The Two-Party Trap

The Law That Wasn’t a Law of Nature

In 1951, the French sociologist Maurice Duverger published Les Partis Politiques, the comparative analysis of electoral systems and party formation that produced what subsequent political science has come to call Duverger’s law.¹ The law’s central claim was empirical and structural: single-member-district plurality electoral systems tend, across observable cases, to produce two-party competition; proportional or preferential systems tend to produce multi-party competition. The mechanism Duverger identified was twofold. The mechanical effect was that single-member-plurality systems, by definition, allocate one seat per district to the candidate with the most votes; smaller parties whose support is broadly distributed cannot win seats even when their cumulative support is substantial. The psychological effect was that voters, recognizing the mechanical effect, strategically defect from preferred-but-non-viable smaller parties to support the larger party that approximates their preferences and that has a realistic chance of winning. The combined effect, across electoral cycles, is the consolidation of competition into approximately two stable parties whose competitive equilibrium is reinforced by the structural features of the underlying electoral system.

Duverger’s law was, when published, a generalization from observable cases. Subsequent political-science research across seventy-five years has substantially confirmed the generalization across the comparative-electoral-systems literature.² The cases that have appeared to violate the law — Canada’s persistent third parties, the United Kingdom’s Liberal Democrats, the more recent fragmentation of European party systems — have, on closer analysis, generally involved either regional concentration of third-party support (Canadian Bloc Québécois, Scottish National Party) that produces the geographic concentration the mechanical effect requires for non-major-party seats, or the underlying transition of the electoral system away from pure single-member-plurality (the UK’s adoption of mixed systems for devolved parliaments, the European parliament’s proportional systems). The law’s structural prediction — that pure single-member-plurality systems produce two-party competition — has held across the comparative cases.

The American electoral system is, in operational terms, the most thorough application of single-member-plurality competition in any major democracy. The U.S. House of Representatives has, since the Apportionment Act of 1842, been elected from single-member districts in every state except for the brief period (1842-1929) when the requirement was imperfectly enforced.³ The Senate is elected through statewide single-member-equivalent contests (one or two seats per state, but elected separately rather than through any form of proportional or preferential mechanism). The presidency is elected through the state-by-state Electoral College, which operates as a series of state-level winner-take-all contests in forty-eight of the fifty states. Every state-level legislative chamber in the United States is, with limited exceptions, also elected through single-member-plurality (a few states retain multi-member districts for specific chambers, but the modal pattern is single-member-plurality across all levels of American government). The combined system is the structural condition that produces, by Duverger’s law, the two-party competition that has organized American politics across essentially the entire modern political-party period.

The two-party pattern in American politics is, in this respect, not a constitutional feature of American government — the Constitution makes no reference to political parties and does not require single-member-plurality elections — but a statutory and administrative feature that has been progressively established across approximately one hundred and eighty years. The 1842 Apportionment Act’s single-member-district requirement for House elections was the first major statutory step. The progressive standardization of state-level election administration across the nineteenth century was the second. The post-Reconstruction restructuring of state-level electoral systems in the South, which eliminated multi-member districts in many state chambers as part of the broader white-supremacist consolidation of post-Reconstruction politics, was the third. The progressive elimination of the few remaining multi-member districts in northern state legislatures across the early twentieth century was the fourth. By approximately 1960, the American electoral system had reached the operational form in which it has continued to operate: single-member-plurality across essentially all levels of government, with the resulting two-party competition that Duverger’s law predicts and that the contemporary American politics has demonstrated.

The structural feature of the system that the contemporary debate has been slowest to integrate is that the two parties whose competition the system produces are not, in any structural sense, predetermined. The system produces two-party competition; the system does not specify which two parties. The Whig-Democrat competition of the antebellum period gave way to the Republican-Democrat competition of the post-Civil War period through a structural realignment in which the Whig coalition collapsed and was replaced by the Republican Party — a process that took approximately twelve years (1848-1860) and that was driven by the slavery crisis the Whig coalition could not absorb. The post-Civil War Republican-Democrat alignment has continued, in nominal terms, for one hundred sixty years; the substantive coalitions the two parties represent have shifted dramatically across the period (the Democratic coalition’s transition from the Solid South of the late nineteenth century to the Civil Rights coalition of the post-1960s period; the Republican coalition’s transition from the post-Civil-War urban industrial coalition to the post-Reagan suburban-and-rural coalition). The system has produced two parties; the identity of the two parties has been structurally contestable, on a multi-decade horizon, when the underlying coalitions the parties represent have decomposed and re-formed.

