Why the media is so polarized — and how it polarizes us

The following is an excerpt from Ezra Klein’s new book, Why We’re Polarized, published by Simon & Schuster and available January 28.

We talk a lot about the left/right divide in political media. But we don’t talk enough about the more fundamental divide that precedes and, in some ways, causes it: the interested/uninterested divide.

In All the News That’s Fit to Sell, economist James Hamilton writes, “News emerges not from individuals seeking to improve the functioning of democracy but from readers seeking diversion, reporters forging careers, and owners searching for profits.” That’s a bit more cynical than I’d be — a lot of us really do want to improve the functioning of democracy — but as a description of the overall economic system that surrounds our work, it’s useful. You can’t understand the news without understanding the financial and audience forces that shape it. The first thing to appreciate is that those forces have changed, and changed dramatically, in recent decades.

Consider the options available to eager political news consumers in 1995. They might have had a hometown paper or two, a handful of radio stations, the three nightly newscasts, the newly launched CNN, and, if they were really hardcore, a couple of magazine subscriptions.

Fast-forward a decade. Those same consumers could fire up Internet Explorer and read almost any newspaper in the country — and most of the major newspapers of the world — online. For political opinion, they had a dizzying array of magazines, any op-ed page they chose, and, all of a sudden, a countless number of blogs. On television, CNN had been joined by Fox News and MSNBC. On radio, satellite began crowding the airwaves with more political commentary. In pockets, the launch of the iPod kicked off the age of podcasting. And the quantity of available political information has only multiplied since then. Never in human history has it been remotely possible to be this politically informed.

In most models of democratic politics, information is the constraint. Voters don’t have the time or energy to read thick tomes of political theory and keep themselves updated on every act of Congress, so they’re dependent on the political professionals — elected officials, campaign operatives, party staffers, lobbyists, pundits — who do. What follows from this model is tantalizing: If information ceases to be scarce, if it becomes freely and easily available to all, the fundamental problem afflicting democratic systems would be solved.

Over the past decade, the dreams of democratic theorists everywhere actually came true. The internet made information abundant. The rise of online news gave Americans access to more information — vastly more information, orders of magnitude more information — than they had ever had before. And yet surveys showed we weren’t, on average, any more politically informed. Nor were we any more involved: Voter participation didn’t show a boost from the democratization of political information. Why?

How choice changes media

In the early aughts, Princeton political scientist Markus Prior set out to unravel this apparent paradox. The way he resolved the problem is, in retrospect, obvious.

Yes, there were more cable news channels, but they were dwarfed in number by the channels that had no interest whatsoever in news — channels that served up round-the-clock cooking, home repair, travel, comedy, cartoons, tech, classic films. The key factor now, Prior argued, was not access to political information but interest.

Yes, you could read the political coverage of any newspaper or magazine in the country online, but you could also read so much more nonpolitical coverage. The explosion in political media was more than matched by the explosion in media covering music, television, diets, health, video games, rock climbing, spirituality, celebrity breakups, sports, gardening, cat pictures, genealogical records — really, everything.

The key factor now, Prior argued, was not access to political information but interest in political information. He made his point by comparing it to television. Like the internet, television multiplied the amount of information available to people, and it spread like wildfire. But unlike the internet, television, at least in its early years, offered little choice. You might own a television because you refused to miss I Love Lucy, but if you had the TV on in the evening, you ended up sitting through the news anyway. Similarly, you might subscribe to the newspaper for the sports page, but that meant seeing the political stories on A1. Politics was bundled alongside everything else, and even the uninterested were pushed to consume political news.

The digital revolution offered access to unimaginably vast vistas of information, but, just as important, it offered access to unimaginably more choice. And that explosion of choice widened that interested/uninterested divide. Greater choices let the devotees learn more and the uninterested know less.

To test this, Prior surveyed more than 2,300 people about their content preferences and their political knowledge. And because he was conducting this survey in 2002 and 2003, the early years of the internet and still reasonably early for cable, he was able to survey people who had internet access, those who had cable access, those who had both, and those who had neither.

