What this covers

Sean Carroll and Ezra Klein examine the mechanisms and history of American polarization, drawing on Klein's book Why We're Polarized. The conversation traces how the two major parties have sorted themselves by identity—race, religion, geography, and psychology—over the past fifty to sixty years, transforming what "Democrat" and "Republican" mean. Klein argues this stacked identity polarization is not primarily a failure of individuals but the product of structural incentives and feedback loops baked into the media landscape, primary system, and the nature of multiparty competition. Because the American constitutional system requires consensus to govern, he contends the realistic goal is not de-polarization but building institutions that can function under sustained partisan division.

The conversation covers distinct forms of polarization—policy sorting versus affective polarization, the latter driven by fear and hatred—and argues that today's partisan attachments are often motivated more by fear of the opposing party than loyalty to one's own. Klein traces how mid-twentieth-century depolarization was actually the historical aberration, enabled by the Dixiecrats blockading ideological alignment, and how the shift to nationalized, choice-based media created conditions where only strong partisans consume political news, triggering further polarization. The discussion also addresses identity-protective cognition—the tendency for informed people to use their cognitive resources to defend group beliefs rather than discover truth—and how parties function as cognitive mediators, allowing voters to outsource political judgment to trusted affiliations. Klein emphasizes that systems shape outcomes more than individual choice, and that trusted-messenger persuasion works better than exposure to hostile opposing voices.

Sharpest takeaway

Klein argues that American political polarization is the product of identities (race, religion, geography, culture, psychology) stacking and aligning with party, driven by structural incentives and feedback loops rather than individual actors, and that because the US system requires consensus to govern, the realistic goal is not de-polarizing but building a system that functions under polarization.

  • The two parties have sorted by identity over 50-60 years, so 'Democrat' and 'Republican' now encode far more information than before
  • Individuals have narrow ranges of action; systems and incentives drive predictable outcomes (the Trump presidency tested and confirmed this)
  • Polarization is the natural state of multiparty systems; mid-20th-century depolarization (caused by the Dixiecrat blockage) was the aberration

The argument · threads8 threads · 38 claims
0.85

Nationalized media and choice-based consumption create feedback loops that intensify polarization.

7 pointscentrality 5/5
  • In media, the only words that truly matter are the headline; because writers exhaust themselves on the underlying piece and slap on a headline at the end, a bad headline negates a good piece—so Vox requires ten headline options per piece.

    “I'm a big believer that the only words that matter in anything are the headline. So at Vox I have a rule that if you write a piece... you have to come up with 10 headline options.”

  • The shift from monopolistic, inoffensive mass media (when watching one network meant catching the news) to nationalized, choice-based, attention-competing media meant that only people with strong political identities still consumed political news, forcing outlets to run more polarizing stories, which in turn further polarizes the audience in a feedback loop.

    “as we entered into the era of cable news, and then internet news, what happened is we sorted by preference... the people who were very invested and interested in politics became much much much more knowledgeable, whereas the people who didn't like political news just opted out altogether.”

  • Because Vox launched in the age of social media, its headlines became more overheated and polarizing than they otherwise would have been, and the ongoing editorial struggle is to keep the publication from being pulled further toward the poles by platforms like Twitter.

    “Vox started in the age of social media, and so like a lot of groups that are in the social media I think that made our headlines more overheated and more polarizing.”

  • Klein theorizes (while noting political scientists lack a settled answer) that the current unprecedentedly close 50-50 competitiveness is partly produced by a nationalized media that, being oppositional to whoever holds power and drawn to conflict and negativity, amplifies the desperation of the out-party, producing a thermostatic balancing effect.

    “I have a theory that the media plays, to some degree... A nationalized political media, plays a thermostatic function, where because the media is reasonably oppositional to whoever is in power, it tends to cover scandals and mistakes.”

