It's become commonplace to speak about online political culture as a set of echo chambers that only serve to reinforce our existing views — but is that really the best way to understand it? Absolutely not, says Duke sociologist Chris Bail in his new book, "Breaking the Social Media Prism: How to Make Our Platforms Less Polarizing."
The echo chamber metaphor tells us something about what's happening online, but not enough to guide us toward discovering solutions, Bail argues. In fact, it can be misleading, because breaking people out of echo chambers doesn't make them less polarized — it does the opposite, as Bail discovered with the first experiment on the pathway that led to this book.
What's more, online polarization is only one facet of the problem. "I believe the rapidly growing gap between social media and real life is one of the most powerful sources of political polarization in our era," Bail writes. You could even argue that people are internally polarized between their online and offline selves — though in very different ways for different sorts of people.
Bail's first book, "Terrified: How Anti-Muslim Fringe Organizations Became Mainstream" was a nuanced exploration of how a tiny handful of fringe anti-Muslim organizations managed to hijack the public discourse about Islam (Salon interview here.) Innovative explorations of online data were a central part of his analysis, along with a diverse mix of other approaches. His new book deals with generalized versions of many of the same themes explored in "Terrified," using a similarly mixed-method approach to gathering data — most notably via in-depth interviews with subjects of online experiments, whose results in turn are compared with a wide range of other research. But what's most telling is Bail's central insight into the root of the problem.
"We use social media platforms as if they were a giant mirror that can help us understand our place within society," he writes. "But they are more like prisms that bend and refract our social environment — disturbing our sense of ourselves, and each other." While more attention has been focused on the polarizing dynamics of extremists, theirs is not the only story that matters.
"The most pernicious effects of the prism operate upon the far larger group of social media users who are appalled by online extremism and eager to find middle ground," Bail observes. We are also misled by perceptions of much greater polarization than actually exists, a "feedback loop between the social media prism and false polarization," as Bail puts it. "One of the most important messages I'd like readers to take away from this book," he writes, "is that social media has sent false polarization into hyperdrive."
What's happening isn't necessarily something new and strange, Bail argues, just because the setting may be.
"We are addicted to social media not because it provides us with flashy eye candy or endless distractions, but because it helps us do something we humans are hard-wired to do: present different versions of ourselves, observe what other people think of them, revise our identities accordingly," Bail writes. This applies to moderates as well as extremists. "Although we scan our social environment, consciously or unconsciously, we are often quite wrong about what other people think," he continues — and the distorting prism of social media only compounds this problem.
The prism metaphor represents a shift in analytic frameworks so clarifying and compelling that it reminds me of the Copernican revolution. "Our focus on Silicon Valley obscures a much more unsettling truth: the root source of political tribalism on social media lies deep inside ourselves," Bail argues. Echo chambers are still there, just as the Moon still revolves around the Earth, but the larger landscape has been radically transformed, and things fit together in promising new ways.
Bail spends his first three chapters dealing with "the legend of the echo chamber" and what happens when we break out of it, and then looking at typical extremists and moderates, before focusing squarely on the social media prism itself.
Offering a thumbnail diagnosis, Bail says, "The social media prism fuels status-seeking extremists, mutes moderates who think there is little to be gained by discussing politics on social media and leaves most of us with profound misgivings about those on the other side, and even the scope of polarization itself." ("Status-seeking extremists," it should be noted, are not the same thing as strong partisans. How and why their views are held sets them apart.)
Getting rid of social media is unrealistic, he argues — it's become too much a part of our lives. But there are both bottom-up and top-down ways of reshaping our online experience. We can all make our own online experience more consensus-seeking, rather than divisive, and entire social media platforms could shift incentives — or new platforms could be intentionally created for that purpose.
Every one of Bail's chapters threads together multiple lines of thought — some dating back decades or centuries — interweaving the frontiers of online social science research with the traditions they emerge from. In the first chapter, he highlights the origins of social network research in the late 1940s with sociologists Paul Lazarsfeld and Robert Merton, for example, and the origins of the term "echo chamber" in a 1960s book by political scientist V.O. Key.
