As sentiment about big tech companies has worsened, emerging conventional wisdom has held that social networks are primary causes — and accelerants — of polarization in the United States. The rise of social networks has been roughly correlated with the rise of authoritarians here and elsewhere. Surely social networks, with their algorithmic feeds pushing the most emotional posts to the top of our attention, are warping our politics?
New research suggests that this may not be the case. In a working paper published this year
, Levi Boxell, Matthew Gentzkow, and Jesse Shapiro found that polarization had increased faster in the United States than anywhere else — but that in several large, modernized nations with high internet usage, polarization was actually decreasing.
“One theory this lets us reject is that polarization is a byproduct of internet penetration or digital media usage,” wrote Ezra Klein, my Vox Media colleague, in a piece last month
. “Internet usage has risen fastest in countries with falling polarization, and much of the run-up in US polarization predates digital media and is concentrated among older populations with more analogue news habits.”
Ezra expounds on these dynamics in his fascinating new book Why We’re Polarized
, discussion of which has dominated both my Twitter and podcast feeds since it came out last month. (I particularly enjoyed his chats with Jill Lepore
, Jamelle Bouie
, and Ta-Nahesi Coates
.) Recently I invited Ezra to answer a few questions I’ve had about the Boxell study and other issues raised by his book, and he was gracious enough to agree.
Recent research doesn’t quite let social networks off the hook. But Ezra’s analysis has challenged some of my beliefs about how Facebook, YouTube, Twitter, and their collective effect on our politics. I hope you enjoy the conversation — and I encourage you to pick up his book here
Casey Newton: One of the more counterintuitive arguments that you make in your book, at least for me, is that social networks aren’t polarizing in the way that we often think thye are. How did you reach that conclusion, and why do you find the research persuasive?
Ezra Klein: Hmmm. I wouldn’t go so far as to say they’re not polarizing. What I’d say is they’re not core to the broad story of polarization, much of which predates social media. There’s also not much evidence for the echo chamber effect, at least not in the way people tend to think of it. I cite, for instance, a well-designed experiment in which Democrats and Republicans on Twitter were paid to follow people from the other side. The exposure made the Republicans more conservative and the Democrats, if anything, more liberal, though the effect wasn’t statistically significant. I also cite some experiments on cable news, which has similar dynamics, where people were forced to watch, and the big finding was that the only people who had their minds changed were those who didn’t want to watch it — when they altered the study design so people could watch something non-political, those folks did, and the persuasive effect melted away.
That’s all to say that the people who read politics Twitter or watch cable news tend to do so because they already know what they believe, and they’re following politics to track whether their candidates, party, or ideas are winning or losing. They’re not easily persuadable.
I found this all surprising because of a conversation I had last year with someone who worked on these issues for one of the big social networks. They told me that the people most likely to post about politics are the most partisan. As a result, if you look at Facebook or Twitter a lot, you tend to only see the most partisan opinions. Over time, this person said, that can’t help but have a polarizing effect.
Is it possible that the research on this subject to date simply hasn’t been done over a long enough time frame to have captured an effect?
I think it does have a polarizing effect, but it’s primarily polarizing because it further polarizes elites, who then act in more polarized ways, which create more polarized choices and situations that the mass public has to respond to. An example of this is impeachment. Donald Trump and Rudy Giuliani got very invested in a Joe Biden conspiracy theory promoted on Breitbart and Fox News. That led them to invest considerable administration resources in trying to prove the conspiracy true, or tar Biden with it. That led to the whistleblower report, and then to the hyper-polarizing impeachment process. So there was a dynamic here in which the ultimate political elite — the president — responded to the polarized political media he consumed by doing something that then everyone else had to respond to. You don’t need a big audience to change American politics, you just need the right audience.
That, I suspect, is more how social media is polarizing: political elites are on Twitter every day, and for all the warnings that Twitter isn’t real life, it feels like real life to them. They’re stuck in a hyper-polarized informational system, and it influences the decisions they make, the candidates they support, the messages they emphasize, the stories they focus on. When people say Twitter isn’t real life, they mean it’s not representative of mass opinion. They’re right! But neither is American politics. Political elites have outsized effect on the structure of politics, and if they become more polarized, and act in more polarized ways, that will ultimately polarize the public simply by presenting them with very polarizing choices to respond to.
