chevron-down Created with Sketch Beta.
May 06, 2024 ABA Task Force for American Democracy

Decreasing the Political Polarization of the American Public

Richard H. Pildes

This Working Paper addresses the extent to which the American public has become polarized, the causes behind that and what innovations might be utilized to address it. Improving our democracy and making it more effective depends in part on overcoming the distortions and gridlock caused by partisan polarization within the electorate and among elected officials. Accordingly, the goal of this Working Paper is to capture as broadly and creatively as possible new approaches and fresh thinking that can help overcome the extent to which political polarization has paralyzed the public and its elected officials. While some structural innovations to counteract polarization and misalignment could occur in time for the next midterms, at least in some states, it is likely that significant structural reform—especially applicable to presidential elections—must be a longer-term project extending to 2028 and afterwards.

Problem Statement

To what extent has the American public become polarized? What are the causes behind that, and what innovations (e.g., top two voting, ranked voting, etc.) can be utilized to address it?


There’s an old saying that American politics is a game played between the 40-yard-lines, referring to a football field. While that used to be true, it no longer is, and instead America’s two political parties—and their voters—are moving further and further apart, as starkly illustrated in this graphic:

Democrats and Republicans more ideologicdally divided than in the past

Distribution of Democrats and Republicans on a 10-item scale of political values

Distribution of Democrats and Republicans on a 10-item scale of political values

Source: Survey conducted June 8-18, 2017. PEW RESEARCH CENTER

To stick with the football field analogy, it’s now as if the two parties are at the 20-yard-lines, or maybe even closer to the end zones. Democracy is more difficult when the two parties are polarized in this way—each located closer to the opposite ends of the political spectrum than to the middle. When voters face a choice between two 20-yard-line candidates, and the voters themselves are largely split into two polarized groups with only a relatively few voters left in the center, then even in a 50-50 “purple” state the election will produce an outcome that swings wildly one way or the other. A small shift one way or the other by this narrow slice of centrists can produce a victory for a candidate whose view are far from the middle. If in the next election there is a small shift in the opposite direction, then the change in the result is not incremental, but itself a huge leap from one 20-yard-line to the other.

These gyrations in electoral outcomes make it hard to maintain stable government policy development. Moreover, when a legislature is populated by politicians who have won their elections this way, compromise becomes increasingly difficult. When politicians from opposing parties are from their respective 40-yard-lines, finding compromise in the middle does not require either side to move very far. But when both sides are at their respective 20-yard-lines movement towards the middle is much more challenging. This problem of polarization is bad enough in a 50-50 “purple” state. But the situation is even more problematic by the added factor of uncompetitive states and districts. When a state or district is lopsided one way or the other, whether because of regional variation (Massachusetts being very different from Mississippi) or gerrymandering (the deliberate manipulation of districts to make them more lopsided), the consequence is that only the primary election matters. Under conditions of polarization, the primary will produce a nominee far from the center, and if the general election is not competitive, there is no electoral penalty for the nominee being further and further from the center.

These conditions of polarization in turn breed animosity. Politics is no longer respectful deliberation over policy alternatives among fellow citizens, but instead a vicious fight between antagonist tribes that view each other as villainous enemies. And when the nation as a whole is almost evenly split between these two warring sides, every election starts to feel “existential”—as if the survival of the country depends upon one’s own side defeating the other.

Some of the causes of polarization are sociological and beyond the capacity of electoral institutions to cure. Nonetheless, different electoral procedures interact with polarization in very different ways. Some electoral procedures magnify existing polarization within the electorate, producing outcomes even more polarized than the voters themselves. Other electoral procedures have the capacity to do the reverse: produce outcomes that are less polarized than the voters.

Given the dangers to democracy of polarization, including the threat it poses to the stability and effective functioning of government, it becomes imperative to explore the possibility of electoral innovations that can counteract polarization and mitigate its pernicious effects.

Innovations Aimed at Reducing Polarization and Misalignment

Alaska-Style Non-Partisan Primary with a Ranked Choice Voting General Election

All candidates who qualify for the primary ballot, regardless of their political party affiliation, run against each other in the primary. (The political parties separately may choose to endorse their preferred candidate on the ballot, but the ballot includes all candidates who self-identify as a member of that party.) All voters, also regardless of their political party affiliation, vote for whichever single candidate on the primary ballot they like the most. The four candidates who receive the most votes in the primary advance to the general election. (There can be two or more candidates from the same party who advance.) In the general election, voters cast a ranked-choice ballot, known as Ranked Choice Voting (RCV), which permits voters to rank the candidates in order of preference. Voters can choose to rank all four or rank a fewer number. The winner of the election is determined by an “instant runoff” procedure: if no candidate has a majority of first-choice votes, then the candidate with the fewest first-choice votes is eliminated, and the ballots that ranked the eliminated candidate first are redistributed to whichever other candidate is ranked second on the same ballot. This elimination-and-redistribution procedure is repeated until a single candidate is the top-ranked remaining candidate on a majority of the ballots.