  • Reading the other side does not reliably moderate views: a study by Chris Bail paying Twitter users to follow out-party influencers made Republicans more conservative and did not make liberals less liberal, because being exposed to the other side talking to its own people in an offensive way triggers entrenchment rather than persuasion.

    “they paid them to let them insert in their Twitter feeds members of the other political party... And what happened was that it made Republicans more conservative to see a bunch of liberals in their Twitter feed.”

  • As the share of persuadable voters fell from roughly 20% to under 7%, campaign strategy shifted from courting cross-cutting moderates (Bill Clinton, compassionate-conservative George W. Bush) toward base mobilization (Trump, and arguably Obama), which in turn further polarizes the electorate in a feedback loop.

    “as the number of persuadable voters has gone down, what you have are candidates who are running base mobilization strategies... as the political parties begin running more base mobilization strategies, what they do is they further polarize the electorate.”

  • Accurately reporting on the Trump White House—stating that the President lied or made bigoted statements—itself reads as partisan slander, so the difficulty of describing reality clearly has itself become a polarizing force.

    “For me to state clearly what is going on in the White House... is to say things that sound like almost slander... For me to say clearly what is going on has just become itself more polarizing.”

0.85

Systemic features like weak primaries, veto points, and close competition force extremism and paralysis.

7 pointscentrality 5/5
  • Polarization combined with the American system's many veto points increases paralysis, and because parties can rarely enact their agendas, citizens are left frustrated by unsolved problems and unable to judge parties on actual outcomes—a dynamic that would be healthier if majorities could simply govern and be evaluated.

    “polarization with that many veto points... what polarization tends to do is increase paralysis... absent the ability of parties to actually govern, you're just caught in this endless struggle for power that you never quite achieve”

  • Unlike most polarized democracies where a majority can govern, the American political system was designed to resist parties and requires high levels of consensus to act, so polarization uniquely produces paralysis and makes ambitious governance nearly impossible over the long term.

    “polarization with that many veto points... what polarization tends to do is increase paralysis... absent the ability of parties to actually govern, you're just caught in this endless struggle for power that you never quite achieve”

  • We live in an era of weak parties but strong partisanship: party organizations lost control of nominations as primaries became binding and small-D democratic, but because few people vote in primaries, candidates must fear a small group of activists and loyalists—pulling them toward extremism—while the party itself cannot gatekeep figures like Trump.

    “we live in a period of weak parties and strong partisanship... Donald Trump could never have been the Republican Party nominee cause he would have never won delegates at a convention. He could only be the Republican Party nominee after reforms made to the primary process”

  • This is the most competitive era in American history; unlike past 'sun and moon' eras of long one-party dominance, both parties are always close to winning or losing power, and that constant knife's-edge competition (per Frances Lee) creates a desperation that fuels no-holds-barred political tactics.

    “this is the most competitive era in American politics ever... everybody's always so close to either winning or losing power that it creates this desperation, this no-holds-barred nature to the political tactics”

  • America is not a majority-rule democracy but a system that amplifies the power of rural, mostly white, and affluent voters; a recent analysis found that because of the Electoral College's tilt, Republicans should be expected to win 65% of presidential contests in which they narrowly lose the popular vote, so the GOP wins despite losing more votes and is not forced to moderate.

    “because of the electoral college's tilt towards Republican-leaning areas, that Republicans should be expected to win 65% of presidential contests where they closely lose a popular vote.”

  • Individuals in politics have much narrower ranges of action than political journalism's narratives suggest; the system drives outcomes and shapes incentives so strongly that one can predict where actors will end up, even with large variation among the individuals at play.

    “the system is driving outcomes and shaping incentives such that, while obviously there is some range of free will operating within American politics... it's pretty narrow. And people are, much more to the point, the incentives and systems around them, and polarization is in some ways the master story and master incentive system of them all, are shaping the way they understand the information”

  • Asymmetric polarization—the right straying further from American political norms than the left—is well documented in political science, but when Klein examined the literature he found narrativizations (about Mitch McConnell, ousted speakers) rather than a good causal account of why such figures emerged and were rewarded on the right and not the left.