Content preferences — which is to say, how much people wanted to consume political information versus how much they wanted to consume other forms of entertainment — had little effect on the knowledge of those without cable and internet access. Even if you wanted more political information, you didn’t have easy access to it, so the interest didn’t translate cleanly into information.

What the digital information revolution offered wasn’t just more information but more choice of information

But among those with cable and internet access, the difference in political knowledge between those with the highest and lowest interest in cable news was 27 percent. That dwarfed the difference in political knowledge between people with the highest and lowest levels of schooling. “In a high-choice environment, people’s content preferences become better predictors of political learning than even their level of education,” Prior wrote.

Prior was conducting this research in the early 2000s, before Facebook and Twitter, before mobile internet and YouTube algorithms, before MSNBC’s leftward turn, before BuzzFeed and HuffPost, before Breitbart and the alt-right, before Vox. The internet has become much better at learning what we want and giving us more of it since then. The competition for audience, and the threat to journalistic business models, has become much more intense since then. And all of this has changed both how political news is produced and how it’s consumed.

Political media is for the politically invested

In an age of choice, political journalism is a business that serves people interested in political news and that tries to create more people interested in political news. And to be interested in politics is, for most people, to choose a side. How could it be otherwise? The differences between the parties and their coalitions are profound. They are ideological, geographic, demographic, temperamental. Whether your side wins or loses is a literal matter of life and death — perhaps not for you, but, given the stakes for health insurance and foreign policy, certainly for someone.

In today’s media sphere, where the explosion of choices has made it possible to get the political media you really want, it’s expressed itself in polarized media that attaches to political identity, conflict, and celebrity. That is to say, it expresses itself in journalism and commentary that is more directly about the question of why your side should win and the other side should lose.

I’ve produced a lot of this kind of journalism. I cover politics because I think policy is important, which is to say, because I think who wins and who loses policy fights is important. And, obviously, my views on those questions are rational, judicious, disinterested, and objectively correct. The problem is lots of other people are doing that kind of work, too, and some of them come to different conclusions than I do.

But rather than argue over who’s right, I want to step back and look at how a political media system increasingly organized around that axis deepens political identity, hardens polarization, and raises the political stakes.

The simplest measure for assessing political journalism is whether it’s giving those who follow it a more accurate understanding of American politics. As one disturbing window into this question, consider a fascinating study published by Douglas Ahler and Gaurav Sood in 2018.

In it, Ahler and Sood conducted a survey asking people “to estimate the percentage of Democrats who are black, atheist or agnostic, union members, and gay, lesbian, or bisexual, and the percentage of Republicans who are evangelical, 65 or older, Southern, and earn over $250,000 per year.” They were asking, in other words, how much people thought the composition of the parties fit the caricatures of the parties.

Misperceptions were particularly high when people were asked to describe the other party. Democrats believed 44 percent of Republicans earned more than $250,000 a year; it’s actually 2 percent. Republicans believed that 38 percent of Democrats were gay, lesbian, or bisexual; the correct answer is about 6 percent. Democrats believed that more than four in 10 Republicans are seniors; in truth, seniors make up about 20 percent of the GOP. Republicans believed that 46 percent of Democrats are black and 44 percent belong to a union; in reality, about 24 percent of Democrats are black and less than 11 percent belong to a union.

Here’s the kicker: As the charts below show, the more political media people consumed, the more mistaken they were, in general, about the other party. This is a damning result: The more political media you absorb, the more warped your perspective of the other side becomes.

The old line on local reporting was “If it bleeds, it leads.” For political reporting, the principle is “If it outrages, it leads.” And outrage is deeply connected to identity — we are outraged when members of other groups threaten our group and violate our values. As such, polarized media doesn’t emphasize commonalities, it weaponizes differences; it doesn’t focus on the best of the other side, it threatens you with the worst.

As that last paragraph suggests, I’m about to step into some dangerous territory, so let me say this clearly: I’m not asserting moral equivalence, and in the book this article is adapted from, I have much more to say about the ways and reasons the left and the right — including their media spheres — have diverged.

But virtually everyone in political media is competing for audience attention and loyalty amid a cacophony of choices. We all make different decisions about how to compete for that audience, but since we are all trying to attract other human beings, there are certain similarities in our approach.