In Chapter 4, he goes back even further. "One of the most ancient ideas in Western thought is that rational deliberation will produce better societies," Bail notes, an "idea [that] gained momentum during the Enlightenment." It's a noble idea with a persistent vaporware problem. Bail gives a lightning-quick description, passing from French salon culture through Jürgen Habermas' account of mass communications as echoing aspects of salon culture in a newly-created mass public and on to early internet fantasies of realizing salon culture online, which he describes as "a heavily idealized vision" that "may now seem whimsical," but whose basic logic "continues to motivate many technology leaders." (Note the name of this online publication, founded in 1995.)
As Bail puts it, our experience suggests that "social media are less like an eighteenth century salon and more like a sprawling football field on which our instincts are guided by the color of our uniforms instead of our prefrontal cortexes." There are ways to mitigate the situation, as he argues in his last two chapters. But doing so requires a lot of careful rethinking about the behavior, motivations and perceptions of both extremists and moderates.
Destructive extremist trolls do a lot to drive polarization, rooted in their own sense of powerlessness. "Many people with strong partisan views do not participate in such destructive behavior," Bail writes. "But the people who do often act this way because they feel marginalized, lonely, or disempowered in their offline lives. Social media offer such social outcasts another path."
In a section titled "Lonely Trolls," Bail notes that one such extremist "repeatedly mentioned that he had 'a couple thousand quote followers,' and he was truly proud to court count several prominent conservative leaders among them." But it turned out this person "only had about 200 followers," and "the high-profile conservatives he thought were following him we're actually people with copycat accounts."
Though isolated in their offline lives, trolls often coordinate with one another online, including launching attacks on their perceived enemies, which "serve a ritual function that pushes extremists closer together." Some extremists are political converts, particularly keen to prove their new loyalties as a kind of ongoing purification ritual.
Another purification ritual that extremists of all sorts engage in is to attack moderates on their own side. What's more, some closely monitor their followers, and can be even more savage in attacking anyone who stops following them. This leads to a broader comment about cult-like dynamics. "Proving one membership in a cult often becomes a sort of ritual," Bail writes, "in which members reward each other for taking increasingly extreme positions to prove their loyalty to the cause."
Bail concludes his chapter on extremists by identifying two interrelated processes driving such radicalization: It normalizes extremism on one's own side and exaggerates that on the other side. The more intensely extremists interact with each other, the easier it becomes to believe that everyone thinks that way. Thus, Bail writes, "At the same time that the prism makes one's own extremism reasonable — or even normal — it makes the other side seem more aggressive, extreme, and uncivil."
But extremists are only part of the story, Bail argues. "The most pernicious effects of the prism operate upon the far larger group of social media users who are appalled by online extremism and eager to find middle ground." The overrepresentation of extremists doesn't just drown out the voices of such "moderates," but discourages them from speaking up in the first place — not just for fear of attack by extremists, but also for fear of being mistaken for extremists themselves. Most people care more about social relations than they do about politics — particularly national politics. "Moderates Have Too Much to Lose," as one of Bail's sections is titled.
So the decision not to engage with politics online is a perfectly rational one for the vast majority of people. But it doesn't have to be, if the online experience can be changed. What's central to doing that is disrupting the aforementioned feedback loop between social media and false polarization. In the chapter dealing with bottom-up approaches, Bail describes three learning strategies "to hack the social media prism." First comes learning to see and understand how the prism distorts both our own identities and other people's. Second is learning to see ourselves through the prism and to monitor how our behavior gives the prism its power. Third is learning how to break the prism by changing those behaviors, replacing them with more productive ways of engaging with ideological allies and opponents alike.
It's a challenging task, but recent social science research suggests it's more doable than you might think. Bail and his colleagues have spent years developing new tools to help facilitate the process (available at Duke's Polarization Lab.).
One of the simplest tools is the "Troll-O-Meter": Answer six questions about an account and you can calculate the probability that you're dealing with a troll. Further help is offered with a chart of the most common terms used by political trolls over the last three years and the advice, "Take a look through the last dozen tweets of the person you think might be trolling you."