In addition to Alaska, a five-candidate variation of the same basic system has received approval by a majority of voters in Nevada, but Nevada’s constitution requires that the same proposal receive a second vote of approval from the electorate before taking effect. Both of these changes to the state’s primary system were brought about by citizens’ initiatives.

Alaska has had only one year of experience with its new system so far. The evidence indicates that it had somewhat of a moderating effect. For example, moderate Republican incumbent Senator Lisa Murkowski prevailed, whereas in the traditional electoral system she likely would have lost the GOP primary to her Trump-endorsed opponent. (She might have been able to prevail in the general election as a write-in candidate, as she did in 2010, when she lost the GOP primary to a Tea Party challenger.) In an important state senate election, a moderate Republican similarly was able to prevail who otherwise would have lost. With respect to the two elections held for the state’s single seat in the U.S. House of Representatives, one a special election to fill a vacancy and the other the regular election, the Democrat won and Trump-endorsed Sarah Palin lost—which can be interpreted as a rejection of the most polarizing candidate—but there was another more moderate Republican who did not win, and who would have won the special election if a different and more anti-polarization form of ranked-choice voting had been employed. This victory by the Democrat against two Republicans, who combined had more first-choice votes than the Democrat, has angered Republicans both in the state and nationally about the use of ranked-choice voting.

In this regard, it is important to emphasize that the “instant runoff” procedure used in the Alaska system is only one version of ranked-choice voting and other versions (including the “Maximum Consensus Voting” procedure described below) would operate less detrimentally to Republicans in this kind of situation and has greater anti-polarization tendencies. Various electoral reform organizations are, however, actively pursuing the adoption of this system in other states through the process of citizens’ initiatives.

Alaska-Style System with a “Consensus Voting” Form of Ranked Choice Voting

This system would be the same as Alaska’s system described above, except for the method of identifying the winner from the ranked-choice ballots. In other words, the procedure for the nonpartisan primary would be identical, and the top four candidates would advance to the general election, where voters would cast ranked-choice ballots. The difference would be that the election’s winner would be identified by a “Consensus Voting” tabulation method.

“Consensus Voting” is a currently-used term for a centuries-old concept most often associated with an eighteenth-century French philosopher (Marquis de Condorcet), but actually dates back even earlier to a thirteenth-century Majorcan thinker (Ramon Llull). The idea is that if there is one candidate among several whom a majority of voters prefer to each other candidate (that is when compared one-on-one), this single most-favored candidate should be elected. The term “Consensus Voting” is appropriate for this idea insofar as a candidate who is most-favored in this way is the candidate who achieves the broadest possible consensus across the entire electorate. For this reason, it is the electoral method most effective at combatting polarization and avoiding a misalignment between the electorate’s preferences and the election’s outcome.

Perhaps the simplest way to introduce Consensus Voting within an Alaska-style system would be to add what we can call a “Consensus Proviso” to the “instant runoff” procedure that is currently used to tabulate the ranked-choice ballots. The Consensus Provisio would simply stipulate as a matter of law that whenever the ranked-choice ballots identify one candidate whom more voters prefer to each other candidate on the ballot, when each pair of candidates is compared head-to-head, then this single most-favored candidate wins the election. Otherwise, when no such candidate exists (as can occur infrequently), then the basic “instant runoff” procedure elects the winner. If Alaska had employed this Consensus Proviso as part of its electoral system in 2022, then one of the Republican candidates—Nick Begich—instead of the Democrat would been the winner of the special congressional election given the ranked-choice ballots cast, and the Republican backlash to RCV would have been a lot less severe and perhaps largely avoided.

There are other ways besides the Consensus Proviso to modify an Alaska-style system so that it elects the most-favored candidate defined according to the principle of Consensus Voting defined above. One way would be to adjust the “instant runoff” procedure so that instead of eliminating the candidate with the fewest number of first-choice votes on the ranked-choice ballots, eliminate instead the candidate with the fewest total votes on the ranked-choice ballots. This Total Vote Runoff variation of the basic “instant runoff” method was explained in a Washington Post column and elaborated upon in a subsequent law review article. An arguable advantage of this method, in comparison to simply using the Consensus Proviso, is that it will tend to elect a less polarizing candidate somewhat more often. It also arguably is more straightforward insofar is it just one procedure, rather a procedure with a separate proviso added on.

Another way to tabulate ranked-choice ballots that most closely conforms to the principle of Consensus Voting, insofar as it seeks always to elect the most consensus-achieving candidate, is one we could call Maximum Consensus Voting. Its tabulation procedure elects the candidate who, when compared one-on-one to each other candidate, either is never disfavored by more voters or is disfavored by the narrowest margin when measuring the widest margin by which each candidate is disfavored against another. A candidate preferred by a majority of voters compared to each other candidate is necessarily elected by this procedure. When there is no such candidate, the winning candidate is the one who comes closest to being preferred by a majority of all voters compared to each other candidate, because the winning candidate is the one defined by being the fewest voters away from a majority to the extent that a majority prefer another candidate to this one. Defined this way, the winning candidate is the one who provokes the minimal disagreement among voters—and thus achieves the maximum possible degree of consensus. For this reason, Maximum Consensus Voting is the procedure that most directly counteracts the polarization that pulls voters apart rather than bringing them together.

