Over the past several decades, governments and corporations have found that gathering data about their citizens, customers, and employees lets them manage people more efficiently. As a result, much of the infrastructure of society now relies on reducing the person to a set of behaviors, attributes, or opinions. This history — and its troubling consequences — has been well-documented in recent years. What has been less recognized are the insidious ways that individuals are increasingly encouraged to apply these same methods to better manage their own social and personal lives.
In his new book, Breaking the Social Media Prism: How to Make Our Platforms Less Polarizing, Chris Bail offers needed insights into the distortions that result when human persons are reduced to a set of data points. Inadvertently, it also has much to show us about why the most oft-discussed solutions for fixing social media seem so unequal to the problem, and why the massive investment of public attention to these platforms has yet to amount to much.
Bail, a sociologist who directs Duke University’s Polarization Lab, begins the book by debunking a common piece of advice pundits offer for depolarizing social media: diversify your news feed. Bail and his colleagues designed an experiment to get Twitter users out of their partisan echo chambers by having them follow bots that would retweet posts from voices on the opposite side of the aisle. The result, perhaps surprisingly, was that those who learned more about the views of the other side wound up having their own partisan commitments strengthened. Diversifying news feeds did not fix polarization; if anything, it increased it.
Much of Bail’s book attempts to make sense of this. His explanation draws from a series of studies designed to help us perceive the gap between real persons and the representations they project on social media.
To understand this gap, Bail and his fellow researchers conducted in-depth interviews while also using quantitative measures of individuals’ political opinions and social media posts. The contrast between the pictures these different methods provide is often startling. One woman they talked to described herself as “in the middle” politically, but it turned out she usually voted for Democratic candidates, even though her views on particular issues better aligned with the platform of the Republican Party. Further conversation revealed that she didn’t really have firm views on a lot of issues; like many Americans, she had more important things to worry about than developing strong political opinions. But after following a bot that retweeted conservative accounts, she articulated more-progressive opinions on a range of issues.
Bail and his team found this to be a common result: “Unenthusiastic Democrats and Republicans who followed our bots…. not only came to identify more strongly with their political party, but they developed new partisan views about issues that they had previously known little about.” Bail surmises that this is because “unenthusiastic partisans … do not carefully review new information about politics when they are exposed to opposing views on social media and adapt their views accordingly. Instead, they experience stepping outside their echo chamber as an attack upon their identity.” They discovered “that there was a war going on,” and felt they had to “choose a side.” It seems that when we see other people talking about politics on social media and forging their identity based on politics, we are prompted to think of our own identities and communities in terms of political categories as well.
Why do we do this? “In an era of growing social isolation,” Bail observes, “social media platforms have become one of the most important tools we use to understand ourselves — and each other.” But they are not a mirror that accurately represents us. They are “more like a prism that refracts our identities.” They leave us “with a distorted understanding of each other, and ourselves.” The resulting gap between who we are in person and who we are online continues to grow. This gap, Bail concludes, is “one of the most powerful sources of political polarization in our era.”
Bail’s argument about social media needs to be set within the broader context of how social scientists have used big data to measure and manipulate public opinion. Much has been written about this history, but the recent and accessible If Then: How the Simulmatics Corporation Invented the Future, by Harvard historian Jill Lepore, is a good place to begin. Lepore chronicles how a group of marketers, data scientists, and computer programmers founded the Simulmatics Corporation in the 1950s to model political and consumer sentiment. She summarizes their ambition this way:
If they could collect enough data about enough people and feed it into a machine, everything, one day, might be predictable, and everyone, every human mind, simulated, each act anticipated, automatically, and even driven and directed, by targeted messages as unerring as missiles.
Drawing on survey data, they organized American voters into 480 types, and then produced a model to predict how a candidate’s policy positions would affect the outcome of the election. For instance, they ran simulations showing that if John F. Kennedy foregrounded his religious convictions, then he would gain voters in the states that mattered, while losing relatively few to anti-Catholic bias. Similarly, they predicted that if he took a strong stance favoring civil rights, then he would improve his odds in the Electoral College. Lepore notes that, while it’s not clear how much these computer simulations influenced the decisions the Kennedy campaign made during the summer of 1960, the course they wound up taking was largely the same as the one advised by Simulmatics. Kennedy, of course, won the electoral vote handily.
