Echo Chambers Don't Cause Polarization

Out-group animosity makes exposure to opposing ideas unattractive.

Recently, Tim Keller reviewed Duke professor Chris Bail’s book Breaking the Social Media Prism. I read Bail’s book earlier this year, and I thought it was good, even though I quibbled with some points here and there. Bail, and Keller, dig into how social media is more about building identities than sharing ideas, and their digging is insightful. In the end, Keller is celebrates Bail and others’ pursuit of, as Keller calls it, “a social media platform in which ideas rather than identities can actually be debated.”

I celebrate these efforts, too, even if I do think the efforts are a bit quixotic. I think such a use of social media runs contrary to the entire medium of the social internet at its core. It’s like wanting to build a beautiful yacht when you live in the middle of the desert—the yacht sounds amazing, but it’s sort of doomed to fail despite its beauty because of the geographical context of its construction.

The Root of Polarization

Perhaps the most insightful bit of Bail’s book is on the root of polarization and the reality that the polarization we all feel isn’t fruit of echo chambers. That’s the logical explanation that many, including myself, have reasoned for years. “The reason we’re so polarized is because we’re closed off in ideological tribes and not interacting with people different from us,” has long been the mentality of many who study social media and its wide-ranging sociological effects.

But now we have data that shows otherwise, and we’ll examine that. First, here’s Keller on Bail’s work regarding this idea:

Bail starts with the problem of social and political polarization and asks how social media contribute to it. The common answer is that algorithms keep us in echo chambers or bubbles where we only hear news and opinions from our own side, and this drives division and extremism. But Bail points to research showing that, on the contrary, daily exposure to opposing political and cultural views (and not just to the nasty, caustic versions of those views) only makes people stronger in their views or even more extreme. People who regularly listened to the opposite opinions did not adjust their views and become more balanced or moderate because for many people social media have become places where they are curating a self. And therefore they see opposing views as attacks on their identity (31).

This is helpful, even as it feels a bit disorienting. Keller continues by explaining Bail’s (correct) belief that social media is more about building identities than sharing ideas. Keller writes:

So social media are not primarily about public discussion of ideas. The ideas are ways to define oneself and signal belonging to a group, as well as to assign identities to others by associating them with groups you oppose. This is the reason social media have perfected the art of bad-faith readings—interpreting a person’s words in the most uncharitable sense possible. There is no effort to understand the argument in its strongest form and respond to it. Rather, the goal is to associate the thinker with shameful “out-groups.”

This “out-group” shaming that Bail and Keller get into a bit is worth more thought. Let’s look at one recent study on how in-group and out-group conflict, not echo chambers, is at the heart of polarization.

Out-group Animosity and Polarization

A good bit of research has been conducted even in the last couple of years that shows “echo chambers” are not the primary reason social media contributes to polarization. At least two studies published in the Proceedings of the National Academy of the Sciences since 2019 explore the relationship between echo chambers and polarization. The gist of the findings is that echo chambers have a much subtler effect on polarization than once thought and could even work against polarization. However, one study published in that same journal just a couple of months ago highlights the real effect of echo chambers online: conflict with “out-groups.”

In a study called “Out-group animosity drives engagement on social media” authors Steve Rathje, Jay K. Van Bavel, and Sander van der Linden investigate the performance of divisive content on social media. What they found is that the most engaging content on social media is content posted by one political group about an opposing political group. This content is, obviously, overwhelmingly negative, but it gets twice as many shares and retweets as content about the group doing the posting. This means that, say, posts by Democrats criticizing Republicans get twice as much engagement posts by Democrats promoting liberal ideas, and vice versa.

Perhaps the most interesting conclusion from the study, though, comes when the authors engage with the new data about echo chambers and their limited effects on polarization. The authors write (bolding mine for emphasis):

Even if people are exposed to more cross-partisan content than expected…our findings suggest that opposing views on social media may be excessively negative about one’s own side. This may help explain why exposure to opposing views on Twitter can actually increase political polarization…Thus, the severity of online echo chambers appears to be a less important issue than the kind of content that tends to surface at the top of one’s feed, since exposure to divisive in-party or out-party voices is unlikely to be productive.

Being exposed to other viewpoints drives polarization because the “other viewpoints” that algorithms are most likely to float to the tops of our feeds are demeaning toward our viewpoints. This means that if a pro-life Twitter user wanted to see pro-choice Twitter content, they are more likely to see a tweet that says, “Pro-life people hate women,” than they are, “Pro-choice rights are women’s rights!”

Content that is antagonistic toward opposing groups gets more engagement and then, because algorithms promote content that gets engagement, that content snowballs in popularity and has a higher chance to achieve some level of virality than content that is kind or “moderate” in nature.

The polarization problem on social media lies with the conflict-laden content we consume more than it does with a lack of exposure to diverse ideas. This is what Bail says in Breaking the Social Media Prism. Our focus has long been on “echo chambers” or “filter bubbles,” which are no doubt a problem for a host of reasons, but not because they are the root of polarization.

What drives polarization on social media is not echo chambers as much as it is our unwavering will to fight. Can we agree to put down our pixelated swords to pursue peace so far as it depends on us?