Why audience-driven media is identitarian media

Historically, not only did the audience have less choice in what media to consume, the media didn’t have much information about the audience. The networks had ratings. The newspapers had subscription renewals. Everyone received letters. But that was it.

I used to regularly guest-host on cable news. The emotional rhythm of that workday crested at 4 pm, when the Nielsen numbers came out and everyone stopped to compare how their show did against the competition. If you beat your competitors, you could rest easy. If you didn’t, you had to worry. And if you lost a few times in a row, you’d start getting calls from upstairs. Maybe your programming should stick closer to the news of the day. Maybe you needed shorter intros, or longer intros, or more guests, or more heat.

Cable news is journalism, but it’s also a business. Chris Hayes, who anchors MSNBC’s 8 pm newscast and is among the most thoughtful, civic-minded journalists in the industry, referenced a Will Ferrell joke from Anchorman 2 on his podcast, saying, “What if instead of telling people the things they need to know, we tell them what they want to know?” That is, he says, “the creation story of cable news.”

“At some level,” he continued, “we’re wedding DJs. And the wedding DJ’s job is to get you on the floor.” The point is not that this leaves no room for serious journalism. As Hayes says, there are good wedding DJs and bad wedding DJs, and the work of being a cable news host is making sure you’re one of the good ones. But this is the business context in which cable news decisions are made.

Then came the rise of real-time digital analytics. Every digital newsroom in the country, including Vox, subscribes to some service or another that tracks traffic in a gamified, constantly updating interface. The most influential is Chartbeat, which shows you every article on your site, indicates the number of people on each article at any given second, and colors the dots representing those people to tell you how they found the article. Green dots mean they found you through a search engine. Purple dots mean they came from a social network, usually Facebook, Twitter, or Reddit. It’s pure pleasure to watch the display for an article you worked hard on fill with dots.

But we don’t just want people to read our work. We want people to spread our work — to be so moved by what we wrote or said that they log on to Facebook and share it with their friends or head over to Reddit and try to tell the world. That’s how you get those dots to multiply. But people don’t share quiet voices. They share loud voices. They share work that moves them, that helps them express to their friends who they are and how they feel. Social platforms are about curating and expressing a public-facing identity. They’re about saying, “I’m a person who cares about this, likes that, and loathes this other thing.” They are about signaling the groups you belong to and, just as important, the groups you don’t belong to.

BuzzFeed’s secret: Identity = virality

The rise of BuzzFeed made this subtext into text. Its co-founder and CEO, Jonah Peretti, originally built BuzzFeed on the side as a skunkworks for experimenting with how viral content spread online. The answer soon became clear: Identity is the slingshot.

“A classic early BuzzFeed post, and later video, was ‘13 Struggles All Left-Handers Know to Be True,’” Peretti tells me. “Another early classic was ‘Signs You Were Raised by Immigrant Parents.’ That one’s a racial identity but also an immigrant identity.”

There are so many more. One of BuzzFeed’s most popular series was “X Things Only a Y Would Understand.” A Google search for those keywords brings up articles like “14 Things Only Anxious People Will Understand,” “19 Things Only People With Fibromyalgia Will Understand,” “53 Things Only ’80s Girls Can Understand,” “30 Things Everyone Who Went to College Will Understand,” “27 Struggles You’ll Only Understand If You Were Born Before 1995,” “38 Things Only Someone Who Was a Scout Would Know,” “19 Comics Only Night Owls Will Understand,” “19 Things You’ll Only Understand if You Had Strict Parents,” “18 Photos That Only People Who Had Braces Will Understand.”

This is identity media in its purest form. Sharing the scouting article says you were a Scout, and you were a serious enough Scout to understand the signifiers and experiences that only Scouts had. To post that article on Facebook is to make a statement about who you are, who your group is, and, just as important, who is excluded.

In political media, identity is affirmed and activated with slightly more oblique headlines. But the underlying dynamic is the same: This public figure that you and everyone in your group loathe said something awful. This poll came out saying you and your group are going to win or, better yet, that your out-group is going to lose. This slashing column explains why you’re right about everything and why your opponents are wrong.