The ambitions of the Simulmatics Corporation seem quaint in light of today’s big data analytics. Facebook’s advertising system makes demographic targeting easy. Old-school retail politics — kissing babies and shaking hands and going to the fair — remain necessary only as a backdrop for photos that can then be used in various ads.
And it’s not just the political realm where data analytics is replacing human interactions. Instead of buying books based on recommendations from a shop owner or trusted reviewer, consumers look to books recommended by Amazon’s algorithm. Instead of getting a loan from a local bank based on a lender’s personal assessment of the financial risks, software evaluates businesses and individuals based on credit scores and other data points. Instead of being assessed by a social worker, disabled people find out what assistance they are eligible for based on some inscrutable algorithm. All such decision-making relies on digital data and if-then processing: If you bought this book, then you will also want this one; if you are of this race or age or gender or religion, then you will hold this political opinion; if you buy a house in this ZIP code, then you will repay your mortgage.
The algorithms and customized predictions at work today enjoy a sophistication the data scientists behind Simulmatics would have drooled over. But now, as then, politicians and corporations find it advantageous to interact with digital representations of voters, consumers, and users.
This is all part of a broader, ongoing social transition from real to digital, from full-orbed, enfleshed, tactile complexity to encoded approximations. Email representations of a conversation, MP3 recordings of live music, Zoomed classes, smoothly curved roadways represented as a series of very small straight lines on Google Maps — all these binary approximations serve real goods. But they reduce complexity and thereby distort our understanding of reality. They are not lossless substitutes for the real thing; they increasingly color our perceptions and expectations as we engage in real life.
One way to make sense of Bail’s “social media prism,” then, is to say that our feeds and timelines have made us all social scientists, aspiring Simulmatics analysts. Social scientists don’t study actual people; they study the data trails left behind by people and by their interactions. Likewise, all of us are now invited to relate to others — and ourselves — based on digital approximations rather than real-world interactions. As inhabitants of a digitally mediated world, we all find ourselves like Simulmatics data scientists, trying to make sense of our neighbors based on scraps of data gleaned from what they post on Facegramitterok.
But what they post is a warped representation both of themselves and of society at large. The prismatic distortion of social media, Bail argues, mutes moderates and amplifies extremists. The result is that “social media has sent false polarization into hyperdrive.” Bail explores several facets of this distortion, but the gist is that political extremists tend to have weaker social bonds in their real-world lives and so look to online interactions to find purpose and community.
Since extreme claims and insults generate more responses on social media, participants gain social rewards for staking out strong positions — rewards they are likely to lack in their regular lives. Bail observes that social media can be so addictive precisely because it enables users to iteratively curate an identity:
It makes it so much easier for us to do what is all too human: perform different identities, observe how other people react, and update our presentation of self to make us feel like we belong.
Online interactions now are tinged with the methodology of social science, the if-then thinking based on identities and reductive assumptions. Social media becomes a kind of democratized mass exercise in social science.
Perhaps you think you are immune to this, but the social media feed is a virulent parasite, and it infects the way we approach even our real-life relationships: The question “If I post this, what responses will I get?” soon informs and guides “If I say this, what responses will I get?” We can beta-test and fine-tune online, and the results shape how we conduct conversations in person.
As L. M. Sacasas has recently argued in his newsletter, “The timeline now tends to be the principal point of mediation between the world out there and our perception of it.” The grammar of social media — what Marshall McLuhan would call its “patterns of perception” — is becoming the grammar of all social relations. Even those who are never or rarely on social media still find themselves shaped by the distortions and practices of these platforms.
If the content of our in-person conversations with others is drawn from the medium of the timeline, our relationships will be polarized and flattened. If we assess the state of public discourse and political sentiment based on our reading of social media representations, we will give too much weight to the significance of our neighbors’ political opinions on a narrow range of hot-button issues. Instead of relating to human persons, we will interact with identity categories. Instead of seeing people in real time, we will see them as if illuminated by a slow strobe light, or as if they were in an old film reel with few frames per second: Our brains can give us the illusion that we are seeing the full range of motion, but in reality we are constructing images of others based on discrete posts, positions, and identity performances. It’s not merely that we are caught in systems that treat us as digital representations of our real selves; it’s that we now are formed and habituated to relate to digital approximations of our fellow citizens, our friends and family members, and even ourselves.