The contemporary moment is, on the multi-decade horizon, a moment in which the structural decomposition of both major coalitions has been visible across the post-2008 period. The Democratic coalition has experienced the progressive realignment of its working-class component to the Republican coalition (most visibly in the 2016 cycle and across the subsequent realignments) and the addition of educated-suburban voters who had previously voted Republican; the Republican coalition has absorbed the displaced Democratic working-class voters while shedding the educated-suburban component that had been part of the Republican coalition since the Reagan period. The contemporary major-party coalitions are, in operational terms, substantially different from the major-party coalitions of even fifteen years ago. The structural pattern across the period has been the realignment of the existing two parties’ coalitions rather than the emergence of a third party that displaces one of them; this is the operational pattern that Duverger’s law would predict. The system continues to produce two parties even as the two parties’ substantive coalitions decompose and reform.

The Mechanical Distortion

The mechanical effect of single-member-plurality electoral systems produces a specific category of distortion that the comparative-electoral-systems literature has documented at length, and that the contemporary American political debate has largely treated as anomaly rather than as structural feature. The distortion is the systematic mismatch between vote shares and seat shares that the mechanical effect produces, particularly in the legislative chambers whose districts are subject to substantial drawing discretion.

The mismatch operates through two related mechanisms. The first is the conversion of vote shares to seat shares through the geographic distribution of voters across districts. A party whose support is geographically efficient — distributed broadly across many districts at modest majorities — wins a higher seat share for a given vote share than a party whose support is geographically inefficient — concentrated in a smaller number of districts at large majorities. The mechanism produces, structurally, the asymmetric outcomes that the contemporary American legislative landscape exhibits. The Democratic coalition’s geographic concentration in dense urban areas, which is the cumulative outcome of the post-1960s migration patterns and the post-2000 sorting of college-educated voters into urban centers, produces structural inefficiency in the conversion of Democratic vote shares to House seat shares. The Republican coalition’s broader geographic distribution across suburban, exurban, and rural areas produces structural efficiency. The combined effect, which has been the operational outcome across the post-2010 cycles, is that the Republican coalition has typically held House majority seat shares with vote shares that, in some cycles, have been substantially below the Democratic coalition’s national vote share.⁴

The second mechanism is the deliberate manipulation of district lines through gerrymandering — the partisan drawing of district boundaries to maximize one coalition’s seat share while minimizing the opposing coalition’s. The contemporary American gerrymander, enabled by the GIS-based district-drawing technology that has been operational since approximately the 2000 redistricting cycle, has produced district maps whose precision in concentrating opposing-party voters into a small number of districts (called “packing”) and dispersing them across multiple districts that the drawing party can win at modest majorities (called “cracking”) has substantially exceeded the historical norms.⁵ The 2010 redistricting cycle, in which the Republican Redistricting Majority Project (REDMAP) effectively organized state-level Republican legislative majorities to control the post-2010-census redistricting, produced district maps in several states (North Carolina, Pennsylvania, Wisconsin, Michigan, Ohio) that subsequent litigation found to be partisan gerrymanders of substantial magnitude.⁶ The Supreme Court’s 2019 decision in Rucho v. Common Cause held that partisan gerrymandering claims were political questions beyond federal judicial review, effectively foreclosing federal-court remedy for the partisan gerrymanders that the redistricting cycles have produced.⁷

The combined effect of the geographic-distribution mechanism and the gerrymandering mechanism is, in operational terms, the substantial decoupling of legislative-seat-share outcomes from popular-vote outcomes. Across the post-2010 redistricting cycle, the Republican coalition has typically held House seat shares that exceeded the Republican national vote share by approximately three to seven percentage points; comparable patterns have appeared in many state legislative chambers.⁸ The decoupling is not the outcome of a single redistricting decision; it is the structural consequence of the mechanical effect operating against the contemporary American political-geographic distribution, amplified by the gerrymandering technology that the post-2010 period has institutionalized. The decoupling is the principal structural feature of contemporary American legislative politics that the contemporary policy debate has been slowest to address.