A lot of these pieces are accurate, and some of them are genuinely useful. I have written many of them myself, and edited countless more. But cumulatively, it’s a sharp change from the days in which most political content people saw was self-consciously trying to avoid offending anyone. The stories that thrive when your business model is a local monopoly that needs a news product that’s appealing to every kind of person who might shop at a department store are different from the stories that thrive when your business model is people who strongly agree with your stories sharing them with their friends.

Identities are malleable things. They can be activated or kept dormant, strengthened or weakened, created or left in the void. The flood of identity-oriented content deepens the identities it repeatedly triggers, confirms, or threatens.

Many of us who wrote about politics on the internet before the rise of social media lament the feeling that something has been lost, that a space that once felt fresh and generative now feels toxic and narrow. In her book Trick Mirror, Jia Tolentino offers a description of what changed that feels right to me, which is that social media shifted the “organizing principle” of online discourse:

The early internet had been constructed around lines of affinity and openness. But when the internet moved to an organizing principle of opposition, much of what had formerly been surprising and rewarding and curious became tedious, noxious, and grim.

This shift partly reflects basic social physics. Having a mutual enemy is a quick way to make a friend—we learn this as early as elementary school—and politically, it’s much easier to organize people against something than it is to unite them in an affirmative vision. And, within the economy of attention, conflict always gets more people to look.

When I entered journalism, the term of art for pieces infused with perspective was “opinion journalism.” The point of the work was to convey an opinion. Nowadays, I think a lot of it is closer to “identity journalism” — the effect of the work, given the social channels through which it’s consumed, is to reinforce an identity.

But an identity, once adopted, is harder to change than an opinion. An identity that binds you into a community you care about is costly and painful to abandon, and the mind will go to great lengths to avoid abandoning it. So the more media people see that encourages them to think of themselves as part of a group, and the more they publicly proclaim — through sharing and liking and following and subscribing — that they are part of a group, the deeper that identity roots and the more resistant the underlying views become to change.

Reading the other side doesn’t change our minds

Many people worry that modern media generates polarization by locking us into echo chambers. We’ve cocooned ourselves into hearing information that only tells us how right we are, and that’s making us more extreme.

There is an optimistic theory embedded in this story: If only we crossed the informational aisle, our enmity and polarization would ebb.

Beginning in October 2017, a group of political scientists and sociologists decided to test this theory. In the largest study of its kind conducted, they paid 1,220 regular Twitter users who identified as either Democrats or Republicans to follow a bot retweeting elected officials, media figures, and opinion leaders from the other side. The participants took regular surveys asking about their views on 10 issues ranging from immigration to government waste to corporate profits to LGBTQ acceptance.

The result of the month-long exposure to popular, authoritative voices from the other side of the aisle was that respondents became more, not less, polarized. “We find that Republicans who followed a liberal Twitter bot became substantially more conservative posttreatment,” write the authors. “Democrats exhibited slight increases in liberal attitudes after following a conservative Twitter bot, although these effects are not statistically significant.”

The difference between the Democratic and Republican responses is interesting and merits more study. But the key finding is that neither group responded to exposure to the other side by moderating its own views. In both cases, hearing contrary opinions drove partisans not just to a deeper certainty in the rightness of their cause, but to more polarized policy positions — Republicans became more conservative rather than more liberal, and Democrats, if anything happened at all, became more liberal rather than more conservative.

I spoke to Christopher Bail, one of the study’s authors and the head of Duke University’s Polarization Lab. “For a long time, people have been assuming that exposing people to opposing views creates the opportunity for moderation,” he told me. “If I could humbly claim to figure out one thing, it’s that that’s not a simple process. If Twitter tweaks its algorithms to put one Republican for every nine Democrats in your Twitter feed, that won’t increase moderation.”

There is evidence that structuring positive, collaborative interactions can promote understanding. But very little in either political media or social media is designed for positive interactions with the other side. Most political media isn’t even designed for persuasion. For all the reasons we’ve discussed, the bulk of opinionated political media is written for the side that already agrees with the author, and most partisan elected officials are tweeting to their supporters, who follow them and fundraise for them, rather than to their critics, who don’t.

