Politics, polarization, and the future of democracy

Yale political theorist Hélène Landemore helped manage a convention of French citizens—a large, randomly-selected group of people brought together to make policy recommendations—tasked with reconsidering the country’s policies on assisted dying. Photo credit: Katrin Baumann/CESE

It’s a milestone year for democracy in the United States. In July, the country will mark its 250th anniversary as a constitutional republic—a government “of the people, by the people, for the people,” in the words of Abraham Lincoln.  

But it’s also a tumultuous moment for democracy in the US and around the world, driving many people to ask: Are people—and politicians—too polarized to have constructive conversations and find solutions? Is democracy still fit for purpose?

FAS faculty are engaging with these timely questions from all angles. They’re conducting research that highlights the thoughts of everyday voters. They’re probing the impact artificial intelligence may have on democracy. They’re even coming up with, and helping implement, innovative approaches to democratic participation.

If there's one thing many scholars at Yale have found in their research, it's that improving democracy isn't about eliminating disagreement. Rather, it's about improving the quality of our conversations with one another, and finding ways to make democratic decision-making faster, more efficient, and more representative of everyday people.  

“I don’t think we should aim for a world where everyone agrees,” says Gregory Huber, Forst Family Professor of Political Science and Acting Director of Yale’s Institution for Social and Policy Studies (ISPS). “We don't need politics if there's only one person, or if we all agree. Politics is fundamentally about managing disagreement, so we should expect there to be disagreement—not to make it go away.”

The Institution for Social and Policy Studies (ISPS) supports much of the domestic research on democracy happening at Yale and brings together interdisciplinary faculty to conduct a wide range of social scientific work. From understanding the stakes of political polarization, to establishing citizens’ assemblies, to interrogating the impact extreme political parties have on society, FAS faculty are leading the way on reimagining models of democracy that are fit for our increasingly complex modern world.  

“I think people are lonely, tired, and exhausted from the toxicity of politics, and it cannot go on like this. We need to do something,” says Hélène Landemore, Damon Wells ‘58 Professor of Political Science. Landemore is the leader of an ISPS research group on “Citizens’ Assemblies” nested under ISPS’s Democratic Innovations program, and the organizer of Governing X, a conference series seeking to address concrete governance problems from a theoretical, practical, and normative perspective.  

“My view is that a first step is to get people in a room to talk to each other again, and perhaps even vent for a while, so they can learn to know each other, trust each other, and cooperate.”

How polarized are we, really?  

The idea that politics is increasingly polarized—that is, divided along party and ideological lines—has become a given.  

“The empirical finding that started to emerge about twenty years ago in the US is that people started to feel comparatively warmer towards their party than they did towards the out party,” Huber explains. (The “out party” is the political party you don't consider yourself a part of.)  

This phenomenon is called affective polarization. “People puzzled over what this meant, and it's an area of ongoing research: should we care that people don't feel warmly towards the out party? When they think about the out party, what are they thinking of?”  

Huber’s research found that these days, people often think of partisan elites, like the President and other household-name politicians, when they think of Democrats or Republicans—not a neighbor down the street.

But Huber and other Yale faculty have questioned just how divided average voters really are, as well as whether—and when—they’re willing to change their minds. He has deeply investigated affective polarization using a variety of social science methods, including field experiments, lab experiments, survey experiments, and straightforward surveys. 

“Our work has, I think, pretty conclusively shown that affective polarization is about issues; it's about substantive disagreement rather than about partisanship per se. So that's good and bad news,” Huber explains. “On the one hand, if it was about just which team you were on, it'd be like Yankees and Mets fans: there's no way to resolve that. But it's about substantive politics.”

He has conducted some of that work alongside ISPS colleagues Joshua Kalla, Associate Professor of Political Science and of Statistics and Data Science, and Alan Gerber, Sterling Professor of Political Science and the Director of ISPS.

It’s not completely clear to Huber that affective polarization is bad in and of itself, but rather a reflection of deeply held beliefs. “It may be that issues are actually what motivates us to be partisans—that we have intense feelings about these things.”