Users are invited to "Check out our tools for identifying and connecting with moderates who do not share your political views, as well as our issue-tracker that identifies the topics where research indicates you are most likely to find compromise."
Bail's discussion in the book, as well as the online instructions and explanations, help explain the logic of the approach, but three insights are worth highlighting. First is the concept of a "latitude of acceptance," meaning a range of attitudes one finds reasonable, even if one might not initially agree with them. Encountering ideas within one's latitude of acceptance makes one more likely to engage, and perhaps even end up agreeing.
Second is the value of listening. Rather than just jumping in feet first, Bail says, "Take some time to study what those people care about and, more importantly, how they talk about it." Arguments that resonate with the worldviews of others are inherently more persuasive. Third is to avoid talking about polarizing opinion leaders. People have low confidence in leaders generally, and such conversations tend to divert attention from ideas and issues back to identities.
While these bottom-up strategies can improve online discourse, in his last chapter Bail argues that "the only way we can create lasting improvement is to create a new playing field." This might seem improbable given the dominance of Facebook and Twitter, but "taking the long view teaches us that platforms come and go," he writes, and he's not looking for a new behemoth.
"I think there is room for a new platform for political discussion," Bail argues. "Would everyone use it? Of course not." But the social science is clear: "Most people get their opinions about politics from friends, family members or colleagues who proactively seek information about politics, regularly engage with others about such information, and care enough about issues to try and influence people in their social networks who trust their opinion."
Bail is agnostic about how such a platform might be created, but does discuss an experimental effort to explore how such a platform might work: an anonymous issue-based discussion forum that proved both depolarizing and enjoyable for participants. Whether that could be scaled up as a business, nonprofit or government-funded entity remains to be seen. But the basic principle seems clearly established, and the need is inarguable.
Existing social media platforms are politically dysfunctional because they were never supposed to be otherwise. "What's the purpose of Facebook?" Bail asks. "The company calls its mission is to 'bring the world closer together' but the platform began as a sophomoric tool that Harvard undergraduates used to read each other's physical attractiveness." And other platforms, such as Twitter and Instagram, had equally banal beginnings. No one's really tried to build a platform that would actively and intentionally promote the practice of democracy. With the clarion call of this book, perhaps that may change.
But will that be enough? Bail's analysis of the problem of online polarization is clarifying and compelling, but it's not the only mega-problem facing us, and I couldn't stop thinking about that as I read the book. Nor could I ignore other efforts to build consensus and strengthen dialogic politics, including ones I've written about before, such as "deliberative polling" with James Fishkin, or "citizens' assemblies" with Claudia Chwalisz.
So I had some questions to ask Bail about how his work fits into the larger framework of problems and possibilities facing us today. This supplementary interview, conducted by email, has been edited for length and clarity.
You analyze our current social media environment and point to ways it could be made less polarizing and more conducive to good government. Your analysis focuses on polarization as a group-identity based problem. While it seems reasonable that reducing polarization is necessary, it's not necessarily a sufficient condition for a healthy democracy. Reducing polarization after the Civil War led us to three generations of white supremacy under Jim Crow. You're advancing a major rethinking of online political culture, and it seems crucial to address that.
I am certainly aware of the broader public debate about the place of polarization, vis-a-vis other pressing social issues. I have a few general concerns about approaching this issue in a zero-sum manner. The first is that there are almost no counterfactuals that can help us realistically understand the effect of depolarization efforts on societal well being. We cannot analyze an alternative reality where the Civil War didn't happen, or where subsequent depolarization efforts did not happen. Also, it is nearly possible to tease out the impact of those efforts from the many other sources of social malaise at the time — to give only two examples, economic factors related to the restructuring of the economy of the U.S. South, or the long-term impact of war.
My second general concern when people wonder whether polarization is really a pressing social concern is: What is the alternative? Many of the most pressing challenges of our era — changing beliefs about race or the climate, for example, are not simply questions about passing legislation; they are fundamentally about winning hearts and minds. In other words, I worry too often that we are equating polarization with voting alone, and not the broader set of issues that determine what kind of country we aspire to be, or the value of social cohesion more broadly.
