Weapons of Math Destruction: Chapter 10: The Targeted Citizen Summary & Analysis

O’Neil imagines creating a petition for tougher regulations on WMDs and posting it to Facebook. As soon as she hits “send” on the post, the petition belongs to Facebook. The site’s algorithm gets to decide what to do with it and whom to show it to, based on the data it has about each of O’Neil’s “friends.” For many of them who don’t engage with many posts and who never circulate petitions, it’ll get buried in their feeds—but for others, it’ll pop up at the top. Facebook isn’t the “modern town square” it might seem to be. Its powerful algorithms and news-molding infrastructure have allowed it, in many ways, to “game” the U.S. political system.

During the 2010 and 2012 U.S. elections, Facebook created experiments to hone a tool called the “voter megaphone” that would allow people to spread the word about voting. Facebook was encouraging over 61 million American users to get out and vote by leveraging peer pressure against them. At the same time, they were studying how different types of updates influenced voting behavior. Because the profits of companies like Facebook, Google, Apple, Microsoft, Amazon, and Verizon are heavily regulated by government policies, these companies often spend a lot of money lobbying and donating to the political system. Now, they can influence Americans’ political behavior and, as a result, the shape of American government.

Facebook’s grand experiment with the voter megaphone showed that they increased turnout by nearly 350,000 people—a big enough group to swing whole states. Facebook used the initial 2010 experiment to study how our friends’ behavior impacts our own—and in 2012, they took things a step further. For two million politically engaged users, researchers tweaked the algorithm to show these people more news instead of social updates or funny videos. Researchers wanted to see if getting news from friends would change people’s political behavior—and it did. Voter participation in the group went up by three percent.

Editors at news outlets, of course, decide what their readers see, and from what angle they see it. But when a news outlet covers a story, everyone can see it. When Facebook delivers news, the process is mysterious—it’s not the “neutral go-between” it might seem to be. Most users are unaware that the company is tinkering with their news feeds and filtering what they see (and which of their posts their network can see). Research indicates that showing Facebook users positive updates puts them in better moods, while showing them negative updates from friends puts them in worse ones. Emotional states can be transferred through the internet—and through Facebook’s algorithms.

O’Neil doesn’t believe that Facebook’s researchers are actively trying to game the political system. But she does believe that Facebook has the power to determine what people learn about the political system, how they feel about it, and how they participate in civic life as a result. Facebook and Google haven’t yet turned their algorithms into political WMDs, but the “potential for abuse is vast.”

In the spring of 2012, Mitt Romney had all but secured the Republican presidential nomination. He traveled to Florida for a fundraiser at the home of Marc Leder, an investor who’d given over $300,000 to Romney’s campaign. Romney assumed he was walking into a closed setting with a likeminded group of people—but as he let loose during his speech with traditional Republican talking points, he underestimated his audience. One of the caterers captured Romney claiming that 47 percent of the U.S. population were nothing but “takers,” posted it to the internet, and let the world see.

Most politicians tailor their pitches for lots of different subgroups on the campaign trail. This is, essentially, a form of modern consumer marketing that’s driven by carefully gathered data. In the incident at Leder’s house, Romney was speaking based on one set of data—but he ignored that other groups might be in attendance.

Nowadays, Big Data has given politicians lots of powerful tools for targeting “micro-groups” of citizens for votes and donations through carefully honed messages. In July of 2011, the Obama campaign started hiring analytics experts who would help create and target groups of like-minded voters. Rayid Ghani, one of the campaign’s data scientists, had previously worked on projects for a consulting that analyzed grocery stores’ consumer data. This information was used to create customized shopping plans for many kinds of shoppers: coupon-clippers, brand loyalists, foodies, and so on. Now, Ghani was trying to see if similar calculations would work on swing voters (those who aren’t firm supporters of any one candidate or political party).

After conducting deep interviews with a few thousand people from different groups, Rayid and his team set out to find voters who resembled them by sifting through the data and demographics of the people they’d interviewed and creating mathematical profiles of them. Then they scoured national databases to find people with the same profiles so that they could target each group with advertisements to see what metrics appealed to them. By the end of all these tests, the researchers had their group of 15 million swing voters.

Four years later, Hillary Clinton’s campaign would build on the Obama research team’s methodology to create a data system that would let them manage millions of voters. But many of the methods used to gather data that campaigns use aren’t necessarily legitimate. Data firms keep tabs on users’ “likes” and use those to rank them on the scale of “big five” personality traits (openness, conscientiousness, extroversion, agreeableness, and neuroticism). Then they develop targeted ads based on this information. Not all of these methods are useful—but some are, and they’ve essentially turned the voting public into a kind of fluctuating financial market.

Lobbyists and interest groups, too, use these microtargeting tactics to reach more people. This can be dangerous—certain political groups can create targeted advertisements full of false information that spread rapidly. The anti-Obama “birther” campaign got off the ground this way, as has a lot of misinformation about things like abortion and immigration. Even television is moving toward personalized or microtargeted advertising—and as individualized ads become more common, it’ll be harder to understand or even access what our neighbors and friends are seeing. Political marketers have scores of information about us, but we have little about them or their methods.

Microtargeting is vast, largely hidden, and unaccountable or unregulated—so it is, in O’Neil’s estimation, a WMD. And it’s actively undermining and threatening U.S. democracy. Additionally, what’s so frightening about political microtargeting is that it’s not aimed only at the rich or the poor—it’s aimed at everyone except for those who aren’t expected to vote. This creates a feedback loop that keeps uncertain or disenfranchised voters out of the civic system.