    “the right has become a more... It has responded to polarization by straying further from the norms of American politics than the left has... while there had been narrativizations of it there actually was not a good causal story for it.”

0.81

Polarization reflects deep ideological sorting across multiple dimensions that changed party meaning.

6 pointscentrality 5/5
  • Polarization is not the same as disagreement; the two parties have sorted themselves by ideology, race, religiosity, geography, culture, and psychology over the past 50-60 years, so that the labels 'Republican' and 'Democrat' now encode far more information than they used to.

    “what has happened is the two parties have sorted themselves by ideology, race, religiosity, geography, culture, and psychology over the past 50 or 60 years such that what Republican and Democrat means, how much information those labels encode is much, much larger than it was then.”

  • Today's self-styled independent or low-information voters are as certain about the differences between the two parties as strong partisans and high-information voters were in the 1970s—not because earlier people were lying, but because the parties' differences are now genuinely far larger (easier to tell a donkey from an elephant than from a mule).

    “today people who are self-styled independent voters or low-information voters, they are as certain when polled about the differences between the two parties or the two candidates, as people who are strong partisans or high-information voters were in the '70s.”

  • Republicans in Congress voted for the Civil Rights Act at a higher proportion than Democrats, but because Barry Goldwater ran against it while Lyndon Johnson championed it, what could have been a bipartisan accomplishment became a partisan cleavage, making the Republican Party a home for white identity backlash and the Democrats the party of non-white Americans.

    “Republicans voted for the Civil Rights Act both in the House and Senate at a higher proportion than Democrats did. But then, in the election, Barry Goldwater runs against the Civil Rights Act and Lyndon Johnson is for it. And so, as opposed to this being a bipartisan accomplishment, it quickly becomes a cleavage between the parties.”

  • The Dixiecrats—the conservative, racially conservative Southern wing of the Democratic Party that ran the South as an authoritarian racial hierarchy—formed a blockade preventing the two parties from coalescing around ideology, which is why mid-20th-century America had liberal Republicans and conservative Democrats and was unusually depolarized.

    “The Dixiecrats, which were the Southern wing of the Democratic Party, they were quite conservative but importantly, very racially conservative. They functionally ran the South as a authoritarian system built on protecting a racial hierarchy system... what they created was almost a blockade around the two parties coalescing around ideologies.”

  • Whereas in the early 20th century the density of a place did not predict its partisan politics, today no place in America denser than about 900 people per square mile is Republican, and all major cities are Democratic—an illustration of how geography has sorted by party.

    “Back, if you go into the early 20th century, the density of a place does not predict its partisan politics. But now there is no place in America more dense than 900 people per square mile, if I'm remembering the stat right, that is Republican.”

  • The correlation between how a district votes for President and for House has risen from around 0.5 in earlier periods to about 0.98 in 2018, and state-to-state presidential swings have fallen from roughly eight or nine points in the 1970s to under two points, showing voters have become far more fixed and nationalized in their preferences.

    “the correlation in your vote between House and presidential in terms of partisan lean is something like, during these periods, 0.5. In 2018, I believe, it was 0.98.”

0.76

Affective polarization driven by fear and identity protection is the core problem, not policy disagreement.

5 pointscentrality 5/5
  • When identities that once cut across parties (union members, religious groups, races, regions) stack on top of each other and align with party, the nature of disagreement becomes deeper and more threatening, making affective polarization—fear and hatred of the other side and loyalty to one's own—a rational response.

    “when you begin to stack identities the nature of disagreement becomes much deeper, it becomes much more threatening. And more affective polarization, hatred and fear of the other side, loyalty or identification with your own side, is a quite rational response.”