By contrast, without some form of Consensus Voting procedure, the “instant runoff” method used by Alaska can easily produce polarized outcomes. To return to the football field analogy, assume that the four finalists are one candidate on each of the 40-yard-lines and one candidate on each of the 20-yard-lines. With Alaska’s instant runoff procedure, if the electorate is polarized and the two candidates at the 20-yard-lines have the most first-place votes—but are also ranked last by voters on the opposite side of the spectrum—then the two candidates at the 40-yard-line will be the ones eliminated, and the winner of the “Instant Runoff” procedure will be one of the two 20-yard-line candidates. Conversely, insofar as all voters prefer either of the two candidates at the 40-yard-line to the election of the candidate at the 20-yard-line who is opposite to them, the Maximum Consensus Voting method will elect one of the two 40-yard-line candidates and not a 20-yard-line polarizer.

This difference in outcome between the basic Alaska method and Maximum Consensus Voting is based on the exact same rankings on all the ballots. In other words, the electoral “inputs” remain the same; it’s just the electoral “output” that’s different. One output, Alaska’s method, more closely replicates the existing polarization among voters. The other output, Maximum Consensus Voting, remediates this polarization and produces a winner more aligned with the views of the entire electorate.

A ballot initiative campaign could be organized to promote the principle of Consensus Voting and why it would lead to better representative democracy on behalf of all citizens. A ballot initiative campaign could be aimed either at adopting a specific form of Consensus Voting, like the Consensus Proviso or Maximum Consensus Voting, or instead it could simply mandate the use of some form of Consensus Voting, leaving to the state’s legislature the choice of which form of Consensus Voting to adopt. Indeed, states could experiment with different forms of Consensus Voting and, as laboratories of democracy, learn from each other’s experience.

Ranked Choice Voting in a California-Style “Top Two” Non-Partisan Primary

California uses a non-partisan primary just like Alaska’s, except that only two candidates advance to the general election (where a simple ballot, requiring voters to express their choice between the two finalists, is used) instead of four (requiring RCV in the general election). Both California and Alaska use basic plurality voting for their nonpartisan primaries, meaning that voters only identify which single candidate they like the most, and the tally of ballots is simply which two (or four) candidates get the most votes.

But it would be possible to use RCV in a California-style top-two primary. Depending on which form of RCV were used, it could greatly reduce the effect of polarization. Using basic plurality voting in the top-two primary mimics the polarization of the electorate, because the top two candidates are likely to be further left and further right in a polarized society—at the 20-yard lines, instead of the 40-yard lines. Thus, the choice in November would be limited to a choice between two extremes. By contrast, if Maximum Consensus Voting were the form of RCV used in the top-two primary, the two November finalists would be the two least polarizing candidates.

Using RCV in a top-two primary would also permit a hybrid form of RCV that combined two different tabulation methods. As indicated above, the regular “instant runoff” method that Alaska uses is very different from any Consensus Voting form of RCV. Because the science of election system design has identified pros and cons for each method (unfortunately, there is no completely perfect tabulation method for RCV ballots), a top-two system enables the two November finalists to be the “winner” of each method based on the same ranked-choice ballots. Voters in November would then themselves make the direct head-to-head choice between which of these two candidates they wish to elect.

The potential impact of this type of system depends on the type of RCV used. If only the regular form of instant runoff is used in the nonpartisan primary, then its capacity to reduce polarization is significantly limited. As discussed above, two polarizing candidates are likely to prevail if the RCV elimination process is based on fewest first-choice preferences, rather than fewest total votes or some other tabulation method consistent with the Consensus Voting principle. But as long as one of the two finalists is chosen by a method consistent with Consensus Voting, then the general election voters in November will have the opportunity to elect an anti-polarizing candidate. If both finalists in November are chosen by a Consensus Voting method, like Maximum Consensus Voting, then the election’s winner is assured to be a depolarizing candidate.

A ballot initiative campaign in a given state could lead to the adoption of this approach, which in its hybrid form could be seen as a compromise between two different versions of RCV, and therefore potentially more palatable to Republicans.

A Top-Three General Election after a California-Style Nonpartisan Primary

It is also possible to conduct an election involving three candidates in a way consistent with the principle of Consensus Voting but without using ranked-choice ballots. The election would have voters directly express their preferences between each pair of candidates: A versus B, B versus C, A versus C. The election’s winner would be whichever candidate prevails in the pairwise comparison against both opponents. In the rare case that each candidate would prevail in only one pairwise comparison, creating a kind of three-way tie, then pursuant to the method of Maximum Consensus Voting the election’s winner would be whichever candidate’s single pairwise defeat has the narrowest margin of defeat. A top-three general election of this nature could follow a California-style nonpartisan primary in which the top three candidates advance to the general election.