One simple way to put this is that the 2020 election looks likely to be between Bernie Sanders and Donald Trump. Both of those candidates have been buoyed by intense social media fandom, in a way that helped them triumph over less controversial, more coalitional, competitors. They present a far starker — and thus more polarizing — choice than, say, Al Gore and George W Bush, who ran their campaigns in ways that muddied the differences between them in 2000. I think social media is part of why politics is selecting for sharper-edged candidates, and that’s leading to a different political reality that even those who aren’t on social media need to face.
Before I started reading your work on this, my instinct was that platforms should take steps to become less polarizing. But you make the point that the mid-century period where American politics were least polarized may have been a historical anomaly — one that emerged from a racist compromise between Democrats and Dixiecrats. In this view, it’s not at all clear that polarization itself is a problem — in a democracy, we’re meant to fight about things!
Let me break that into two pieces. First, the alternative to political polarization is often political suppression — disagreements get suppressed rather than resolved. That’s certainly how the political system treated civil rights for much of the 20th century, bottling up bills in the Rules committee, filibustering them in the US Senate, or neutering them through agreement with the Southern Dixiecrats. But that doesn’t mean polarization is always and everywhere good, nor that the social media networks shouldn’t think about how to reform themselves. I have a longer discussion in the book about the kind of speech and voices that are selected for by algorithms that are sorting off the intensity of emotional response, but I don’t think it’s a great foundation atop which to structure political communication.
You explain how polarization makes governance much harder in the United States, and all the very serious problems that come along with that. But it’s now much less clear to me what platforms ought to do about it, if anything. If platforms found a way to promote agreement and consensus-building, should they? (Twitter has been studying this for almost two years now, with almost nothing to show for it so far.)
In general, I don’t think the platforms are going to fix the problems of American politics. But I do think they could fix the problems of the platforms. People have a million ideas here, but I’ll just state the most obvious: I think the move towards algorithmic feeds that select for content that triggers an intense emotional response is just a bad way to structure communication. I think supercharging our social instincts often brings out the worst, not the best, in us — few look back fondly on the social dynamics of high school cafeterias, and for good reason.
One thing I’ve been doing recently is reading past critics of television, like Neil Postman and Jerry Mander. And something that’s striking about their arguments is they were largely (though not entirely!) right. In many cases, the rise of televisual culture created problems much worse than anything they could’ve imagined, and an approach to politics-as-entertainment that would’ve read like parody if they had included it as a thought experiment in their books. I think there’s a dominant view that every new communications medium comes with critics but we’re still here and the medium prospered so the critics must’ve been wrong, right? No. Sometimes the critics were right, and what they feared came to pass, and we just learned to live with it.
I don’t know what I’d do with these platforms if I ran them. Being inside a system, as I argue in the book, warps your judgment. But to use Twitter’s emphasis on healthy conversations as an example, what if the way to have a healthy conversation is simply inimical to the nature of Twitter — that is to say, what if you’d never, starting from first principles, built healthy conversations around 280-character bursts that are judged by the social reaction they create among an audience that’s consuming them in an environment that is, to say the least, not conducive to reflection?
Finally, you wrote a good piece last week about how a Michael Bloomberg victory in the Democratic primary would set a bad precedent, in part because it would represent the triumph of raw spending power over any other candidate attribute. (As of last week Bloomberg had spent $417 million, much of it on Facebook, Google, and Twitter ads.)
So far, it seems that Democratic voters are rejecting Bloomberg’s candidacy in large numbers. Assuming you’re right, and that Sanders wins the nomination, will that suggest that these days it’s better for a candidate to be polarizing than to be rich?
I don’t think there’s any doubt of that, actually. We saw it in 2016, too: Money matters, but if it was the only thing that mattered, Jeb Bush would be president now.