When we talk about political media, we tend to cut a sharp line between the political elites who create the media and the audience that consumes it. But that’s a mistake. No one consumes more political, and politicized, media than political elites. This is part of the reason political media has an enormous effect on politics, even though only a small fraction of the country regularly consumes it.

Politicians are increasingly addicted to Twitter, with the president being only the most prominent example. Fox News has whipped the Republican Party into a number of government shutdowns, and much of Trump’s most offensive rhetoric comes on a direct conveyor belt from conservative media feeding him conspiracies that he transforms into presidential proclamations.

Indeed, the impeachment effort House Democrats launched against Trump stems from Trump believing a set of anti-Biden conspiracies pushed by Breitbart editor-at-large Peter Schweizer and heavily promoted on Fox News. Most Americans had never heard of Hunter Biden, much less followed vague insinuations about Ukrainian prosecutors. But the president was sufficiently persuaded that he threw the weight of his administration into an investigation, setting off a chain of events that changed American political history and further polarized the country.

Politics is, first and foremost, driven by the people who pay the most attention and wield the most power — and those people opt in to extraordinarily politicized media. They then create the political system they perceive.

Journalists are hardly immune to these forces. We become more polarized, and more polarizing, when we start spending our time in polarizing environments. I have seen it in myself, and I have watched it in others: When we’re going for retweets, or when our main form of audience feedback is coming from highly partisan social media users, it subtly but importantly warps our news judgment. It changes who we cover and what stories we chase. And when we cover politics in a more polarized way, anticipating or absorbing the tastes of a more polarized audience, we create a more polarized political reality.

The media creates, it doesn’t just reflect

The news is supposed to be a mirror held up to the world, but the world is far too vast to fit in our mirror. The fundamental thing the media does all day, every day, is decide what to cover — decide, that is, what is newsworthy.

Here’s the dilemma: to decide what to cover is to become the shaper of the news rather than a mirror held up to the news. It makes journalists actors rather than observers. It annihilates our fundamental conception of ourselves. And yet it’s the most important decision we make. If we decide to give more coverage to Hillary Clinton’s emails than to her policy proposals — which is what we did — then we make her emails more important to the public’s understanding of her character and potential presidency than her policy proposals. In doing so, we shape not just the news but the election, and thus the country.

While I’m critical of the specific decision my industry made in that case, this problem is inescapable. The news media isn’t just an actor in politics. It’s arguably the most powerful actor in politics. It’s the primary intermediary between what politicians do and what the public knows. The way we try to get around this is by conceptually outsourcing the decisions about what we cover to the idea of newsworthiness. If we simply cover what’s newsworthy, then we’re not the ones making those decisions — it’s the neutral, external judgment of news worthiness that bears responsibility. The problem is that no one, anywhere, has a rigorous definition of newsworthiness, much less a definition that they actually follow.

A simple example comes in the treatment of presidential and pre-presidential rhetoric. On some level, anything that the president says, or that a plausible candidate for president says, is newsworthy. Yet only a small minority of what is said by presidential candidates, or even presidents, gets covered as major news.

When President Obama gave a speech on manufacturing policy at an Ohio steel mill and when Sen. Marco Rubio held a town hall discussing higher education costs in New Hampshire, they struggled to get the press to take notice. Trump, meanwhile, routinely gets cable networks to air his rallies live by lying flagrantly, lobbing racist and sexist insults, and generally behaving outrageously. Whether this is strategy or intuition, the result is the same: Trump hacked the media’s true definition of newsworthiness, and it lets him control the agenda. This was true well before he won the presidency — indeed, it might be why he won the presidency.

In their book Identity Crisis, political scientists John Sides, Lynn Vavreck, and Michael Tesler find that “from May 1, 2015, to April 30, 2016, Trump’s median share of cable news mentions was 52 percent.” There were 17 Republican candidates running for president, so Trump was getting more than half of all the media coverage, with the other 16 candidates splitting the remainder.