Because it is a Consensus Voting method, this proposal would also significantly counteract polarization as long as one of the three candidates advancing to the general election was a nonextreme candidate. It would also facilitate creation of the kind of third-party being advocated by Liz Cheney, Mitt Romney, Joe Manchin, and others seeking an alternative to the current two-party system. This reform could be adopted in a variety of states by ballot initiatives. Given the current interest in the possibility of a third party, this reform could generate significant enthusiasm.

Maine-Style Partisan Primaries & Consensus Voting General Election

Like Alaska, Maine uses RCV for its general elections. But unlike Alaska, Maine has partisan rather than nonpartisan primaries. This makes a big difference, given the ordinary “instant runoff” form of RCV that both Alaska and Maine employ. Partisan primaries are not nearly as effective as counteracting polarization as nonpartisan primaries, because partisan primaries prevent nonpolarizing candidates from the same party from competing in an election involving the entire electorate instead of just the partisan primary electorate. This difference is especially pronounced when the form of RCV in the general election is the kind of instant runoff that Alaska and Maine use rather than a Consensus Voting form of RCV. In the Maine system, the winners of the partisan primaries of the two polarized major parties will win the RCV general election, reproducing the existing polarization of the electorate. If the two major party nominees are at the 20-yard lines, not the 40-yard lines, then the Maine system will force a choice between the two 20-yard-line candidates.

By contrast, if a Consensus Voting form of RCV is employed in the general election, then even with the use of partisan primaries, there is a chance for a centrist nonpolarizing party to form and compete effectively. If this new third party occupies the space between the 40-yard lines, while the two longstanding major parties remain stuck at their respective 20-yard lines, then the new third party will quickly become powerful in a Consensus Voting system. The third-party nominee likely would be preferred by voters from either side of the polarized divide than the major party nominee on the opposite side of the divide, and thus the third-party nominee would win using a Consensus Voting method for tabulating the ranked-choice ballots. This centripetal effect of the Consensus Voting procedure would force the two major parties to move closer to the center in order to be competitive against the new third party. Thus, the overall effect of the electoral system would be to depolarize the competition between the political parties.

In sum, without Consensus Voting, nonpartisan primaries are essential to redressing polarization. But with Consensus Voting, even partisan primaries can be part of a system that counteracts polarization. As indicated above, without Consensus Voting the Maine system will be ineffective in counteracting polarization, but with Consensus Voting it will.

Assuming a Consensus Voting form of RCV is used in the general election, is it better to have Maine-style partisan primaries or Alaska-style nonpartisan primaries? There are arguments on both sides of this question. In favor of nonpartisan primaries is the openness of the competition among all candidates. The entire electorate is not limited in its choice of which Republican or which Democrat it prefers, even if it also has a genuine option of a third-party candidate. By contrast, the argument in favor of partisan primaries (according to many political scientists) is that parties channel electoral competition more efficiently. There isn’t the danger of two Democrats or two Republicans splitting votes, as there is in a nonpartisan primary that permits multiple candidates from the same party to compete against each other. In a Maine-style system with Consensus Voting, each party would present a single nominee who best represents that party, and general election voters would choose the nominee of whichever party best represents the collective views of the whole electorate.

Independent Citizens’ Redistricting Commissions

An independent citizens’ redistricting commission is distinctive insofar as it is populated by ordinary citizens, like a jury, instead of by professional politicians. The goal is to have these citizens draw fair maps that are not gerrymandered in favor of either political party. Independent Citizens’ Redistricting Commissions have been implemented in Arizona, California and Michigan as well as other states by ballot initiatives. In Ohio, there is a current effort to adopt this kind of commission after the failure of a previous politician-controlled commission. In California, Governor Schwarzenegger was instrumental in getting that state’s proposal adopted and in Michigan, it was one citizen who instigated a social media campaign in favor of the proposal.

Where adopted, citizen commissions have been successful in removing the power to draw district lines from politicians who have a self-interest in gerrymandering those lines to their advantage. These commissions have been accused of being captured by political interests and not truly independent, but these accusations have tended to come from those who would benefit from partisan gerrymandering and thus oppose the implementation of a fairer system. On balance, there is good reason to believe that these commissions are a net benefit compared to leaving the redistricting power in the hands of self-interested politicians. The specific details of how a citizens’ commission is designed, and its members chosen are important, and any future adoption of this approach should benefit from lessons learned with the experience of existing commissions.

The idea of independent citizens’ redistricting commissions has been gaining momentum in light of public frustration with gerrymandering. Congress came very close to adopting this approach as part of its omnibus “For the People Act” electoral reform, but it failed when Democrats in the Senate refused to negate the filibuster for this type of reform. Had Congress separated this proposal from the rest of the omnibus bill, it is at least conceivable that there would have been 60 votes in the Senate to overcome the filibuster, but that effort was never made. Depending on the outcome of elections for the rest of this decade, there potentially might be the opportunity to try again to enact this proposal in Congress.


“Self-districting” is a system in which voters choose which communities they would like to join for purposes of legislative representation. These communities could be based on geography, like traditional legislative districts, but they also could be organized around other attributes of social affiliation, like occupation, age, ethnicity, or ideology. For example, there could be a “farmers” district, or a “millennials” district, or a Latino district, or a “Constitutional Conservatives” district.

