1. Defining the Problem(s)

Democracy requires shared legitimacy in electoral and governing institutions, regardless of who wins and who governs. For a metastasizing share of the American electorate and a growing share of the political class, this shared legitimacy is collapsing. The other party is not the opposition anymore. It is now the enemy and a threat to the well-being of the nation. They must be kept out of power at all costs. Such thinking is a hallmark sign of political extremism.¹

Some party polarization is healthy and normal in modern democracy. Societies are diverse. Voters benefit from meaningful alternatives. But contemporary American polarization is decidedly unhealthy. This is “pernicious polarization,” where all politics flattens into an all-encompassing us-against-them fight.² Democracies that have experienced similar binary partisan flattenings degrade quickly, often sliding into civil war and authoritarianism.³ Addressing other concerns, such as effective governing or, say, dealing with a pandemic or a climate crisis, becomes like arranging a bookshelf amid an earthquake.

But why has our politics grown so toxic and destructive? How we diagnose and define the problem shapes how we respond. The main argument of this paper is that we have a system-level problem. A system-level problem requires a system-level solution.

Because parties are the central institutions of modern representative democracy, and the dangers follow from this partisan-identity conflict, we must alter the dynamics of this conflict. We must “break the two-party doom loop.”⁴ This requires re-organizing and re-orienting our party system.

This first requires adjusting how we analyze and diagnose the problem. If “extremism” threatens the democratic system (it does), we must ask ourselves: Where is this extremism originating? What is its source?

Is it coming from the politicians? It is tempting to blame “extreme” politicians because we see such politicians constantly in the news (they are the darlings of social media and click-bait journalism). But why have elected officials become too extreme?

Other explanations blame the voters, particularly the primary voters. But then why have voters, especially primary voters, grown more extreme? And what’s the connection between extreme voters and extreme politicians?

Politicians shape what voters think. Voters constrain what politicians can do and what they should prioritize. It is a recursive feedback loop. But voters and politicians are connected by a core intermediary institution: the political party. Many definitions of political parties exist in political science, but conceived here broadly to include the emergent coalition of aligned interest groups, activists, and media, as well as the politicians and the formal party apparatus.⁵

Political parties do not exist in isolation. Parties compete against each other. Competition shapes them, as the parties seek new strategies in response to each other.⁶ But voters and politicians and interest groups and activists and donors also shape parties, frequently constraining them.⁷

Politics is a complex system, full of dynamic patterns and interactions. The complete system is not simply the mathematical aggregate of its parts. It operates as something more, perhaps almost alive, and certainly emergent.⁸ The individual parts (the voters, the politicians, the parties, and other groups) are meaningless in isolation from each other. Instead, they move in an evolutionary dance with each other.

This requires a higher unit of analysis: the system. The more typical (read: publishable) academic approach is to treat voters and politicians as independent actors for robust statistical analysis.

In journalism, the more common approach is also to treat politicians as independent actors, as subjects of individual stories or characters in a drama. Similarly, to the extent voters show up, they show up as individual actors too, in diners, focus groups, or as persons-on-the street. We don’t inquire into their sources of knowledge as much.

Reformers, relying on these studies and narratives, slice the electorate into different segments of politicians and voters. Then, by an electoral tweak or two (an open primary here, a redistricting committee there), maybe they can give “moderates” a boost over “extremists.”

But who counts as a moderate? And who counts as an extremist? It feels like we should all be able to agree. But definitions quickly become subjective. People often assume that they are reasonable and grounded in reality, while their opponents are extreme and based in falsehoods.⁹ Perhaps the only agreement we all share is a “my-side bias.”¹⁰ So what, again, is moderation?

Absent universal truths, pluralist democracy is the best alternative for those who are not sure they are right. It is the very process of democratic contestation and public debate that helps us define the boundaries of reasonableness and extremism. If we can’t have that debate, then we can’t have democracy.

Here’s where extremism comes in: “Extremism refers to the belief that an in-group’s success or survival can never be separated from the need for hostile action against an out-group. The hostile action must be part of the in-group’s definition of success.”¹¹ Extremists impose purity tests on their in-groups, and closely police their direct interactions with out-groups. Extremists highlight only the most negative aspects of out-groups, distorting reality into a dusky and murky funhouse mirror.

Over the last decade, we’ve seen a sharp rise in extremist behavior in American politics, particularly on the political right, with issues of race and national identity at the forefront. But again, why? Has something pushed the people into extremist anti-system thinking? What was extreme 10 years ago no longer seems so extreme today. How did so many people change what they consider acceptable? Surely, things will be different 10 years from now.

Perhaps the core problem is that extremism is not an individual-level property, but a system-level property. High levels of group polarization and high uncertainty are core drivers of political extremism.¹² Closely divided, razor’s-edge elections contribute to high levels of uncertainty.

One factor unites almost all the causal processes driving polarization: “Once categorical boundaries between ‘us and them’ are drawn, a whole host of destructive social processes may kick in.”¹³

But try to break the causal chain apart, and everything affects everything. An exhaustive review of recent literature on political polarization found a complex web of reinforcing causal mechanisms and cognitive biases driving partisan polarization.¹⁴ A similar finding comes from the Proceedings of the National Academy of Sciences (PNAS), which recently published a “Dynamics of Political Polarization Special Feature.” The special issue brought together a mix of political scientists and complex-systems scientists to explore how many aspects of polarization have self-reinforcing dynamics. In summarizing and integrating the findings, Scott Page observed how “each type of polarization strengthens the other through feedbacks.”¹⁵

When everything reinforces everything else, Page goes on, “no sequence of small interventions will likely reverse our course.” The core problem is a flattening of dimensionality, a fancy way of saying that everything has become reduced to a single us-versus-them fight, distributed bimodally. “As a rule, a bimodal distribution for anything suggests that something strange is afoot.” Several papers in the series note bimodal distributions are almost always signs of impending collapse in many types of systems.

