4. The Contemporary Choice: Will We Repeat the Mistakes of the Past or Build Something Better for the Future?

We turn now to the present-day challenge.

It is familiar in the boiling discontent, the distrust of elites, and general antisystem feelings. It is distinct that hyperpartisan polarization and political extremism pose a second, and perhaps more urgent threat to the continued existence of liberal democracy in America.

As in previous moments of democratic discontent, aspiring reformers are promoting a wide range of proposals. Each reform has its own theory of the case. Each makes its own assumptions about politics and how voters and politicians might respond. Nonetheless, some theories of reform are fantastical, while others are better grounded in history, political science, data, and common sense.

Some reforms envision a kind of politics that has never existed before on a mass scale—a politics free of political parties, with everyone acting independently and with individual voters devoting unprecedented time and energy to political engagement.

Others, and I put myself in this category, start instead from the basic reality that most people don’t want to think about politics most of the time; that political parties are inevitable and necessary; and too much independent action makes collective governing impossible.

Certainly, reform experimentation is valuable. We must always approach reform with humility. Reforms also take time to work. But incremental reforms are unlikely to change system dynamics. If the problem is a system problem, small tweaks here and there are unlikely to have much impact if they do not get at the core dynamics driving the breakdown.

Very few reforms work exactly as intended, hence the skeptic’s lazy retort to any reform: “What about the unintended consequences?” To which the plucky reformer can always respond: “But maintaining the status quo also has unintended consequences.”

The year 2023 is not the year 1787. The collective knowledge of human civilization is exponentially larger, and this includes knowledge of how to manage the hard problems of large-scale self-governance. Just as the Framers attempted to learn the lessons of history and early political science in writing the U.S. Constitution, so we should also learn the lessons of history and political science in updating our democratic vision. We know plenty, based on what has (and hasn’t worked) elsewhere.

This paper has tried to distill some of the most relevant lessons from history and political science into a framework for evaluating contemporary reform paths. This framework starts with the centrality and inevitability of political parties in modern representative democracy. Political parties organize modern democracy. This is a necessary reality. Therefore, we should make political parties as good as we can—as representative, responsive, and effective in performing the essential roles that only parties can fulfill.

Political parties organize and cohere citizens. Political parties set the agenda and shape conflict. Political parties allocate and organize power. Power, organization, and agenda-setting are the central features of modern democracy.

We ignore these patterns at our peril. The idealistic American reform tradition may disdain organized power and dislike political parties as corrupt manifestations of power. It may wish for a more direct, unmediated democracy. But this wish is just that.

The realities are more complicated. Politics is a complex system with many interacting forces. Effective organization always translates to greater power than haphazard coordination. As the proverb goes, a single stick easily breaks. A bundle of sticks tied together resists considerable force.¹

Political parties are the most important source of organized power in modern democracy. Though many other sources of organized power obviously exist, politically they must interact with the political parties—usually as one group within a larger party coalition.

Pro-parties reform starts with this reality. Conceptually, pro-parties reform views the political system as a system, with parties as the central organizing institutions of this system. It says we need both better parties and a better party system.

This gives us a framework for evaluating any reform: Does it improve the party system? More specifically, does it encourage and facilitate healthy and effective parties that can perform their central role in modern representative democracy with honesty and integrity?

The goal of better parties, of course, is not better parties for their own sake. The goal is better parties for the sake of better self-governance through better representation, through … better parties. Better parties and more parties are merely the means. Better self-governance through better representation is the end.

More parties can stabilize the doom loop and produce system-level moderation by adding dimensionality to the political system and creating an opportunity for moderate parties to form on both the left and right, to pull politics back to the middle. Political parties offer identities that shape how citizens engage in politics, especially how they vote.

To calm down extremism and restore balance in politics, new moderate parties are necessary. Political parties are the institutions best equipped to forge compromises, build new identities, and break the doom loop. The problem of hyperpartisanship cannot resolve without partisanship, but different partisanship.

The alternative to pro-parties reform is pro-moderate candidates reform. The goal in these reforms is to solve a particular problem at a particular moment by finding a way around the political parties. In this view, the political parties are the problem. The solution is better candidates. Here, “better” means more moderate and more compromise-oriented. In particular, on the political right, it means “pro-democracy” and “non-MAGA” as defined by accepting the results of the 2020 election and not engaging in election denialism.

In this category, I put open primaries and ranked-choice voting, and the combination of open primaries and ranked-choice voting, Final Five Voting.

The thinking here is consistent with the long anti-parties reform tradition in the United States. This thinking looks at the dominant power within the two parties and tries to dislodge it by creating an alternative pathway specifically for more moderate candidates. In this manner, it is consistent with approaches previous generations of reformers took in trying to elevate political outsiders by taking away power from political insiders.

The great irony, of course, is that in the 1960s, the political outsiders sought to upset the balance of power within the parties because they thought both parties were too compromise-oriented. But the same theory holds, which is that parties are unnecessary intermediaries.