What can we do? Bail rightly rejects several commonly offered solutions. Simply opting out of social media doesn’t insulate us from its effects; it’s now too deeply embedded in our social and political lives. Existing platforms can’t just be tweaked to ameliorate their harms — the fundamental problems, as we’ve seen, can’t be addressed by merely blocking bad actors or tweaking algorithms or opening up echo chambers.
Bail gives some advice for how we can use the existing platforms more wisely — what he calls “hacking the prism.” To some extent, recognizing and naming the reality of the prism can reduce its power to define our social understanding. In addition to seeing the distorting effects of the prism, Bail urges moderates to remain active on the platforms and model better forms of discourse. Instead of “stepping boldly outside your echo chamber,” which is likely to reinforce negative polarization, Bail recommends taking “baby steps” — for example, following thoughtful people who share some of your core convictions but come from different perspectives or draw different conclusions. Other ways Bail offers to hack the prism include listening more than we post and learning to frame our political positions in ways that resonate with our political opponents.
Yet given the power of social media to shape our perceptions, how many of us can follow these recommendations and resist being warped by the prism? In light of this reality, Bail and his colleagues also experiment with a new platform that rewards people for having thoughtful, issue-focused discussions. The problem is that the anonymity and narrow political focus of this new platform make it unlikely to become popular, or to encourage its users to relate to one another as persons — or even friends — rather than as potential converts to the “right” political views. Bail’s solutions remain underwhelming, unequal to the nature of the problem we’ve just heard described.
It is a recurring theme of debates about social media over the last five years: sweeping, terrifying diagnoses of how it is altering the very fabric of American social life, followed by meliorist, technocratic fixes. Twitter locks or permanently suspends individual accounts, and Facebook’s Oversight Board provides the impression that the platform is serious about making content decisions, but neither platform is interested in reducing its reach into our personal and social lives. Recent leaks from Facebook indicate that for years the company weighted the emotionally expressive emoji reactions — such as “love,” “wow,” and “angry” — five times more heavily than plain “likes” in selecting which posts to bump to the top of your feed. Rejiggering this algorithm might help around the edges, but simply favoring posts that get the more anodyne “like” won’t fix what ails Facebook. The same goes for other often-offered solutions such as banning the infinite scroll, or removing “streaks” from Snapchat. As the recently announced name change from Facebook to Meta indicates, the company’s ambition remains to mediate all human “connections” through its apps. The prism continues to grow more totalizing.
While Bail would surely critique Mark Zuckerberg’s new vision, the trouble is that his book remains within the limits set by the social science paradigm. Why hasn’t such intense focus from reformers produced more compelling ideas? Part of the problem is surely that the amount of money generated by these platforms makes them resistant to any reform. But the more vexing question, I think, is whether would-be reformers really succeed in escaping the prism.
To Bail, the answer to the woes social media has unleashed can be found by doing more social science:
As we unlock the keys to make our platforms less polarizing, we can use insights from social science to make them a reality…. We must build the methods of empirical observation of human behavior into the architecture of our platforms.
The metaphor here is garbled — I don’t know what it means to unlock keys or to make them (or the insights?) a reality. But what Bail is insisting on is that social science can heal the disorders caused by our relating to one another like social scientists.
The reformist conceit goes something like this: Social media platforms were built based on a particular view of human nature that has since been falsified. The real-world testing in which we have all participated over the last decade-plus shows the consequences of these inaccuracies. Now, social science has revealed what the flaws were. Armed with this more accurate view of how people behave, we can apply it to build a new generation of better platforms. Mark Zuckerberg as a Harvard freshman could hardly have sounded more optimistic about his own new project.
One of Bail’s key findings is that if we judge the state of American political sentiment based on what our fellow citizens post on social media, we will have a skewed perception of reality. Recall the apathetic partisans who, when exposed to people from the other side on social media, came to believe that there is a war on. In fact, Bail finds, “the extent to which people exaggerated the ideological extremism of people from the other party … was significantly larger among those who used social media to get their news.” But then shouldn’t we expect social scientists’ studies of social media to suffer the same distortion?