The way the system would work would be, first, for groups to self-organize and to compete for voters to join them. Given the constitutional requirement of equally populated districts, groups with more voters joining them potentially would be entitled to more than one district. If a group had enough voters for three districts, for example, the group would be subdivided by a districting computer program according to strict geographical criteria, like compactness. In order to comply with the constitutional requirement of equally populated districts, voters who did not choose to join any of the self-organized groups plus voters who joined groups that weren’t large enough to form even one district—and also “remainder” voters when a group’s size was large enough for n districts but not large enough for n+1 districts (these “remainder” voters would be randomly selected from all the voters joining the particular group)—would be placed in one or more residual “unaffiliated” districts. If there were more than one unaffiliated district in a state, they too would be drawn by computer according to strict geographical criteria without regard to any partisan considerations.

After all the districts were formed according to this self-districting method, candidates would run in each district, and the election’s winner would represent that district in the legislature (either a state legislature or Congress, depending on whether the self-districting system was adopted for one or the other, or both). This self-districting system would tend to resemble a form of proportional representation, as each group would win a number of seats in the legislature in proportion to the number of voters who joined that group. The elections to fill these self-districted seats would tend to resemble party primaries, insofar as all the voters in the district would be self-affiliated in their group and choosing which individual would best represent the interests of the group in the legislature. A version of ranked-choice voting would be most desirable for these elections, but any electoral method that selects a single winner from a pool of candidates would work. The elections for the residual “unaffiliated” districts would tend to resemble more traditional legislative elections in the United States, since the voters there would not be all self-affiliated according to the same social attribute; presumably, political parties would nominate or endorse candidates to compete in those districts.

This is a new idea, although New Zealand has adopted something similar insofar as its indigenous population gets to choose between two different sets of legislative districts.

A self-districting system would eliminate gerrymandering because the voters, not politicians, would choose which district to join. Geography would be used only to subdivide the groups formed according to the voters’ own preferences (as well as the few residual unaffiliated districts necessary to comply with the requirement of equally populated districts), and even this constrained use of geography would be based on strict computer-implemented nonpartisan criteria and thus would not enable the manipulation of district lines for partisan purposes. Moreover, because the basic self-districting decision would be decentralized, with each voter choosing which group to join, there wouldn’t be the risk of the process being captured by any political party—a risk that inevitably exists even with independent citizens redistricting commissions insofar as the selection of citizens to serve on the commission might be susceptible to partisan manipulation, or the performance of the citizens on the commission might be subjected to partisan influence.

Under the Constitution, each state has the power to adopt this kind of self-districting system if it wishes. Moreover, with respect to congressional districts, there would be no need to repeal the existing statute that requires single-member districts for Congress. Each district in this self-districting system is represented by a single member and thus satisfies this existing statutory requirement. Indeed, self-districting has the attractive attribute of being a way to achieve proportional representation in Congress without eliminating the single-member district requirement.

Multi-Member Districts with Ranked-Choice Voting

Instead of one representative elected in a legislative district, it is possible to design the electoral system so that voters elect several members—for example, four or five—from the same district. When the system combines these multi-member districts with ranked-choice voting, the system can achieve a kind of proportional representation. To illustrate this point, assume a district elects four members and there are a dozen (or more) candidates competing for these four seats. If voters cast ranked-choice ballots, candidates can be eliminated one at a time until only four remain. Candidates do not need to have the support of the majority in order to win a seat; any candidate who receives over 20% of the votes will be sure to get a seat, thus providing representation to this share of voters in the district.

This system is used to elect Australia’s national senate as well as one chamber of Ireland’s national legislature. It’s also used in local elections in New Zealand, some parts of the UK (Scotland and Northern Ireland), and some U.S. cities, including Cambridge, MA. It was used more widely among U.S. cities during the Progressive Era (1896-1917), but it fell on disfavor and was repealed in most places where it had been adopted. It was also used for a period in some local Canadian elections, but also repealed there. In the United States, there was a concerted effort among Progressive Era reformers to adopt this method, as described in Santucci’s book.

Advocates for this reform contend that it would combat polarization and gerrymandering and can point to its success in Australia and Ireland. The widespread repeal of its limited adoption in the U.S., however, is a sobering reminder that what works elsewhere may not necessarily work here. Among those who favor introducing some form of proportional representation into U.S. legislative elections, either for Congress or state legislatures, there is a debate as to whether this particular form of proportional representation would be the best one to adopt. The mathematical properties of this system make it more difficult to form coherent legislative coalitions, according to Santucci’s analysis, which is arguably one reason for its repeal in the U.S. and Canada.

The organization Fair Vote is a zealous advocate for this particular reform, and there is a bill pending in Congress to adopt this method for House of Representatives elections. This bill, however, has little chance of passage in the near future. It would make more sense to experiment with this method in one or more state legislatures, and it is worth observing that despite strenuous advocacy for this reform for decades, it has not yet been adopted for any state legislature. Perhaps that will change: it has recently been adopted in Portland, Oregon, and Oregon’s voters will have the opportunity to vote on the system in a statewide ballot initiative in 2024.