It would be better, this thinking goes, for voters to elect representatives who represent all their constituents, not just the hard-core partisans. The important thing is to elect representatives who can work on behalf of everyone and find reasonable compromises on all the important issues. Parties interfere with this.

Reforms in this space accept the two parties (and their failed states) as givens. Instead of trying to make parties better, they look for ways around the parties. These reforms focus on candidate recruitment and support, on changing primary elections, and on encouraging independent candidates. They focus on disintermediation, on “tearing open” the existing power structures, hoping to give voters (as a whole, not segregated by party) more direct authority over the system.

What they envision, either implicitly or explicitly, is something akin to the weak and incoherent parties of the 1970s, in which overlapping factions existed and moderates thrived in both parties. The theory is that by opening up the electoral process further, moderates that loosely affiliate with Democrats and Republicans can return to Washington. Some advocates of factionalized politics see this as possible even without electoral reform, merely by force of investment in faction-building.²

An obvious challenge here is that the parties of the 1970s were factionalized and overlapping because American politics was not yet fully nationalized, and because the parties were realigning. Liberal Republicans were common in New England and on the West Coast. Conservative Democrats were common in the South and in the West. Liberal Republicans and Conservative Democrats were the “moderates” because they were cross-pressured on many issues. Given the nationalization and sorting of politics since the 1970s, the Democratic Party brand is completely toxic in about 45 percent of the country, and the Republican Party brand is completely toxic in a different 45 percent of the country. Out-party hatred has transformed American politics.

This changes the meaning of “moderate” as well. In the past, a moderate was somebody who agreed with Democrats on some issues and Republicans on other issues because views on social issues, economic issues, and foreign policy were separate, and politics was multidimensional. The nature of moderation has changed in an era of one-dimensional, us-versus-them politics. To be a moderate is now to disagree with both parties, which is a lonely place to be. In the past, moderates just had a mix of views, and that was okay, because politics was multidimensional, and there were many ways to be a Democrat or to be a Republican.

Today, things are obviously different. There are still a few states and a few places where conservative Democrats or liberal Republicans survive: Alaska, Maine, Montana. But they are rare and endangered because the voting calculations that supported them have changed.

“Better” candidates who are at odds with their party label are not likely to lead to better parties. They are likely to lead to less-coherent parties, which will then have a harder time doing the things parties are necessary to do in a modern representative democracy, in organizing and structuring elections and governing. Parties that are incoherent are not likely to be effective. These failures lead to more frustrations with the existing parties, which fuel the antipartisan, antipolitician, antigovernment energies that make governing even more difficult.

Let us now turn to the candidate-centric reforms: open primaries, single-winner ranked-choice voting, and Final Five Voting.

I label these as candidate-centric reforms because they focus on changing the candidates who run and win elections. They are very unlikely to change the structure of the party system. These reforms are designed to work within the existing party system, in the hope that they can elevate more moderate candidates and rebuild the political center by “changing the incentives” and encouraging candidates to appeal to “all voters” more “broadly.”

Underlying these reforms are flawed assumptions about the electorate. Indeed, one could be forgiven for looking only at polling and observing that about a third of Americans identify as “moderate”³ and that over 40 percent identify as “independent”⁴ and think there’s a latent political center that is not being represented in our party system. However, few dig beneath these survey responses. Fewer than half of self-identified “independents” also describe themselves as “moderates.” Many self-identified moderates are quite happy to vote for their preferred party because their perception of “moderate” is partisan.

Often voters are asked to declare whether they are liberal, moderate, or conservative. Moderate is the default category for respondents who do not think of themselves as either liberal or conservative. As one recent book on political ideology explained, “the moderate category seems less an ideological destination than a refuge for the innocent and the confused.”⁵

Moderate does not mean centrist. It often means someone who doesn’t pay close attention to politics. Self-identified “moderate” voters are the least attentive to policy.⁶ This is likely because “moderates” in the voting population don’t have strong preferences. Typically, they don’t pay enough attention to politics to have well-defined ideological views.⁷

Independents, meanwhile, are on the whole more antisystem than they are moderate. Self-identified “independent” voters on balance are less compromise-oriented than registered partisans—their disdain for the political process turns out to be disdain over both partisanship and bargaining.⁸

Of course, some percentage of the American population is centrist, independent, follows politics, and has thought through the issues. This highly engaged profile is especially common in the world of democracy reformers and especially democracy-reform funders and their friends. However, it is not so common beyond these elite circles.

The more broadly shared view among the American public is generalized disaffection and frustration. Support for the two-party system is very low. Desire for more parties and more options is very high. But thoroughly considered ideology outside of the existing parties’ programs is hard to find.

