, but experiments show people can be reliably manipulated, especially in an emotional context.
As the unfolding Cambridge Analytica story is illustrating, humans are not the unpredictable wild cards of political campaigns. They are the most reliable part of the system, largely because they are slaves to their emotions. This is known as the “hot cognition hypothesis,” that thinking itself is charged with feeling, sensation. Voting is not just emotional, it is also “rapid, instantaneous and involves largely unconscious processes,” as David Patrick Houghton, professor of national security affairs at the U.S. Naval War College, put it in the book Political Psychology. For campaigners, the key to this insight is that not all voters matter. Hacking the vote with Facebook is not about changing everyone’s mind, or even any specific mind. It is about changing enough minds. That, as it turns out, is much easier. Pennsylvania, Wisconsin, Michigan offer a case study. In the 2016 vote, they were all razor-thin votes for Trump, and they won him the White House, illustrating that you do not have to move every vote in order to affect the outcome. What you need to do is find the likely swing voters, and hit them hard. There is a large body of research in cognitive psychology and experimental economics to draw on for such a strategy. Trends emerge when you compare psychology and political behaviour. Right-wingers tend to be tough, persistent, firm, reliable, loyal, orderly, careful, restrained and decisive, while at the same time being rigid, anxious, conformist, prejudiced, stubborn, moralistic and close-minded.
At the other end of the spectrum, left-wingers tend to be enthusiastic, expressive, creative, curious, imaginative, tolerant and free, while also being slovenly, indifferent, unpredictable, eccentric, uncontrolled, individualistic and impulsive. Facebook has known since 2015 that Cambridge Analytica improperly obtained data on 50 million of its users, then failed to abide by an agreement to delete it. A single worker did this via an a pp, this is your digital life, that administered a personality test, ostensibly for academic research, and also harvested details of a user’s friends. Participants gave their consent. Their friends did not.
The app gathered data that linked five major personality traits—openness, con- scientiousness, extroversion, agreeableness and neuroticism — with political inclinations. More than a quarter of a million people did this, creating a database of tens of millions of potential voters, which Cambridge Analytica could then weaponize for use in its political consultancy.