  • More informed partisans are sometimes likelier than less informed ones to hold factually wrong beliefs (e.g., informed Republicans answering wrong about the Clinton-era deficit), because consuming partisan media gives them the tools and the selective information to arrange facts in ways that confirm what their side believes.

    “when they ask people, 'Did the budget deficit get bigger or smaller under Bill Clinton?' There's an actual answer to that question. But the more informed Republicans are likelier to answer it wrong than less informed Republicans.”

  • Identity-protective cognition means that on politically charged problems, greater skill or information does not improve accuracy but provides cognitive resources to convince oneself that what one's group believes is correct—so the gap in performance is largest among the most numerate or high-information people.

    “when you do that, particularly for people who are better at math, math stops being helpful in giving you the right answer. Whether or not you get the right answer is whether or not you believe in gun control... The disparity is biggest among the people who are best at math.”

  • Much partisan attachment is driven less by loyalty to one's own party than by fear of the other party—many Republicans dislike Trump but hate the Democrats, and many Democrats dislike their party but fear Trump's re-election—so out-group fear is a primary motivator of political behavior.

    “sometimes, they don't really trust the party they vote for, but they really, really fear the other party. There are a lot of people who are Republicans, not loving Donald Trump, but they hate the Democrats.”

  • There are distinct kinds of polarization—policy polarization (parties clustering at opposite policy poles) and affective polarization (how negatively people feel about the other party)—and affective polarization, driven by hatred and fear of the other side, has risen much more than actual policy disagreement.

    “we're seeing a lot more affective polarization than policy polarization. The two interact in important ways, but one thing... What has really, really risen up is how negatively people feel about the other party, more than how much they necessarily disagree with the other party.”

0.63

Individual disengagement from national politics and reinvestment in local activism reduces polarization's harm.

5 pointscentrality 4/5
  • The most useful individual response to polarization is to disengage somewhat from nationalized politics and reinvest in state and local political identities and concrete local activism, because local identity historically restrained polarization and local engagement is more efficacious and nourishing than tweeting into the national ether.

    “the thing people should do that is the most useful... is to engage less with national politics and more with state and local politics.”

  • When blogging arose it was hated by journalists as part of a critique of how media did its work, so the idea that it could lead to a journalism career seemed absurd—yet many early bloggers now run media organizations.

    “it initially was hated by journalists. It was... It rose up as part of a critique of how the media did its work. So the idea... that that would somehow ever lead to journalism, I would have laughed at you.”

  • Most self-styled independents are actually partisans who prefer the 'independent' brand; 'independent' is such a positive label that people even rate objectively less attractive faces as more attractive when told the person is an independent.

    “most independents are just partisans in disguise... If you give people pictures of folks and you ask them to rate how attractive they are, but you also say if they're Republicans, Democrats, or Independents, people will rate objectively less attractive people more attractive if they're told they're Independents.”

  • The conventional usage of 'identity politics' arbitrarily applies the term only to marginalized groups making claims (e.g., African-Americans on police brutality), while treating equivalent identity-based claims by dominant groups (rural gun owners, CEOs wanting tax cuts) as 'just politics'; in reality, identity is a universally powerful force that operates across all political actors.

    “rural gun owners who want the Second Amendment expansively interpreted, that's just politics. Or CEOs who want their taxes cut, that's just politics. But identity is an incredibly powerful way that people operate in the world.”

  • What is striking about Trump is not how aberrant the 2016 election was but how normal it was; exit polls across recent elections show Trump's coalition was essentially the Mitt Romney coalition with minor tweaks, so a future historian could not pick the 2016 result out of a lineup.

    “You reconstruct most of Mitt Romney's coalition, and then the Republican Party falls in line behind the party leader, and he ends up more or less governing as a traditional Republican with a very unusual Twitter account. So, if that is not a test of how much polarization in the broader political system incentives contain even the most disruptive figure, I don't know what is.”