Why do so few people hold thought-through political ideologies independent of the two parties? The answer is simple: To the extent that most people engage in politics, it is around elections, and thus they engage as partisans. People learn about political ideologies from elections and from listening to what candidates are talking about. If there were more parties, the public would have more policy ideologies to choose from. But with just two nationalized, sorted parties, only two dominant ideologies are likely to proliferate in the mass public.

Most people do not have the time or inclination to develop a coherent political view independent of the political parties. Instead, most people who don’t pay attention hold idiosyncratic views, which are often quite extreme on particular issues but held together by a distrust of politics. A new political party could help them form a more coherent and practical identity and link them with politics. Neither the Democratic nor Republican Party can accomplish this.

The challenge in interpreting the popular labels of “moderate” and “independent” is that they are both default choices for people who don’t identify in the other two categories offered in the standard poll question. If you don’t think of yourself as “liberal” or “conservative,” you are then “moderate.” If you choose not to identify as a Democrat or a Republican, you are then “independent.” Independent and moderate, therefore, contain multitudes. This is partly why no party has yet united around either of these labels, despite their popularity. There is not enough shared ground among those who identify thusly to overcome the formidable fortress that the single-member district and the pro-two-parties jurisprudence have enacted to discourage challengers.

However, there might be enough for three or four new political parties to form to represent the diversity of perspectives and identities that the “independent” and “moderate” labels are picking up. But without political parties to organize these perspectives into coherent identities, voters will not organize on their own, and individual candidates are unlikely to emerge without parties. And without changes in the electoral system, none of these alternative perspectives will have an opportunity to organize.

As we think about evaluating candidate-centric modes of reform designed to elevate moderate candidates, we must confront the reality that the labels of “moderate” and “independent” do not mean the same things to everyone. The share of true centrist, moderate independents with strong values may be high in the world of democracy reform elites. But it is low in the mass public.

Political reformers and funders are uniquely engaged, and many of them were drawn to democracy reform because of their genuine concern for the hyperpartisan polarization of American politics and the decline of our democratic system. Many “independents” in the mass public are disengaged because they see the same broken system but conclude it can’t be fixed, so why bother? To engage them in politics will take more than allowing them to vote in primaries—especially younger voters who are most likely to identify as independent and most likely to feel the current system has nothing to offer them.

We should not, of course, expect candidate-centric reforms to build more and better parties; this is not their goal, so it would be unfair to hold them to such a standard. However, we can evaluate candidate-centric reforms on their ability to elect more moderates. Lots of academic studies have evaluated their effects. Few have detected any systematic impact.

A core problem is that without partisan labels, voters resort to other mental shortcuts, such as name recognition, gender, race, occupation, home county, or other personal or biographic details.⁹ In short, take away partisan labels, and the advantage goes to well-known white men with lots of money, prestigious careers, and good looks.

Nonetheless, these candidate-centric reforms—particularly primary reform—have gained some traction because they seem feasible. If we can find some ways to make candidate-centric reforms more effective, or better understand where they can have the most impact, they could remain a valuable tool in our current democracy emergency. Thus, it is valuable to understand why these reforms have performed more like wrenches on a rusty bolt, but how some adjustments might loosen at least some stuck bolts.

“Open” Primaries

In much conventional wisdom, primaries are a main cause of polarization. The explanation is this: Since incumbents fear being primaried by more extreme candidates in their parties, many incumbents either adjust to ward off primary challengers, drop out in anticipation of such challenges, or lose.

Certainly, eliminating primaries altogether would make a tremendous difference. This would be a pro-parties reform. The inability of party leaders to select their nominees has made U.S. parties candidate-centric since direct primaries emerged in the Progressive Era.

However, no serious energy has gone into eliminating primaries altogether. Instead, reformers have focused their efforts on modifying the criteria for primary-voting eligibility and candidate advancement, yet these modifications seem to have minimal influence on who votes, who runs, and who wins.

Roughly half of U.S. states have some version of open primaries. Because states have experimented with different approaches to congressional primaries, political scientists have been able to use this variety to explore whether primary type affects the candidates who emerge under different primary voting systems.

In “open” primaries, voters can vote in the primary of their choice. Democrats can vote in the Republican primary, Republicans can vote in the Democratic primary, and Independents can vote in either. In “closed” primaries, one must be a registered partisan to vote in that party’s primary. There are also “semi-closed” and “semi-open” primaries. When combined with registration timelines, primary rules run the gamut from very permissive to very restrictive.

Still, the consensus across multiple studies is that primary type does not affect candidate ideology, on balance. The share of moderate candidates elected is consistent across different primary types. In a recent exhaustive report on primary reform, I reviewed dozens of studies; closed primaries do not elect more extreme candidates than open primaries.¹⁰

The moderation theory behind open primaries has a two-step logic: that if primaries are open, 1) more independents will vote; and 2) more independents voting will have a moderating impact on outcomes.

Turnout in open primaries is slightly higher than turnout in closed primaries. Regardless of who can vote, turnout is around 15-20 percent.¹¹ Since primaries happen many months before November elections, most voters are not following politics closely enough to know a primary is occurring. Most primaries lack substantial competition, resulting in little candidate effort to mobilize voters.