Like Huber, Joshua Kalla often goes directly to voters to understand what they’re thinking and why they make certain political choices. His research often observes active political campaigns, allowing for unmatched insight into how political practitioners and voters actually behave in the real world.  

His work also probes how campaigns, candidates, and other organizations try to change voters’ minds. To do this, he often carries out randomized field experiments, which he likens to clinical drug trials—the gold standard for determining whether a specific intervention actually works in the real world.  

His work has yielded some surprising findings. In recent studies, he determined that conversations between canvassers and voters can decrease polarization and lead people to moderate their beliefs; in another, he found that research participants who consumed cable news from a different TV channel changed their outlook on political events (albeit temporarily, and their views had typically returned to their pre-experiment baseline when surveyed a few months later).  

“I have work showing that when people have deeply held prejudices, and when they talk about those with someone who they might disagree with, not only are they willing to have a civil conversation—which I think in and of itself is a great sign. But they also become more moderate in their views,” Kalla says.  

It appears that our politics tend to work better when people come together and discuss issues face-to-face. And even when we still disagree, Kalla has found, those disagreements don’t preclude voters from changing their minds or voting for surprising candidates.  

“What my co-authors and I argue is, the fact that we disagree with the other side, it's not breaking democracy. We can disagree, but things still function. People still support reasonable candidates. People still support reasonable policies.”  

By and large, Kalla thinks, democracy is still working—even if the temperature of public debate is hotter than many people would like.  

“You might not like the way that every election turns out, but at the end of the day, voters are responding to what's happening in the broader environment, and what's happening in the economy,” he explains. “They're responding to what a politician is standing for and what they're promising. And more often than not, they vote for that person and Congress is then responsive to what the American public wants.”  

Challenges to democracy, both new and old  

Despite positive signs in the realm of person-to-person persuasion, disagreement is indeed a defining characteristic of democracy. But when those disagreements begin to stray beyond what some citizens deem politically reasonable, how do democracies respond?

Isabela Mares, Arnold Wolfers Professor of Political Science, is currently investigating how multi-party democracies in Germany and France respond to the rise of extremist parties—both in the past and in the present.

She is conducting historical research, survey experiments, and interviews with voters, politicians, and legal scholars in both countries. Her goal: to capture the complex trade-offs democracies face when extremist parties compete, challenge long-term democratic norms, and propose radical policies.  

Voters in Germany and France have occasionally elected representatives from more extreme parties to their parliaments, like the neo-Nazi and Communist parties in 1950s Germany, and Communists in 1950s France. Today, Germany’s Alternative für Deutschland (AfD) and France’s National Rally (RN) are forcing centrist parliamentary coalitions to contend with rising support for far-right parties—and to decide whether, and how, to work with their more extreme colleagues.

Historically, mainstream parties have taken a mix of approaches, some of which include demarcation (i.e., refusing to form coalitions in parliament with radical parties) and party bans, which outright prohibit members of extremist parties from participating in politics: they become ineligible for public campaign financing, candidates may not appear on ballots with the party's affiliation, and no new organizations are allowed to form under the same party name.

Taking what are essentially undemocratic steps to “preserve” democracy is not electorally or normatively unproblematic, and Mares points out that these kinds of policies are often rejected even by voters of mainstream parties.  

In Germany, she’s found that a significant number of people voting for the center-right Christian Democratic Union (CDU) party—about 20 percent—oppose a policy of excluding the far-right AfD from government, given that the AfD received about 30 percent of the popular vote in the last parliamentary election. These voters also oppose proposals to ban the AfD party outright, which Germany’s federal parliament, the Bundestag, considered a few months ago.  

Demarcation could also have serious electoral consequences for the CDU and its centrist coalition with the Social Democratic Party (SPD). The current CDU-SDP coalition often adopts policies that are more left leaning than average center right voters prefer. But in the long run, this mismatch between voters’ preferences and the government’s policies may lead to the electoral radicalization of center right voters in future elections. Such a rightward shift could cause the current CDU-SPD to collapse.