“So what is to be done?” Page asks. “How do we escape our current situation? First, we must be aware of why we cannot chip away at this problem. We must take substantial actions.”¹⁶

Once politics tips into this “bimodal state” (two separate distributions), small interventions become pointless. Worse, they may even be counterproductive: “Once in a polarized state, well-intentioned attempts to improve interaction between groups may increase rather than decrease polarization, by encouraging the behaviors that pull people apart…Even if we could turn tolerance up to 11, so to speak, the polarized equilibrium cannot be escaped.”¹⁷

This is a hard insight to absorb, because it suggests that much of the current playbook of just defeating “extremists” in elections is simply taking buckets to a worsening flood.

“When individuals have a say in governance and individual agency, they become less prone to the all-or-nothing mentality that drives extremism.”

Scholars of extremism come to a similar conclusion. To reduce extremism, punishing extremists is not nearly as effective as changing the environment that produces extremism. As Susan T. Fiske argues: “We can change contexts more effectively than merely by restraining bad actors, simply incarcerating or executing a few evil-doers. As long as the wrong contexts prevail, other bad actors will rise up to take their places.”¹⁸ One way to reduce extremism is to reduce uncertainty, particularly around democratic representation and personal liberty. When individuals have a say in governance and individual agency, they become less prone to the all-or-nothing mentality that drives extremism.¹⁹

McCoy and Somer make a similar point in explaining how majoritarian winner-take-all elections drive pernicious polarization.²⁰ Indeed, a theme across multiple studies is that when crosscutting cleavages that sustain multidimensional political conflict and social relations disappear, and society bifurcates into two opposing groups, bad things happen.

These patterns are bigger than elevating certain actors over others. They are system-level properties. However, to view politics as a system means transcending the simpler solution: If only our side could secure additional victories, the opposition would inevitably concede and adopt a more moderate stance. This view has its advocates. A large industry of campaign consultants already exists to help win elections. This industry is embedded within the parties. A large industry of pollsters already exists to test campaign messages, also embedded within the parties. The righteousness of a good fight has a powerful appeal. But such thinking is precisely what drives the doom loop.

In a two-sided conflict, “myside” bias leads to excuses for extremism and antidemocratic behavior. By fixating on the dangers posed by the opposing side, we nurture the necessity of allegiance to our own camp.²¹ It is precisely under these conditions that partisans will tolerate antidemocratic behavior by their side, and thus perhaps unintentionally usher in illiberal authoritarianism.²²

Combatting belief polarization is arduous, and political parties play a significant role here. Political parties organize identities, helping individuals see themselves as part of a larger whole, and providing a shared identity to make sense of the world.²³ Being a ‘moderate Democrat’ or ‘moderate Republican’ is one thing, but identifying oneself as part of a ‘Moderate Party’ with a distinct set of values and priorities is something entirely different. Moderate Democrats are just like Democrats but less so. Moderate Republicans are just like Republicans, but less so. Moderates without a party are just disorganized and powerless dots on a scatterplot of the electorate.

Unfortunately, for those who don’t want to be pulled to extremes, no easy off-ramps exist. For strong partisans at all levels, the only option is to jump completely over to the other side. For example, Republicans who have left their party talk about how they are politically homeless, and how many of their old friends and family members have disowned them. This kind of break is rocky, and not for most people.

Without a stable middle ground, loyalty to one’s faction prevails. Offering more off-ramps allows for gradual belief adjustments, which are more likely than sudden reversals. It would also make politics crosscutting and multidimensional again, which is essential for healthy democratic functioning. Creating space for more parties would scramble the current dynamic by creating more possible political identities. More political parties would allow for flexible and incremental belief shifts that don’t involve repudiation. More parties would allow for new and shifting coalitions that can break out of the binary mindset.

System change must come through political parties—not individual politicians and voters. Political parties organize politics, not individual politicians or voters. Political parties structure the interactions of politicians and voters, and political parties interact with each other. Extremism is a product of a party system out of control. The system will not correct itself, and unorganized moderates are powerless to pull it back. Only political parties have the gravity and force to alter the dynamics. Therefore, we need more parties.

Upsetting a bimodal distribution that resists change will require considerable concentrated force. It can feel intimidating. Yet systems change is possible. All systems have a point of leverage. The two-party system in America is not an immutable law of nature. It follows from a particular set of electoral rules. Electoral rules are amenable to change.

The systems approach to politics treats politicians as actors within a larger system, adapting to the ebbs and flows of the system’s patterns and dynamics. This systems view differs from the moralistic American reform tradition that analyzes politics through individual actors, distrusts organized power, and demands full openness on the assumption that politics can operate as a true free market, just as neoliberal economists argued against regulations. The moralistic reform vision is inherently a narrative of personal responsibility and morality and individualism. It struggles with collective activity necessary to organize power and authority.

We must confront this reform tradition head on, and we will. But first we must understand what healthy and effective political parties actually do in modern mass democracy. As we will see, political parties in the United States have struggled to perform their crucial duties because of the constraints of a two-party system and past reforms that have hindered their capacities. Duopolies are not ideal in any organizational landscape.