One piece of evidence in the independents-are-not-necessarily-moderates column is that independents voting in the 2016 Democratic primary preferred Bernie Sanders to Hillary Clinton, leading Sanders to call for an end to closed primaries. In the 2016 primary, Donald Trump did better in open-primary states than closed-primary states. Trump did best among voters tired of all the messy fighting in Washington.¹²

Others have argued that crossover voting at least gives voters from a party destined to lose in the general election a chance to vote in the primary of the other party. Wyoming, for example, had open primaries in 2022. Liz Cheney thus encouraged Democrats to cross party lines and vote for her in the 2022 Republican primary, which she lost despite some cross-over support. However, this can cut both ways. Partisans can also interfere in the other party’s primary to elect the more extreme candidate, hoping the more extreme candidate would be a general election loser. There is not much evidence of voters doing either. Mostly, the same reliable group of party-faithful voters turn up in primary after primary, regardless of the rules on who can vote.

The bottom line is that there is no clear evidence that converting a closed primary to an open one consistently results in more moderate winners. There may be normative reasons to prefer open primaries to closed primaries, such as that primaries should be open to everyone. But that is a separate debate.

Two-Round Nonpartisan “Primaries”

One type of “open primary,” however, deserves closer scrutiny. Except it’s not a primary. It is a nonpartisan two-round election.

California and Washington State use a “top two” primary, in which the top two candidates, regardless of party, advance to a general election. Alaska now uses a “top four” primary, in which the top four candidates advance to a ranked-choice voting general election. We’ll say more about the Alaska system a little later, since it is more complicated. For now, we’ll focus on the top-two system.

The basic logic of this approach is straightforward. Imagine an 80 percent Democratic district. Most likely, two Democrats would advance to the general election. Imagine one Democrat getting 45 percent, a second Democrat getting 35 percent, and a Republican getting 20 percent. If the 35 percent Democrat is more moderate, they would lose a straight Democratic primary. But the top-two system allows her to advance. And by courting the 20 percent of Republicans in the general election, she could win as the moderate.

Of course, whether this works as intended depends on two intermediate steps.

First, candidate percentage must break down in a manner to generate intraparty competition in the second round.

Second, opposite-party voters must vote for the more moderate candidate. This requires them to a) understand the difference between the two opposite party candidates; and b) be willing to vote for somebody of the opposite party, which they despise.

An incumbent Democrat in a lopsided Democratic district could try to clear the field of Democratic competition so that she will be guaranteed to face a Republican competitor. But if too many Democrats enter a lopsided Democratic district (seeing a likely November win), they could split the vote too much, so that only one Democrat makes it to the general election—that is, if only one Republican wins. But multiple Republicans could enter. Or none could, leaving the field to Democrats. It’s a lot of complicated calculation, based on what others are doing.

Same-party general election contests occur roughly one-sixth of the time, and when they occur, the moderate candidate is more likely (but far from guaranteed) to win. But political elites in the dominant party are often effective at clearing the field for their side, leading to lower levels of same-party competition than a simple partisan-voter index would expect.¹³

Why don’t moderates win all the same-party contests? A same-party contest may not advance a moderate candidate. A Democratic district might advance two very liberal Democrats; a Republican district might advance two very conservative Republicans. Differences may be minimal, noticeable only to engaged citizens.

Without clear labels helping voters to distinguish moderate partisans from more extreme partisans, many voters (who don’t pay close attention) struggle to tell the difference.¹⁴ Indeed, there is considerable evidence suggesting that voters can’t tell the difference without significant additional information—far more than they would typically receive.

But here is where hyperpartisan polarization kicks in. Democrats and Republicans stereotype and smear each other with a broad brush. To many Republicans, there are no good Democrats. To many Democrats, there are simply no good Republicans. Even with additional information available, many voters might be unwilling to hear it or unreceptive to the mere idea of crossing party lines.

Thus, studies estimate that between 40 and 50 percent of “orphaned” voters in top-two elections will abstain from voting in a race when their only option is a candidate of the opposite party.¹⁵ That is, almost half of Democrats see choosing between two Republicans as a lose-lose choice (so why bother?). Almost half of Republicans feel the same way when it’s two Democrats.

Absent the traditional party cues, money plays a more important role because advertising and name recognition become even more crucial. Notably, California’s shift to nonpartisan top-two primaries raised campaign expenditures considerably.¹⁶

Thus, the overwhelming research consensus is that the “top two” system has not produced more moderate representatives, on balance. However, this should not be taken to mean that a two-round system cannot moderate politics. As one of the more recent analyses of the California system concluded, “Taken together, these findings suggest that political scientists’ claims that the top-two primary has had ‘no effect’ are premature and that the key to the system’s effectiveness lies in reformers’ ability to find ways to encourage more same-party competition.”¹⁷

Making a two-round system work better for single-winner offices will involve working with political parties, not against them. This could, for example, involve letting multiple parties nominate one candidate for a first-round election, and then letting parties whose candidates do not advance to the second round an opportunity to offer their ballot line to one of the two finalists through fusion voting. Thus, a “Moderate Party” could compete in the primary, and if its candidate doesn’t advance to the general election, the two remaining candidates could compete for the ballot line. Allowing more parties to compete would help voters better understand the differences among candidates, since parties provide valuable cues.