The third mechanism is the partisan primary system, which has come to dominate American candidate selection across the post-1960s period. The progressive replacement of party-organization candidate selection with direct-primary candidate selection, which the early-twentieth-century Progressive movement substantially advanced and which had become the operational form of American candidate selection by the mid-twentieth century, has produced a structural feature that the comparative-electoral-systems literature has identified as the principal driver of contemporary American polarization.⁹ Partisan primaries, in operational terms, allow the most ideologically committed members of each coalition (the population that turns out at the highest rates in low-turnout primary elections) to disproportionately influence which candidates are nominated for the general election. The mechanism systematically advances candidates whose policy positions are further from the median voter than the candidates a more inclusive selection mechanism would produce. The combined effect, across the post-2000 period, has been the substantial polarization of American legislative politics — measured in roll-call votes, in legislative comity, in the willingness of legislators to negotiate across party lines — at rates that exceed the polarization the underlying voter distribution would predict.¹⁰

The Reform That Works

The comparative-electoral-systems literature has, across decades of analysis, identified a small number of structural reforms that address the mechanical and gerrymandering effects the previous section described. The reforms are not theoretical; they operate in the comparator democracies whose structural electoral outcomes the previous sections have described. The American refusal to adopt them has been the operational pattern of the post-1990 American democracy-reform debate.

The first reform is ranked-choice voting (RCV), also called instant-runoff voting or preferential voting. Under RCV, voters rank candidates by preference rather than selecting a single candidate; the votes are tallied across multiple rounds, with the lowest-vote-share candidate eliminated each round and that candidate’s voters’ subsequent preferences distributed to the remaining candidates, until a candidate achieves majority support.¹¹ The mechanism eliminates the strategic-voting dilemma the Duverger psychological effect produces: voters can rank a preferred-but-non-viable smaller-party candidate as their first choice without the risk of effectively voting against their second-choice candidate, because the second-choice ranking is registered in the subsequent rounds. RCV has been the standard parliamentary-election system in Australia for over a century, has been adopted for state and federal elections in Maine (2018) and Alaska (2022), and has been implemented at the municipal level in approximately fifty American cities.¹² The empirical record from the implementing jurisdictions has consistently demonstrated reduced strategic voting, expanded candidate competition, and the absence of the catastrophic outcomes the system’s opponents have routinely predicted.

The second reform is multi-member districts elected through proportional or semi-proportional methods. Under multi-member-district systems, multiple legislators are elected from each district, allocated proportionally to the vote shares received by competing candidates or parties. The mechanism eliminates the geographic-concentration distortion the single-member-plurality system produces: a coalition’s proportional seat share corresponds, structurally, to its proportional vote share, regardless of geographic distribution. Multi-member districts are the standard legislative-election system in essentially every major democracy except the United States, the United Kingdom, and Canada. The American historical experience with multi-member districts (which operated in many state legislatures through the early twentieth century before being progressively eliminated) is, in this respect, anomalous in the comparative record.¹³ The reform’s contemporary American advocacy, principally through the FairVote organization and the broader proportional-representation reform community, has not been adopted at meaningful scale.

The third reform is independent redistricting commissions, which remove the line-drawing authority from the partisan legislatures that have, in the post-2010 cycle, produced the gerrymanders the Rucho decision foreclosed federal-court remedy for. Independent commissions have been adopted in approximately a dozen American states (most prominently California, Arizona, Michigan, and Colorado), through state-level ballot initiatives that have generally passed by substantial bipartisan majorities.¹⁴ The empirical record of the independent commissions has been substantially positive on the principal measure: the district maps produced by the commissions have been, by the standard partisan-bias metrics, substantially less biased than the comparable partisan-legislature-drawn maps in comparable states.¹⁵ The reform has been adopted at the state level through ballot initiative in approximately the dozen states that permit citizen-initiated constitutional amendment; the reform has not been advanced through the legislative process at the state or federal level in the great majority of American jurisdictions.