0.58

Redesign political institutions to function amid polarization rather than attempt to reduce it.

4 pointscentrality 4/5
  • Because the feedback loops driving polarization are very hard to reverse, the realistic and better goal is to redesign the political system to function well amid polarization—removing self-inflicted risks like the debt ceiling, abolishing the filibuster, adopting proportional representation for multi-party systems, and adding states like DC and Puerto Rico—rather than trying to reduce polarization itself.

    “What I do think is possible, as you could imagine if we wanted to, a political system that worked better amidst polarization... you can also imagine getting rid of the filibuster, you can imagine forms of proportional representation that would lead to multi-party systems”

  • Minds change not through exposure to the offensive 'other side' but through messengers you regard as on your side who happen to hold some different views and are tuning their arguments to you—reading Ross Douthat (a center-right writer in the NYT addressing liberals) works where reading Breitbart, which exists to offend, does not.

    “What you need is people who you actually think of as on your side, they just happen to have some different political opinions than you do, and you're talking to them.”

  • Localized political deal-making—such as a Republican voting for a Democratic bill in exchange for a bridge in his district—long kept polarization lower; the elimination of earmarks under John Boehner was both a symptom and an accelerant of the nationalization that fuels polarization.

    “one of the ways that the parties kept polarization reasonably low... it may be true that you're a Republican, but you're a Republican from this one district in Oklahoma. And if you voted for that bill, they would help you rebuild the bridge in that one district... And then under John Boehner they got rid of earmarks.”

  • Because no one can hold informed opinions on the vast number of issues politics asks about, parties function as crucial mediators: people attach to a party through identity and then take cues from it on individual issues—reasoning forward from whom they already trust rather than backward from independently formed policy preferences.

    “the idea that what is happening is that people are... most people do not sit down, look at the 50 policy items they think are gonna come before Congress this year and reason backwards from them. They reason forward from who they already trust.”

0.40

Visible disagreement over historically suppressed issues is a healthy democratic consequence, not pathology.

3 pointscentrality 3/5
  • The alternative to polarization is often suppression: in the 1960s, polarization rose because long-bottled-up disagreements—such as over civil rights, which the Dixiecrats had kept off the legislative floor—were finally forced into the open, so difficult debates becoming visible is in part a healthy consequence.

    “the alternative to polarization is often suppression. What was very polarizing in the '60s was that a lot of disagreements that had been bottled up were forced into the open, like say over civil rights.”

  • The 1960s, despite featuring political assassinations, urban riots, and mass civil rights, anti-war and feminist movements, were a nadir of polarization—the political system itself was among the least polarized in American history—showing that intensity of disagreement and degree of partisan sorting are distinct.

    “you can go even back to the '60s, which importantly are a nadir of polarization in America. It's one of the periods in which our political system is least polarized. But look around, you have political assassinations, you have student protesters being killed in the streets”

  • High-information, high-salience political systems with multiple parties tend naturally toward polarization because parties, like products in a market, must differentiate; mid-20th-century American depolarization is the historical and global aberration, not the norm.

    “When you have high salience political... High information, high salience political systems with multiple political parties, polarization is usually the result. What is distinct about America... is that most political systems function just fine amid polarization.”

0.18

Trump presidency demonstrated that systemic incentives constrain even aberrant individuals.

1 pointcentrality 2/5
  • The Trump presidency served as a natural experiment: an aberrant figure in ideology, style, and relationship to party elites nonetheless reconstructed the Mitt Romney coalition, the party fell in line behind him, and he largely governed as a traditional Republican with an unusual Twitter account—demonstrating that systemic incentives contain even the most disruptive individual.

    “You reconstruct most of Mitt Romney's coalition, and then the Republican Party falls in line behind the party leader, and he ends up more or less governing as a traditional Republican with a very unusual Twitter account. So, if that is not a test of how much polarization in the broader political system incentives contain even the most disruptive figure, I don't know what is.”