But based on the evidence from California and Washington, the approach of elevating moderates within the two-party system by encouraging same party competition through a free-for-all two-round system has some significant limits, even if it sometimes succeeds in electing the more moderate candidate.

Later, we will turn to Alaska, which uses a Top Four system with ranked-choice voting. But first we must evaluate ranked-choice voting on its own terms.

Ranked-Choice Voting (Single-Winner)

A second reform that has generated considerable interest is ranked-choice voting in single-winner elections. Ranked-choice voting (RCV) is seen as a way to encourage more compromise and moderation in politics; to reduce negative campaigning; to introduce more competition within the two major parties; to end the spoiler effect of third-party candidates; and to ensure majority winners.

Single-winner RCV operates by having voters rank candidates by preference (first, second, third, etc.). If no candidate secures a majority of first-choice votes, the least popular candidate’s second-choice votes are distributed. This continues until a candidate achieves a majority of the remaining votes.

As more cities and states have adopted RCV, scholars have analyzed its impacts. The effects appear somewhat limited, particularly in partisan elections where the electorate is already polarized.

As a recent comprehensive report I coauthored on ranked-choice voting noted, “while single-winner ranked-choice voting does have many positive effects both in theory and in overseas usage, in practice these benefits have been somewhat limited and/or difficult to quantify based on limited usage thus far in the United States. On balance, the benefits of RCV outweigh the downsides, and RCV has many appealing qualities that make it a strong improvement over more traditional single-mark plurality voting. However, the benefits appear to be more marginal than many had initially hoped.”

RCV first gained traction in the early 2000s in several U.S. cities following the 2000 presidential election in which many blamed Ralph Nader for taking votes from Al Gore in Florida and handing the state—and the presidency—to George W. Bush. Had Nader supporters in Florida been able to vote for Nader first and Gore second, Al Gore might well have been the forty-third president of the United States. In 2000, both Gore and Bush presented as moderates. Bush promised to be a uniter (not a divider). He campaigned on his success working with Democratic state legislators in Texas. The third-party challenges were on the progressive and conservative sides—not in the center.

With extreme minor parties and center-oriented major parties, ranked-choice voting effectively allows noncentrist candidates to express dissenting views without affecting election outcomes. However, when major parties polarize, ranked-choice voting is less likely to benefit centrist minor-party candidates. The importance of elimination order and the requirement for significant first-preference votes work against small center-oriented candidates. In a divided political landscape, centrist candidates will struggle to secure enough first preferences to win.

Indeed, increasing evidence suggests that when RCV is attempting to bridge an enormous gap with a bimodal electoral distribution, centrist candidates do not do well. Once the electorate grows polarized, RCV ceases to be a potentially moderating force.¹⁸ Some even suggest it may actually increase the likelihood of candidates running to the extremes in many places, because in lopsided districts, RCV effectively insures them against vote-splitting by moderates.¹⁹ This assumes a centrist candidate even runs. As a candidate-centric reform, ranked-choice voting requires candidates to volunteer themselves to run a minor-party campaign almost certain to end in defeat.

Approval voting is sometimes pitched as a solution to this collapse of the center. But approval voting confuses the issue further because it relies on voters feeling equivocal about multiple candidates, something that no campaign wants to see. The game theory equilibrium of strategic campaign behavior under approval voting is that every voter should vote for just one candidate, which is what it frequently devolves to in practice. It changes nothing, but adds confusion.²⁰

Still, ranked-choice voting can work well in certain settings where politics is inherently candidate-centric and many candidates typically enter so that no candidate often gets a majority. In particular, ranked-choice voting would work well in: 1) nonpartisan local elections, where party labels do not exist by definition; and 2) party primaries, in which every candidate shares the same party label, making the label effectively meaningless. A mayoral primary in a technically nonpartisan city, for example, would be a perfect place for ranked-choice voting.

Local nonpartisan elections and primary elections also share a common feature: The differences among candidates are limited. Most cities, particularly the cities where ranked-choice voting has been implemented, are essentially one-party cities, with Democrats as the overwhelmingly dominant party. Although various Democratic factions exist in city politics, these differences pale compared to the vast divide between Democrats and Republicans.

Similarly, in a party primary, partisan factions are more similar than different. Such elections often attract many candidates, and without ranked-choice voting, an extreme candidate with narrow but committed support could win. Because the differences among candidates in such elections are relatively small, ranking is both a helpful way to find a consensus winner and can encourage coalitions among competing candidates. These limited differences mean that, to the extent RCV can build coalitions and consensus across candidates, the gap that it sets out to bridge is comparatively small.