Nineteenth-Century-Style Fusion Voting

Fusion voting is the practice where a candidate can be nominated by more than one party and voters can vote for the candidate as the nominee of whichever party they prefer. The practice originated in the nineteenth century, at a time when each party printed its own ballots, before the United States adopted the policy of a government-printed ballot. If more than one party nominated the same candidate, then more than one party would print their own ballots with the same candidate. All the ballots cast for the same candidate would count towards the candidate’s overall total, although each party could tell how much it contributed to the candidate’s total.

After the adoption of the government-printed ballot, some states permitted the practice of fusion voting to continue, whereby the same candidate’s name would be printed separately for each party nomination the candidate received. The U.S. Supreme Court has permitted the abandonment of this practice. Nonetheless, New York and Connecticut continue to use it. (Oregon uses an alternative form of fusion voting, which lists with the candidate’s name all the parties that have nominated the candidate; but this alternative form does not permit voters to distinguish which party they wish to support when the vote for the candidate, and thus also does not permit knowing how much each party contributed to the candidate’s total.)

Advocates for reviving fusion voting contend that it would facilitate the development of a “Moderate Party” that could decide in each election whether to co-nominate either the Democrat or the Republican. In doing so, the theory goes, this Moderate Party could pull some votes for a Democrat whom Republican voters would be reluctant to support (and vice versa). If these voters cast their ballots for the candidate as the nominee of the Moderate Party, they don’t have to “pull the lever” for a “Democrat” if doing so is distasteful to them.

There are reasons, however, to be skeptical of this argument. For one thing, a Moderate Party has not developed in either New York or Connecticut, where fusion voting is currently permitted. Instead, fusion voting has facilitated the development of more “extreme” parties that flank the Democrats and Republicans on either side of the ideological divide. Moreover, there is little evidence that significant numbers of Republican voters currently unwilling to vote for a Democrat would be willing to vote for the same candidate if that individual were also nominated by the Moderate Party. Finally, any anti-polarization benefit that might be obtained by fusion voting could be achieved more effectively by a form of ranked choice voting, where voters could cast their ballot for their most preferred candidate and then also indicate second-choice support for a Moderate Party candidate. In this context, the Moderate Party could form an alliance with one of the other parties in an effort to secure such second-choice support, and voters predisposed to support a Moderate Party candidate might in turn be willing to support whichever other party the Moderates made an alliance.

It seems unlikely that voters would be inclined to support fusion voting, which tends to result in a more complicated and confusing ballot without significant perceived benefits from the voter’s perspective. There is an effort underway to convince state courts that fusion voting is required by state constitutional provisions requiring free and fair elections. This litigation effort also faces an uphill battle and applied to congressional elections, potentially risks invalidation by the U.S. Supreme Court under the so-called “independent state legislature” doctrine (which was rejected in its most extreme form but remains as a constraint against interpretations of state constitutions by state courts that lack a sufficiently judicial character).

Consensus Voting for Presidential Elections

Some American voters are clamoring for a meaningful third-party option in the 2024 presidential election, but the existing electoral system is not structured to enable that. It would be necessary to incorporate Consensus Voting into the presidential election system, either in a RCV form or the direct “top-three” alternative to RCV, in order to make feasible a competitive third-party presidential candidate.

Given the Constitutional provisions establishing the Electoral College, however, it is difficult to incorporate Consensus Voting within presidential elections. States currently have the power to adopt Consensus Voting for the appointment of their presidential electors. But if they did this, they would significantly increase the risk of producing a three-way split in the Electoral College, causing no candidate to receive an Electoral College majority, thus sending the presidential election to the House of Representatives under a procedure that gives each state’s House delegation a single vote regardless of population. California, in other words, would have the same single vote as Wyoming. This procedure was last used in 1824, long before California and Wyoming were added to the union. Most observers think it would be disastrous for another presidential election to end up in the House with this archaic procedure in force.

There are two possible ways to adopt Consensus Voting for presidential elections without risking this antiquated House procedure. The first would be a constitutional amendment. The second would be a variation of the National Popular Vote Interstate Compact (NPVIC), which would be a coordinated reform among states with at least 270 Electoral College votes and therefore capable of controlling the outcome of the presidential election without sending it to the House. Currently, NPVIC has been adopted by 16 states (plus DC) with 205 electoral votes, 65 short of the 270 necessary for it to take effect.

The existing version of NPVIC would award all the compacting states’ electoral votes to the plurality winner of the national popular vote. But a Consensus Voting variation on the same idea would award all the compacting states’ electoral votes to the Consensus Voting winner among all ballots cast in the compacting states. Thus, a candidate who was the consensus choice among all the voters of these states necessarily would win an Electoral College majority and thus the presidency. Once enough states joined the compact to control 270 votes, the rest of the states would need to join as well in order to make sure that their voters were not shut out of any chance to influence the outcome of presidential elections. A map of where the existing version of NPVIC has been adopted can be seen here. NPVIC was proposed after Bush won the Electoral College in 2000 while losing the national popular vote.