My third general concern (which sort of creeps into one of your other questions below), is that people are far too quick to equate Republican elected officials with Republican voters. There is quite a bit of evidence that many Republicans hold beliefs about issues as varied as background checks for handguns and the minimum wage that are far away from those of their leaders. This is why I took such care to discuss the "missing moderates" on the Republican side on social media — people like Sara Rendon.
My fourth general concern is that people too quickly equate depolarization efforts with compromise. Attempting to engage with the other side need not result in caving in on the issues that one is passionate about. I believe there is an intrinsic value to mutual understanding in democracy, even if it is not as vital as some of the early theories of democracy might have believed.
I certainly do not want to paint too rosy a picture here.There are extremely concerning developments in U.S. politics which mean there will be no easy fixes to the many issues that confront us. However, I do often worry that the sudden turn against depolarization efforts on the left will be counterproductive, and ultimately make it more difficult to create the lasting social change that so many Democrats want.
Relatedly, there needs to be some form of reality testing. Climate change is real, just as COVID is. (It didn't disappear on Nov. 4, as Donald Trump predicted.) A healthy online political culture that gets us all killed because it ignores reality doesn't seem fully thought out.
I personally agree with this point. But I also think it is dangerous to assume that one party is completely against reality. This is certainly true of many Republican leaders, and it is also true at the extremes of the Republican Party. But most of the data that I have seen indicates Republicans were in fact very worried about COVID. Perhaps not quite as much as Democrats, but — particularly in the early days of the pandemic — the partisan gaps in concern were fairly small, even if they eventually grew over time.
On global warming, it is also dangerous to equate skepticism about, say, the Paris Climate Agreement with concern about climate change. Many of the Republicans I studied over the past few years were in fact concerned about climate change (and believed it was real), but skeptical that the government could do anything to stop it. By the way, there is also evidence that as many as 40% of Republicans believe "the federal government is doing too little to reduce the effects of climate change" (see, for example, this Pew report).
I think the debates about voter fraud are perhaps more concerning indicators of the potential of partisan differences in the definition of reality. Many of the studies done so far, however, use relatively imprecise wording that, in my opinion, make it difficult to parse people who are genuinely convinced that voting fraud happened from people who are simply upset or displeased about the result of the election (especially given growing evidence of expressive response to surveys among Republicans). This article captures my views on this pretty well — once we start to focus in on the people who really, sincerely believe that voter fraud happens, it might be much less concerning.
Your own data — along with other data on Congress, or on ideological and partisan alignment — shows that political polarization is asymmetric. It's true that "both sides do it" and also true that both sides do it at least somewhat differently. How does this affect your analysis?
The goal of my book was not to explain who is responsible for polarization, but to document how social media shapes the process. A proper analysis of the several decades of asymmetric polarization that you describe would require a much deeper historical analysis. I recommend Matt Grossmann's book "Asymmetric Politics" on this point.
You mention James Fishkin's work in passing, the main thrust of which is that reliance on rationality has been, shall we say, naive. But I see your work as pretty much in the same bin, albeit more ambitious. How is what you're trying to achieve different, or is it complementary?
My concern with Fishkin's argument is that rational deliberation alone will produce consensus. I think it leaves out the role of identity and status in shaping inter-group deliberation. I do not think it is possible to have rational deliberation on social media, at scale, until we learn to recognize how identity and status shape the process of deliberation. Even then, I am only cautiously optimistic.
One might say the point of your book is to argue for intentionally designing our online platforms to serve a collaborative public good, instead of our existing unplanned environment. But aren't conservatives already moving toward an intentionally polarizing alternative? How does this complicate the path toward a more healthy online public square?
I assume you are referring to Parler? It's not clear to me at this point that Parler will survive. The analyses I've seen so far indicate that it is mostly Republicans with extreme views who moved to that platform, and that overall growth has stalled. I have not done careful empirical analysis of this issue, however, so I would point you towards the first few working papers that have come out about Parler from the Stanford Internet Observatory and a lab at Boston University.