The fourth reform is automatic voter registration, paired with same-day registration and universal vote-by-mail. The reform addresses the participation-distortion mechanism that the contemporary American voter-registration system produces: the requirement that voters affirmatively register before voting, in many states with substantial procedural barriers, that produces the lowest voter participation rate of any major developed democracy.¹⁶ Automatic voter registration, which has been adopted in approximately twenty American states across the post-2015 period, registers eligible voters when they interact with state agencies (typically the motor vehicles department or the social-services agencies); same-day registration allows voters to register at the polling place on the day of the election. The combined reforms have produced, in the implementing states, substantial increases in voter participation, particularly among the demographic categories whose participation rates have historically been lowest (younger voters, voters of color, voters with lower incomes, voters with disabilities). The reform’s structural effect is the reduction of the participation gap between the voting and non-voting American populations, which is the contemporary American distortion that the comparative-democracy literature has identified as among the most consequential.

The four reforms, considered as a package, address the principal structural features of the contemporary American electoral system that the previous sections diagnosed. The reforms are not unprecedented; they operate at scale in essentially every comparator democracy. The reforms are not constitutionally prohibited; the relevant federal authority over House election rules was held by the Supreme Court in Arizona State Legislature v. Arizona Independent Redistricting Commission (2015) to permit congressional or state-level adoption of any of the reforms.¹⁷ The reforms have not been adopted at the federal level. The structural reasons for the non-adoption are the structural reasons that operate against every fourth-settlement reform in the broader series: the political coalitions whose competitive position the existing system protects are the coalitions whose support would be required to enact the reforms.

The Two Parties’ Two Failures

The Democratic-coalition position on electoral reform across the post-2010 period has been substantially better than the Republican-coalition position on the same questions, in operational terms; the asymmetry is real and is reflected in the platform-comparison summary the platform’s democracy position presents. The asymmetry has not, however, produced enactment of the structural reforms the previous section described, because the Democratic coalition has been substantially constrained by the same structural features the reforms would address.

The Democratic coalition’s specific failures on the structural reforms have followed a consistent pattern. Federal voting-rights legislation has been advanced repeatedly by Democratic-controlled congressional majorities — the For the People Act in 2019 and 2021, the Freedom to Vote Act in 2021-2022, the John Lewis Voting Rights Advancement Act across multiple Congresses — and has consistently failed in the Senate to overcome the filibuster.¹⁸ The legislative attempts have included the principal structural reforms the previous section described — ranked-choice voting in some versions, automatic voter registration, vote-by-mail expansion, election-day-as-federal-holiday — but the legislative attempts have not produced enactment because the legislative coalition required to enact them has not assembled. The Democratic coalition’s filibuster carve-out for democracy reform, which the Manchin-Sinema position blocked in January 2022, was the operational mechanism that would have allowed the legislative attempts to succeed; the carve-out did not assemble; the legislative attempts did not succeed.

The Democratic coalition’s structural relationship to electoral reform is more complex than the platform-comparison summary suggests. The coalition has supported the reforms rhetorically and in legislative attempts; the coalition has, in operational terms, also benefited from specific features of the existing system that would be modified by the structural reforms the platform supports. The state-level Democratic coalitions in California, Illinois, New York, Maryland, and Massachusetts have, in the post-2010 redistricting cycles, drawn district maps whose partisan-bias metrics are comparable to the Republican-coalition gerrymanders in North Carolina, Pennsylvania, Wisconsin, and Texas.¹⁹ The Democratic coalition’s commitment to ending gerrymandering is, in operational terms, asymmetric: the coalition opposes Republican gerrymanders, which is consistent with the platform; the coalition has not consistently opposed Democratic gerrymanders, which is inconsistent with the platform’s structural commitments. The asymmetry reflects the structural reality that the Democratic coalition’s competitive position in some states depends on the existing gerrymandering pattern, and that the structural reform the platform supports would, in those states, reduce the Democratic coalition’s seat share even as it would increase the coalition’s seat share in the states where the Republican coalition currently benefits from comparable gerrymanders.