A Consensus Voting variation on NPVIC would transform presidential elections, making it possible to elect a president who truly is the consensus choice—and thus most representative—of all voters. Depending on what happens in the 2024 presidential election, there might be significant appetite for reform.

Survivor-style Presidential Elections

The TV show Survivor involves dropping contestants one by one until only the winner remains. The key to this procedure is that each round of voting is based on which contestant should be eliminated, not which contestant should win. This same concept could be used in presidential elections, either in the form of RCV or in a series of separate votes (or both, at the option of each voter). While not exactly the same as Consensus Voting, under conditions of polarization Survivor-style voting would have much the same anti-polarization effect. Polarizing candidates strongly opposed by voters on the opposite side would tend to be the first candidates eliminated in a Survivor-style system, leaving more common-ground candidates in the race for voters to choose among. Like a Consensus Voting system for presidential elections, a Survivor-style system would need to be adopted either by constitutional amendment or a variation on the NPVIC plan.

As indicated above, it would depolarize presidential elections. Also, the participatory nature of voting each presidential candidate off the stage, one by one, potentially could increase civic participation and enthusiasm for presidential elections. The immense popularity of the TV Show Survivor, plus the familiarity of Americans with its voting method, suggests that there is a good chance the public could be persuaded that this makes sense for presidential elections.

European-Style Proportional Representation

In many countries, voters elect members of a legislature by voting for a party, and the party receives a number of seats in the legislature in proportion to the number of votes it receives. There are different specific versions of this party-based approach. In “closed” versions, the party alone chooses which of its candidates receive seats in the legislature, depending upon how many seats the party earns from the vote. In “open” versions of the system, voters choose which of the party’s candidates receive the seats the party earns from its proportion of the overall vote. Given the historically candidate-centered, rather than party-centered, nature of elections in the US, it is highly unlikely that Americans would want to adopt a “closed” version of this system. But it is conceivable that Americans could be persuaded to adopt an “open” version of proportional representation. Indeed, the “self-districting” system described above can be understood as an even more “America-tailored” variation on a “open” party-based proportional representation system. Rather than voting for parties without there being any specific electoral districts whatsoever, self-districting enables a measure of proportional representation insofar as parties attempt to compete for districts within the system. A pure proportional representation system, like those in some European countries, abandons electoral districts altogether; even in “open” versions of the system, voters do not choose a specific representative for their own district, but rather contribute to the choice of which candidates form a party’s overall share of the legislature. The connection between representative and constituent is thus “severed” in a pure proportional representation system, even an “open” one, whereas the connection between representative and constituent is maintained in the self-districting system.

Norway, Sweden, Denmark, Belgium, the Netherlands, and Austria are among the many European countries that use some variation of a party-based proportional representation system.

There is a debate among political scientists and other scholars about the extent to which the adoption of proportional representation in the US, were it to occur, would ameliorate existing levels of polarization and improve the functioning of government. Many political scientists strongly advocate in favor of proportional representation, arguing that it would isolate extremist elements of American politics, liberating moderate conservatives from the dominance of the “MAGA movement” within the Republican party. Others argue that proportional representation tends to destabilize government, as legislative coalitions fail to form or break apart quickly. There is also the concern about an extremist party earning enough seats in a proportional representation legislature to extract concessions to become part of a majority coalition. Another point of debate is how proportional representation would function in the context of America’s overall constitutional system, which necessarily includes a president and US senate (as well as state-level gubernatorial elections) that structurally cannot be subject to any form of PR.

Even if a European-style proportional representation system is thought to have desirable attributes, it would seem more likely that Congress or states would adopt a version of proportional representation, like self-districting, with features more familiar to American elections.

Jury-Style Citizen Assemblies

Today, when Americans think of democracy, they think of elections. But there are other forms of democracy beside government by elected representatives. A New England town meeting, in which each adult citizen participates, is one form of direct democracy. Other is the ancient Athenian idea of “sortition”—which is a fancy term for government by means of randomly selected ordinary citizens. A jury is an example of government by sortition, but currently we use juries for factfinding in connect with the judicial process. It is conceivable, however, to populate legislative assemblies with lawmaking authority by means of sortition—randomly selecting ordinary citizens to serve in a lawmaking body.

Given the problems afflicting elections in recent years—from the distortions of campaign finance to gerrymandering—there has been growing interest among political theorists about the possibility of using sortition rather than elections as the means to populate legislative assemblies that would be representative of the body politic as a whole. Some of this interest has been confined to consider citizen assemblies, with members chosen randomly, to be advisory or supplementary to existing traditional elected legislatures. But the more ambitious of these proposals would be to have a citizens assembly chosen by lot replace one or more branches of the traditional bicameral legislature.

Citizen assemblies have been used in Canada, Scotland, and elsewhere with varying degrees of success for the purpose of proposing government reform and other major policy changes. Private-sector advisory citizen assemblies, in the form of “deliberative public opinion polls,” have been conducted in the United States, based on the work of Stanford political scientists Larry Diamond and James Fishkin.