Mares is speaking to politicians, legal experts, and voters alike to gauge the magnitude of these electoral risks, and to present a wider “menu” of democratic possibilities when it comes to navigating the influence of extremist parties.  

The party ban isn’t the only option. Lawmakers could strip disruptive individual politicians of their parliamentary immunity, for example, or even change their country’s electoral system. France has previously employed the latter approach—the country introduced a two-round voting system in both the third and fourth Republic, which may feel to voters like the most democratic option.  

“You make it the same for all parties, and then you allow voters to make the decision,” she says, emphasizing that this gives voters the chance to make their thoughts known on extremist candidates and mainstream parties alike. “It’s still a fair system because the party puts itself at risk, but in the end, it’s the voters who decide.”  

As voters and political parties wrestle over what’s acceptable and desirable within their democracies, so too is society wrestling with how artificial intelligence may complicate voters’ participation in democracy.  

Daniel Karell, Assistant Professor of Sociology, is exploring the implications of AI for civic life and the social sciences.  

A recent study Karell co-authored found that summaries of historical events created by generative artificial intelligence shifted people’s opinions on those events more effectively than summaries written by humans—raising critical implications for voters and democracy more broadly.  

“Our starting point was this basic assumption that people are going to increasingly rely on chatbots to learn about the world,” he says. “For example, if somebody wants to understand, ‘Why was the Civil War fought?’ or ‘Why did Reconstruction fail?’ they would just ask a chatbot.”

Karell and his co-authors used generative AI to produce synthetic summaries of historical events. Some participants read those, while others read human-written summaries, like expert blog posts or Wikipedia articles. Karell notes that the study focused on historical events because they tend to have an outsized influence on the stories we tell about our societies. He and his co-authors also wanted to test historical summaries because they suspect that generative AI can be “a really good tool for creating alternative histories.”

“It's a pretty simple study, but I think it introduces really serious complications,” Karell explains. “This essentially means that we have people increasingly relying on these AI tools to inform themselves, but these tools can be created with whatever biases or perspectives that companies want—or a state could mandate companies to have particular viewpoints or stances.”

Joshua Kalla, too, is testing how AI may be used to try to persuade voters in the near future. A two-year project he’s conducted with ISPS has examined whether chatbots could be more persuasive than human canvassers, which could make the bots a significant political force because of their potential to be deployed at scale.

But he’s found that, so far, it’s difficult to replicate the potency of in-person discussions between voters, candidates, and volunteers. “What makes canvassing particularly useful is that someone shows up at your door on a random Saturday afternoon. You open the door, not knowing what they're going to talk about, but then you start having a conversation with them. I think it's hard to recreate that knock on the door,” Kalla says.  

He also doubts that chatbots are “suddenly going to change American politics” because of the challenges of getting people to interact with them at scale. Unlike a canvasser who shows up at your door, a chatbot requires people to choose to engage with it—which limits the number of people the bots can reach. “The persuasive potential of AI is real,” Kalla points out, “but the delivery mechanism is the bottleneck.”

Politics that ‘brings out the best in people’

If voters can have their minds changed by engaging in conversation, perhaps democratic countries should be creating more opportunities for voters to connect, discuss issues at length, and have a more direct say in policy and governance.

Hélène Landemore, Damon Wells ‘58 Professor of Political Science, is working hands-on around the world to help create those opportunities.

She's been investigating democracy since her first and second books, Democratic Reason and Open Democracy. These early works interrogated the properties of democracy and used formal and historical arguments to assert that democratic regimes ought to produce better solutions to collective problems than more authoritarian ones.  

Democratic decisions are meant to tap the collective intelligence of the larger public, Landemore says, and should yield “better informed decisions and laws that are more representative of the interests and preferences of the public.” After her intensive exploration of democracy, she ultimately concluded that, in order to maximize the problem-solving properties of our governments, we would be better off replacing elections of representatives with groups of citizens chosen by civic lotteries.

She then became active in the observation and governance of several such democratic innovations, including citizens’ assemblies in France. Citizens’ assemblies bring together large, randomly-selected groups of people for a sustained period—weeks or even months—to discuss policy issues like climate justice, urban planning, and assisted dying laws, then make recommendations to their government.