As Bail demonstrates, the problem is not merely that the representation of other people we get on social media is a limited approximation of their real-world self, but that it is a warped approximation. There is a kind of feedback loop at work. If I meet a new neighbor and strike up a conversation, it would be uncouth if the first questions I ask are whether he supports Black Lives Matter, what his opinion is on immigration, or how high he thinks the corporate tax rate should be. Yet it’s all too normal to see people framing their identities on social media in terms of their opinions regarding political and cultural issues. Instead of revolving around the weather, neighborhood happenings, high school sports, or how the fishing’s been, social media discourse tends to fixate on national or international events — that thin stratum of events we all putatively share.
But striking up conversations with strangers to ask about their political views is pretty much the job description of the social scientist. It is the central tool of Bail’s book. Social science carries with it an entire conception of how we make sense of other people, a conception that frames the conversations mediated through social media and also the conversations we have about them. What ground the lens of the social media prism was the treatment of particular, complex human persons as the sum of their political opinions, consumer choices, and identity markers like religion, race, gender, sexual preference, and ZIP code. It seems dubious that like will cure like in this case.
If there’s no technocratic fix for this digitally mediated world, and if we can’t escape its polarizing distortions, where does that leave us? I am reminded of W. H. Auden’s humorous and incisive poem “Under Which Lyre,” in which he warned about the impending dominance of “Pompous Apollo,” the god of the social sciences and their imposed order. What makes Apollo dangerous is the way that he takes over all aspects of thought and self:
If he would leave the self alone,
Apollo’s welcome to the throne.
But as Bail’s book demonstrates, the Apollo of social science seeks to rule the self as well. Auden concludes with a witty “Hermetic Decalogue” that those who wish to oppose Apollo might follow. The most famous rule in this list is
… Thou shalt not sit
With statisticians nor commit
A social science.
But it also contains other good advice:
Thou shalt not answer questionnaires
Or quizzes upon World-Affairs.
Auden’s tone in this poem is akin to that of the Mad Farmer in Wendell Berry’s poem “Manifesto: The Mad Farmer Liberation Front,” who warns:
… Your mind will be punched in a card
and shut away in a little drawer.
When they want you to buy something
they will call you. When they want you
to die for profit they will let you know.
What can we do in response to such a world? What can we do when we are not only its victims but also its perpetrators? The Mad Farmer’s advice is as good a place to start as any:
So, friends, every day do something
that won’t compute. Love the Lord.
Love the world. Work for nothing.
The only way to begin healing the damages caused by distorted digital approximations is to return deliberately to the full-orbed complexity of life in the flesh. Physical life and its practices and communities provide the formation we need to counteract the two-dimensional world of the social media timeline. This is the kind of recovery that David Sax chronicles in The Revenge of Analog or that Matthew Crawford commends in The World Beyond Your Head. Even in a world mediated by the timeline we can follow Robert Farrar Capon in preparing lamb for eight and hosting a dinner party.
Such examples remind us that instead of committing a social science we might, as my friend David Lyle Jeffrey puts it, commit random acts of poetry. These acts of creating beauty confound the if-then logic that threatens to flatten the possibilities we can imagine. Such acts do not constitute a “solution” that would be recognizable by a social scientist or a computer programmer or a government regulator or a misinformation reformer. In this context, however, the most effective responses may be unexpected and unrecognizable — in the parlance of political scientist James C. Scott, they will be illegible.
I’m reminded of a story that Berry likes to tell about his friends Harlan and Anna Hubbard. They lived along the Ohio River on a small, off-the-grid homestead where Harlan pursued his painting and writing and together they made music and hosted many visitors. When the Marble Hill Nuclear Power Station was slated to be built across the river from them, Berry assumed the Hubbards would participate in the local protests against it. He was disappointed that they never even spoke about the plant. Eventually, however, he came to recognize the nature of their response:
I understood that by the life they led Harlan and Anna had opposed the power plant longer than any of us, and not because they had been or ever would be its “opponents.” They were opposed to it because they were opposite to it, because their way of life joined them to everything in the world that was opposite to it. What could be more radically or effectively opposite to a power plant than to live abundantly with no need for electricity?
Similarly, the most effective responses to the distortions of social media will be found in ways of life that provide occasions to relate to one another as family members, friends, and neighbors instead of digital avatars. The social scientists can’t engineer a way out of the digitally distorted world we inhabit, but all of us can work to recover the affection, mirth, and delight found in the various and unpredictable goods of the physical world. We might begin simply by laughing with a friend. After all, as the Mad Farmer reminds us, “laughter is immeasurable.”
How Tech Reform Diminishes Us