Could a new centrist party emerge and thrive under RCV? Anything is possible. But it is not likely. In order to build a successful party under RCV, a party needs to attract compelling candidates who can expand the party’s following and reach. In order to attract compelling candidates, the party must offer a genuine path to victory for such candidates. No minor party can provide such a victory path now, even with RCV (and/or open primaries). Thus, no new moderate party is likely to emerge.

Over time, a hypothetical Moderate Party could build a stronger identity and draw in more voters. But a Moderate Party must first exist. Then it must gain sufficient influence to attract resources. If it comes and goes based on its ability to recruit strong candidates to stand for office (who will come and go), it is unlikely to build power and loyalty. Fusion voting solves this problem by establishing a ballot line that is not dependent on recruiting effectively doomed candidacies.

To be sure, ranked-choice voting lowers the barriers to candidate entry. Challenger candidates from outside the political mainstream can campaign without fear of being spoilers. This is a potential positive benefit of ranked-choice voting, though the extent to which RCV has truly stimulated more diverse candidates (across demographic and ideological categories) is still unclear, based on the very limited studies conducted. RCV stimulates new candidate entry initially upon passage. But over time, as outsider candidates realize it is more difficult to assemble a winning majority, the number of candidates declines.²¹

Notably, in Maine, where ranked-choice voting has been in place for three election cycles, new parties have not emerged, nor have existing third parties expanded their support. A few independent candidates have run. To be sure, minor party candidates have done slightly better in Maine than elsewhere. But the two-party system remains strong in Maine. No new parties have emerged yet.²²

Were national politics broadly centrist, ranked-choice voting could likely work to drain off extremist sentiment while preserving a dominant and moderate two-party system, as it has in the Australian House for a century. However, given the current hyperpartisan polarization, ranked-choice voting is unlikely to stimulate new parties.

Still, ranked-choice voting has some positive properties toward cooperation and compromise in candidate-centric settings where politics is already less polarized or devoid of distinct party labels. Ranked-choice voting is a candidate-centric reform and so works well within crowded candidate-centric environments, such as nonpartisan local elections and party primaries. This is where it should be used.

Open Primaries plus RCV—aka Final Four/Five Voting

In 2022, Alaskans voted for the first time under a new system that combined nonpartisan open “top four” primaries with ranked-choice voting in the general election. The system works as follows: The first round, the primary, is open to all candidates and all voters, similar to the “top two primary.” However, instead of just the top two candidates advancing, the top four advance to the general. Then the general election is decided by ranked-choice voting to ensure a majority winner. Parties have no formal role, but candidates can choose which party to affiliate with.²³

Alaska has now run one election under the “final four” system. We should, of course, be cautious in overinterpreting from a single election. And Alaska is unique in its politics in two important respects. First, it has a strong “independent” tradition, with by far the highest share of voters of any state considering themselves Independent (at 55 percent).²⁴ Second, it is unique in that the urban-rural divide is not a partisan divide. Republicans and Democrats do equally well in rural and urban areas.²⁵ This makes politics more multidimensional (since urban versus rural cuts across both parties), and thus supports the overlap that has made bipartisan coalition governing possible in the Alaska state legislature possible for many years now..

The three statewide elections in 2022 each yielded a different result. Alaskans elected a moderate Democrat to the House in its one statewide race, a moderate Republican to the Senate, and a conservative Republican to the governorship. This likely represents Alaska’s somewhat idiosyncratic politics.

In the high-profile Senate race, Lisa Murkowski, one of the few remaining moderate Republicans, was re-elected, 53.7 percent to 46.3 percent over Kelly Tshibaka, the Trump-endorsed Republican, after transfers. In the first round, Murkowski got 43.4 percent, Tshibaka got 42.6 percent, and a third candidate, Democrat Pat Chesbro (a retired academic) got 10.4 percent. A fourth candidate who had advanced to the general election, far-right Republican Buzz Kelley, dropped out in September and endorsed Tshibaka.²⁶

Under a traditional Republican primary, Murkowski likely would have lost to Tshibaka. The Final Four system enabled her to advance to the general election without leaving the Republican Party or running independently. Had she run independently, she would have likely defeated Tshibaka, since Democrats did not mount a significant campaign. This mirrors her victories in 2010 (as a write-in candidate) and 2016, when Democrats also mounted no serious campaign. Notably, Murkowski is the only senator elected three times without a majority of first-preference votes in any of those elections.