It gets worse. “Trump received 78 percent of all coverage on CNN between Aug. 24 and Sept. 4, 2015,” and by November 2015, “Trump had received more evening network news coverage—234 minutes—than the entire Democratic field. By contrast, Ted Cruz had received seven minutes.” This was a choice the media made, and not for the best reasons. In February 2016, for instance, the chair of CBS said of Donald Trump’s candidacy, and the ratings it drew, “it may not be good for America, but it’s damn good for CBS. … It’s a terrible thing to say, but bring it on, Donald. Keep going.”

Sides, Vavreck, and Tesler argue that in a chaotic, crowded primary, the media coverage Trump received was crucial to legitimizing his campaign: “Republican voters had received no clear signal about who the front-runner was or should be. The resulting uncertainty meant that this signal needed to come from somewhere else. It was news media coverage that would fill this void.” The coverage of Trump also made it impossible for his challengers to get their messages heard.

As president, his rambling monologues, which are unusually detached from both factual rigor and his administration’s policymaking decisions, are treated as worthier of airtime than the more careful, factual, and policy-predictive speeches of his predecessors.

“Journalism academics have always known that newsworthiness, as the American press defines it, isn’t a system with any coherence to it,” Jay Rosen, a journalism professor at New York University, told me. “It doesn’t make any sense. It’s just a list of factors that occasionally come together to produce news. There’s no real logic to it, other than it’s a list of things that can make something news. The advantage of it is that it leaves maximum leeway for editors to say, ‘This is news,’ and, ‘That’s not news,’ and so it’s news if a journalist decides it’s news.”

In practice, judgments of newsworthiness are often contagious; nothing obscures the fact that a decision is being made quite like everyone else making it, too. In the modern era, a shortcut to newsworthiness is social media virality; if people are already talking about a story or a tweet, that makes it newsworthy almost by definition.

This can lead the country into odd, angry cul-de-sacs. I remember returning from an offline vacation only to find the entire political media at war over a viral video in which students from Covington Catholic High School wearing MAGA hats appeared to harass Nathan Phillips, a Native American elder playing a drum. In the original video, which took place during a protest at the National Mall in Washington, DC, the teens were seemingly mocking, smirking, and making tomahawk chop motions at Phillips. A longer video muddied the waters, offering evidence that the teens were harassed by members of the fringe Black Israelites group beforehand. Soon enough, the media was filled with takes and counter-takes, and President Trump was weighing in. “Nick Sandmann and the students of Covington have become symbols of Fake News and how evil it can be,” he tweeted.

Nick Sandmann and the students of Covington have become symbols of Fake News and how evil it can be. They have captivated the attention of the world, and I know they will use it for the good – maybe even to bring people together. It started off unpleasant, but can end in a dream!

— Donald J. Trump (@realDonaldTrump) January 22, 2019

What was striking, walking into this debate without the (dis)advantage of being present for its initial escalation, was how angry everyone was over something that objectively didn’t matter. How was this newsworthy?

The answer was that it had been dominating social media all weekend, and that had made it newsworthy. And why had it dominated social media? Because it was a perfect collision of political identities: MAGA-hatted teenagers against a peaceful, drumming Native American elder. Liberal news outlets turning the country against conservative, Christian children from a religious school. It was an object lesson in how social media’s preference for identitarian conflict focuses the media on identitarian conflicts, even when those collisions are almost comically obscure.

These are dynamics that Trump exploits daily. He weaponizes outrageousness, offensiveness, and identity cues to capture a share of political coverage unknown in the modern era. He’s shown that in a competitive media environment — particularly one responsive to social platforms — you can dominate the media by lobbing grenades into our deepest social divides.

The media is how most Americans get their information about politics and politicians, and if the media is tilting, or being tilted, toward certain kinds of political stories and figures, then the political system will tilt in that direction, too.

Trump is a product of the tilting, but he is not the first, and he will not be the last. The political media is biased, but not toward the left or right so much as toward loud, outrageous, colorful, inspirational, confrontational. It is biased toward the political stories and figures who activate our identities, because it is biased toward and dependent on the fraction of the country with the most intense political identities.

You can order Why We’re Polarized, or find a full schedule of tour events, here. You can also listen to an excerpt from the audiobook book by subscribing to The Ezra Klein Show, or by streaming it here.