Landemore has since honed her views on the benefits and challenges of citizens’ assemblies, and how they could complement, and perhaps one day even replace, elected ones. These “mini-publics” could “fundamentally change the way we understand democracy and the way we practice it,” she says.  

Citizens’ assemblies reduce democracies’ reliance on elected officials, who, as a group, are often unrepresentative of the broader population. They also bring more and different people into the process of democratic deliberation and decision-making—and they tend to “bring out the best in people,” Landemore says.

“I think citizens’ assemblies foster an atmosphere of shared purpose and mutual respect that allows participants to form these bonds, even while they strongly disagree about the issues at hand. The structured deliberation, shared fact-finding, and collective responsibility for outcomes generates trust, solidarity, and even a renewed sense of patriotism,” she recently told Yale News about her work with citizens’ assemblies.

She has also witnessed the capacity of citizens’ assemblies to build trust between regular people and experts.  

Landemore has witnessed that participants in citizens’ assemblies in France, for example, received experts’ guidance “with some suspicion [at] first.” But the two groups developed a rapport as the experts provided sustained knowledge and support.  

“The relationship is transformative of both sides,” she notes. “The citizens learn to appreciate the expertise and knowledge and lose a lot of their initial suspicion. Meanwhile, the experts, some of them do come in with a certain posture of: ‘we know best and we're going to educate you.’ But then they learn, ‘actually, we don't know everything.’”  

The assemblies should be sure to have “citizens on top and experts on tap,” she emphasizes. “It’s really the citizens who have the final say about the recommendations that come out of the assembly, and the experts are just there to support their work and inform their deliberations. Not to decide in their place.”

Her newest book, Politics Without Politicians, is Landemore’s effort to share a more inspiring, deliberative, inclusive, and joyful vision of democracy than the system that is our current status quo—one focused on elite competition. It is also an effort to share the potential of citizens’ assemblies with the broader public in an approachable, even democratic way.  

Landemore is also the director of design and chair of the governance committee for the upcoming Connecticut Citizens Assembly on Property Taxes, which will be one of the first statewide citizens’ assemblies in the United States. It will bring together at least a hundred Connecticut citizens to discuss and deliberate how to fund and deliver public services like parks, schools, and police.  

She believes regular citizens, rather than elected officials, are uniquely equipped to face up to the complex trade-offs required to fund and deliver better public services. “The reality is that elected officials don't really have an incentive to run on these issues, because you're not going to get elected on the promise of raising taxes or sharing the taxes locally with other towns,” she points out. “So, we need the voice of the Connecticut citizens to say something like, ‘what do you think are the main problems, and how would you like to solve them?’”

Creating space for civic thought

While continuing to teach and research the past, present, and future of democracy, Gregory Huber says that faculty also play a critical role in fostering a culture of discussion and civic engagement across the entire Yale community.  

“We strive to be a place where people who disagree interact with one another, and even when they choose to disagree, do so in a constructive, useful, and civil manner,” he says. Even when academics are simply debating methodology or the state of an academic field, he adds, “When we engage in those debates in a structured, respectful, and professional way, we model a set of behaviors that are relevant for thinking about how we should interact with each other across partisan lines.”

That spirit of openness is embodied in a number of other programs supported by ISPS: the Yale Politics Initiative, a seminar series that invites notable people active in politics to Yale to meet with small groups of students; the Millstone Fellowship, which connects undergraduate students interested in public service with summer internships in government; and Yale's new Center for Civic Thought, which engages students and faculty in civic education by prioritizing small, rigorous conversations that are deepened by reading and linked to practical expertise.

“Many of these initiatives are designed to make sure that people on the Yale campus don’t find themselves vulnerable to the assumption that people on the ‘other side,’ who they may never talk with, are vastly different than they are,” says Huber.

When we do take the time to thoughtfully connect with people we disagree with, he notes, “we often learn, ‘oh, they agree with us about some things, they disagree with us about some things, but wow—they're all humans.’”