It is possible that the new rules helped to save Lisa Murkowski’s career. But Murkowski survived to be re-elected only because Democrats effectively stood down and supported her. Murkowski raised $11.2 million in campaign contributions and spent $10.4 million. Chesbro raised $188,164 and spent $178,681—which, in current campaign spending, is the equivalent of sneezing into the wind.²⁷

Murkowski’s reelection strategy relied on Democratic cooperation as much as the voting system. As a relatively moderate Republican incumbent, Murkowski is unique, and the Alaska system suits her re-election. However, few senators or representatives resemble her profile, and few states have Alaska’s unusual partisan political geography, in which the urban-rural divide is not a partisan divide.

In the House race, Mary Peltola, a moderate Democrat, won in both the special election (to replace Don Young) and the general election rematch. In both elections, Republicans split their vote between Sarah Palin and Nick Begich, with Palin slightly edging out Begich. Had Alaska used the old system of separate partisan primaries, Palin and Peltola likely would have won their respective primaries, and Peltola would have won the general election, since Palin was widely disliked statewide. Begich (considered the slightly more moderate Republican) might have defeated Peltola head-to-head. But order of elimination matters. As with the Senate race, the fourth-place finisher in the primary, Tara Sweeney (a moderate Republican) dropped out, seeing no path to victory.

In the governor’s race, Mike Dunleavy, a conservative Republican, was re-elected with a majority of first-round preferences. He almost certainly would have won re-election under the old system.

That Alaskans elected a moderate Democrat, a moderate Republican, and a conservative Republican to different offices in the same election speaks to Alaska’s unusual politics, which seems to be unusually personalistic and idiosyncratic. Alaska’s state legislature has also explored various approaches to bipartisan governing over the years.

It seems unlikely that new parties will emerge in Alaska, since aspiring candidates can grab the labels of the existing parties for free without having to build their own, thus further confusing voters as to the meanings of “Democrat” and “Republican.” Murkowski is not building the Moderate Party of Alaska.

In the state legislature, meanwhile, a bipartisan coalition leads the Senate as of 2023: nine democrats and eight of the 11 Republicans. However, Alaska has a long experience with bipartisan governing coalitions. Between 2006 and 2012, a bipartisan coalition ran the Senate.²⁸ Prior to the current session, a bipartisan coalition governed the House going back to 2013. Alaska’s state legislature has consistently been one of the least polarized, and with the highest share of elected Independents.²⁹ Notably, Alaska’s Senate, with just 20 members, is the smallest State Senate in the country, and its House, with just 40 members, is the smallest state House. In tiny chambers, members may be more likely to form personal connections, fostering bipartisan relationships.

Among states, Alaska’s strong independent streak and long history of bipartisan governance would make it a state most likely to adopt a reform like the top-four system. Alaska has the highest share of genuinely independent voters.³⁰ Obviously, evaluating a reform’s impact after one cycle is challenging. Short time frames likely yield limited changes, and one case is one data point. Politicians and parties adapt over time as they learn. It is also difficult to assess a reform based on a single state, especially when that state is unique in so many respects.

A top-five (Final Five Voting) version of this system also was approved narrowly by Nevada voters in 2022, though it will have to win again statewide in 2024 in order to become law, a daunting prospect since the initiative did not face any organized opposition in 2022, but almost certainly will in 2024.

The Problem of Candidate Entry in Candidate-Centric Reform

These candidate-based approaches to reform rely on candidates. But where do candidates come from?

In a candidate-centric political system, candidates must volunteer themselves and decide that they want to run for office. Who runs? Men are more likely to volunteer themselves than women.³¹ Rich people with connections, especially lawyers, are far more likely than working-class people to volunteer themselves.³² Political extremists and those who fit well with their party are more likely to volunteer themselves than moderates and those who fit poorly with their party.³³ White candidates are more likely to run for higher office than candidates of color.³⁴ Older individuals are likelier to run for office than younger ones.³⁵

Party leaders play some role in candidate recruitment. In a candidate-based system, party leaders seek self-sufficient candidates who can fundraise independently, connect directly with voters, and align with the party’s prevailing ideology. Party leaders and voters also must make calculations about candidate “electability”—a self-fulfilling proposition that cuts against candidates who don’t fit the traditional mold of white and male.³⁶ As a general pattern, the more open and inclusive the candidate selection process (across countries), the less likely female candidates and candidates of color are to be selected.

Many interest groups also play active roles in recruitment. These groups typically have more extreme policy demands. Indeed, political parties in the United States often outsource crucial candidate recruitment and support functions to outside groups, because such groups are eager to do so and parties themselves have limited resources to accomplish this.³⁷

In a more party-centric political system, political parties could do much more vetting and candidate development and more actively ensure that their candidate lists represented the diversity of the electorate.³⁸

A major shortcoming of candidate-centric reform is that it does not wrestle with the core question of the candidate pipeline and desire of candidates to run for office. Instead, the view assumes that candidates will just emerge if the opportunity presents itself. Yet, running for public office in the United States is a tremendous commitment and a tremendous personal sacrifice. In other, more party-centric democratic systems, political parties do far more to recruit and support their candidates. Most U.S. candidates launch their campaigns independently.