Depending on how ambitious the particular form of this proposal might take, its impact potentially could be enormous. If elected legislatures were entirely replaced by jury-like citizen assemblies, the effect would be transformative. Election campaigns would disappear (at least for those offices to which this reform applied). The relationship between citizens and their randomly selected representatives would be altogether different as well. If citizens were unhappy with the public policies chosen by the randomly selected representatives, they would have to hope that the next term’s random selection would produce better outcomes.

Even if citizen assemblies did not entirely replace elected legislatures, their impact could be hugely significant. Imagine if a state legislature were composed of two chambers, one traditional elected and the other chosen by sortition. Insofar as the two chambers diverged on the public policies they pursued for the state, the public as a whole would have the opportunity to assess which of the two better served its interest. The sortition-selected chamber thus would serve as a check on distortions caused by flaws in the electoral process, while at the same time the elected chamber could attempt to convince the public that it had a better understanding of the public’s needs that the randomly chosen citizens assembly.

This proposal is obviously “radical” (in the basic sense of term) insofar it has the potential to change the way government works all the way down to its roots. Therefore, it necessarily is less likely to be adopted, at least on a widespread basis anytime soon. Even so, given how deep the sense is that government currently is not working, there may be opportunities for at least some exploration of this very different alternative. Even if there was only a more sustained use of citizen assemblies as advisory to existing legislatures—imagine a “shadow Congress” composed of randomly chosen ordinary citizens who proposed legislation for the actual Congress to consider—this idea has the potential for paving the way to improve government decision making.

Australia-Style Compulsory Voting

This one is simple: instead of permitting citizens to vote, the government could require citizens to vote—obligating them to pay a modest fine if they don’t. The justification for this idea is that voting should be viewed as a duty of citizenship as well as a right. Australia is the most prominent country to have adopted compulsory voting but there are other countries that have done so as well. In Australia it was enacted by legislation and was sponsored by individuals concerned by the previous drop in voter turnout.

If adopted in the United States, compulsory voting would have a dramatic effect on voter turnout, especially in nonpresidential elections, which tend to have low turnout. Moreover, if voting were compulsory, debates over facilitating access to the ballot would be transformed, as the government would be obligated to make it easy for voters to fulfill this civic duty. Also, because turnout rates differ significantly among different socioeconomic groups, compulsory voting would equalize the electorate across all groups.

The United States historically has had a libertarian streak to its national culture, and thus the idea of compulsory voting has been viewed as contrary to the American character. But the change in perspective on the functioning of democracy would be so significant if this reform were adopted that perhaps it is worth considering.

This document has been submitted to the Task Force for American Democracy for consideration and has been posted and/or circulated for information purposes only. The views expressed herein represent the opinions of the author(s) and not those of the Task Force or the ABA. They have not been reviewed or approved by the House of Delegates or the Board of Governors of the American Bar Association and, accordingly, should not be construed as representing the position of the Association or any of its entities. This publication is freely available to download, copy and distribute provided there is attribution to the ABA Task Force for American Democracy, and provided this notice is reproduced on all copies.

    Richard H. Pildes

    The Ohio State University

    Edward B. Foley holds the Ebersold Chair in Constitutional Law at The Ohio State University, where he also directs its election law program. He is a contributing opinion columnist for the Washington Post, and for the 2020 election season, he served as an NBC News election law analyst.


    His most current book, Presidential Elections and Majority Rule (Oxford University Press, 2020), excavates the long-forgotten philosophical premises of how the Electoral College is supposed to work, as revised by the Twelfth Amendment to the U.S. Constitution, and then uses this historical analysis to provide a feasible basis for reform of state laws that would enable the Electoral College to operate according to majority-rule objectives it was designed to achieve.


    His book Ballot Battles: The History of Disputed Elections in the United States (Oxford University Press, 2016) was named Finalist for the David J. Langum, Sr. Prize in American Legal History and listed as one of 100 “must-read books about law and social justice”.


    As Reporter for the American Law Institute’s Project on Election Administration, Foley drafted Principles of Law: Non-Precinct Voting and Resolution of Ballot-Counting Disputes, which provides nonpartisan guidance for the resolution of election disputes. He has also co-authored Election Law and Litigation: The Judicial Regulation of Politics (Wolters Kluwer 2014).


    During his fellowship at Stanford University’s Center on Democracy, Development, and the Rule of Law, Foley wrote Due Process, Fair Play and Excessive Partisanship: A New Principle of Judicial Review of Election Law, 84 U. Chicago Law Review 655-758 (2017), which was cited in briefs in Gill v. Whitford and Benisek v. Lamone (the Supreme Court gerrymandering cases). In addition to his Washington Post opinion columns, his op-eds and other essays have appeared in the New York Times, the Atlantic, Politico, and Slate, among other publications, and he frequently writes online commentary on election law issues of public interest.


    Foley clerked for Chief Judge Patricia M. Wald of the U.S. Court of Appeals for the D.C. Circuit and Justice Harry Blackmun of the United States Supreme Court. He has also served as State Solicitor in the office of Ohio’s Attorney General, where he was responsible for the state’s appellate and constitutional litigation.


    Professor Foley is a graduate of Columbia University School of Law and Yale College.