Candidate-Centric Reforms That Don’t Reshape Party Dynamics are Unlikely to Succeed

Candidate-centric reforms fit easily into the traditional American reform narrative of individualistic reform. In theory, all we need are better politicians, acting on their own judgment, connecting directly to their constituents.

In a short-term democracy emergency, with too many dangerously illiberal candidates, candidates obviously matter. Better candidates are indeed better. And short-term fixes that help better, pro-democracy candidates win, are valuable. But short-term fixes are just that: short term.

And just as hyperpartisan polarization and extremism in America have developed over many years, they won’t vanish instantly. They are certainly unlikely to disappear without addressing some of the system-level dynamics that continue to drive polarization and extremism—specifically the binary us-versus-them, all-or-nothing political conflict, which is a significant accelerator of extremist behavior.

It is certainly possible that in some particular, targeted circumstances, changing primary rules, adopting ranked-choice voting, or combining the two could help a particular (better) candidate advance, or work to defeat another (more extreme) candidate. In certain circumstances, rules changes can have meaningful consequences in a particular moment.

But if reforms are narrowly targeted to help (or defeat) particular candidates in a particular election, they may not hold up well under changed conditions. The history of party reform in America has been the history of factional fights, with reformers trying to open up the process to help their preferred candidates advance and without thinking much into the future.

Reforms built around advancing particular candidates or particular factions in particular moments are vulnerable to backlash. We see this in the history of ranked-choice voting in the United States. Between 1915 and 1948, many U.S. cities implemented ranked-choice voting, often using the proportional form of it. In every city save one (Cambridge, Massachusetts), RCV was eventually repealed by partisans who wanted to strengthen their party. Pro-RCV reformers didn’t think about how to organize the government. They did not build parties into their approach. But parties organize the government. Without parties committed to defending reforms (because their existence depends on such reforms), candidate-centric reforms can change based on the rise and fall of changing candidates. As Jack Santucci explains in his important history of ranked-choice voting in the United States, More Parties or No Parties: “Reformers will need some way to organize government. In the long run, reform may not outlast the reform coalition.”³⁹

We see this today in the Republican backlash to ranked-choice voting, which Republicans believe has helped Democrats.⁴⁰ In 2022, Tennessee and Florida banned ranked-choice voting. In 2023, at least four Republican state legislatures advanced legislation to ban RCV after the Republican National Committee came out strongly against it (following the success of Alaska Democrat Mary Peltola in defeating Sarah Palin).⁴¹ Similarly, Wyoming banned “open primaries” in 2023. In 2022, Liz Cheney had encouraged Democratic voters to cross over and vote for her in the Republican primary.⁴²

To be sure, the backlash against RCV, particularly after Alaska, reflects the deeper challenges of the party system. In a binary, hyperpolarized system, any change that even as much as appears to give one side an advantage gets sucked into the maw of hyperpartisanship. A reform that does not change the existing party system meaningfully may not survive the party system’s attempt to kill it. A reform that keeps the two-party binary is unlikely to help both parties equally. And when the reform reinforces existing partisan divisions, it is difficult to expand and to sustain.

Some reforms institutionalize an organized constituency whose success depends on the reform. These reforms are durable. Reforms that do not organize a constituency to defend the reform are not durable. This is a key finding in broader studies about reform sustainability. As Eric Patashnik has written in his landmark study of reform sustainability, reform is a dynamic process—“the losers cannot be counted on to vanish without another fight, and new actors may arrive on the scene who will seek to undo a reform to further their own agenda.”⁴³

Sustainable reforms disrupt long-standing patterns of governance, recast institutions, upset existing power monopolies, and create feedback effects that “render it difficult or unattractive for the government to reverse course.” Sustainable reforms bring about a “conscious, nonincremental shift.” Sustainable reforms break up existing orders, recast identities and political affiliations. Thus sustainability of reform depends on the “reconfiguration of political dynamics.”⁴⁴ Policy backlash is a real problem across multiple policy areas, especially in an era of high partisan polarization.⁴⁵

Unless reforms meaningfully reallocate power and reshape political affiliations, they are unlikely to be sustained. Incrementalism is as likely to lead to reversion as it is to build up. To endure, change in any policy area must realign power dynamics on the side of sustainability. The political losers in any policy change are unlikely to go quietly if they retain their organizational coherence and alignment. This explains why U.S. cities almost universally repealed ranked-choice voting within a few decades of enacting it, Santucci found. It did not alter the two-party system. And the major parties eventually killed RCV, because it undermined their power enough to be annoying, but not enough to build new and competing sources of power to protect it.⁴⁶ Indeed, as the preceding discussion of candidate emergence suggests, candidate-centric reform does little to change the candidate pipeline to create a permanent constituency to support any reform.

Changing the dynamics of the system involves thinking bigger than just changing candidates, or elevating some voters over other voters. It involves changing the party system. We now turn to that challenge.