The Republican-coalition position on electoral reform has been structurally different and has been internally consistent across the post-1990 period. The Republican coalition’s framework on electoral rules has been, in operational terms, the maximization of the coalition’s competitive position through the existing structural features the previous sections described, combined with the introduction of additional procedural barriers (voter ID requirements, registration restrictions, early-voting reductions, vote-by-mail restrictions) whose empirical effect, on the political-science literature analyzing the post-2010 implementations, has been to reduce participation among the demographic categories that have historically supported the Democratic coalition.²⁰ The Republican-coalition framework’s empirical case for the procedural barriers — that the barriers prevent voter fraud — is substantially refuted by the cumulative empirical literature on actual voter fraud rates, which finds rates so low that the procedural barriers cannot be justified on fraud-prevention grounds without producing participation effects that substantially exceed the prevented-fraud rate.²¹ The framework’s actual structural intent, which the political-science literature has documented across numerous specific cases, is the participation suppression rather than the fraud prevention.

The convergent failure of the two parties has been the absence of any serious legislative framework for the structural reforms that the comparative record validates. Both parties’ competitive positions are protected by the existing system; both parties have, in operational terms, declined to support the structural reforms that would modify the system’s competitive features. The Democratic coalition’s support has been rhetorical and inconsistent; the Republican coalition’s opposition has been operational and consistent. The combined effect has been the preservation of the existing system across forty years of nominal reform discussion.

The Cascade That Closes

The electoral system described in the previous sections operates, considered in isolation, as the structural mechanism through which the two-party duopoly that the campaign-finance lock-in described in The Donor Class depends on is preserved.²² Considered against the broader fourth-settlement framework the previous essays in this series have described, the system becomes the structural reason the broader settlement has not been advanced.

The donor-class lock-in described in the campaign-finance essay operates through the financing of two specific political coalitions. The two coalitions exist because the electoral system produces two-party competition by Duverger’s law. The donor class’s structural influence depends on the small number of competing coalitions whose campaigns the donor class can effectively fund and whose policy positions the donor class can effectively shape; a system that produced multiple competing coalitions, each with smaller individual financing requirements and each with more diverse donor bases, would substantially attenuate the donor class’s structural influence. The two-party trap and the donor-class lock-in are, in this respect, the same structural feature viewed from different angles. The electoral system produces the two parties; the two parties accept the donor-class financing because the electoral system requires the financing scale that the donor class supplies; the donor class influences both parties because both parties depend on the same financing source. The structural reform of either component requires the structural reform of the other.

The fourth-settlement reforms described in the previous essays — the housing reform of The Quiet Foreclosure, the healthcare reform of The Accidental System, the education reform of The Credential and the Debt, the tax reform of The Code That Built the Donor Class, the climate reform of The Transition That Was Refused — are, in operational terms, blocked by the same two-party duopoly that the electoral system produces. The blocking operates through the donor class’s structural influence on both parties’ coalitions, on the framework the previous essay described. The structural alternative — the multi-party competition that the comparative democracies operate under, and that the structural electoral reforms would allow — would substantially modify the political-economy environment in which the fourth-settlement reforms would be considered. The reform of the electoral system is, in this respect, the structural prerequisite for the reform of the broader political-economy lock-in.

The broader fourth-settlement framework the series has sketched is, in this respect, structurally incomplete without the electoral-system reform. A reformed campaign-finance system that operated within the existing single-member-plurality framework would still produce the two-party competition the system structurally requires; the reformed campaign finance would reduce the donor class’s structural influence within each party but would not eliminate the structural feature that the two-party competition produces. A reformed electoral system — ranked-choice voting, multi-member districts, independent redistricting, automatic registration — would substantially modify the structural environment in which the campaign-finance reform would operate. The two reforms are complementary; neither is, in operational terms, sufficient on its own.

What’s at Stake

The Intelligent Party’s policy framework on democracy and elections, articulated in the platform’s democracy position, is the four-component proposal that addresses the structural features the previous sections diagnosed.²³

The first component is ranked-choice voting for all federal elections. The component is the structural reform that the previous section described as the contemporary policy with the most extensive American implementation record (Maine 2018, Alaska 2022, approximately fifty municipal jurisdictions, Australia for over a century). The component eliminates the strategic-voting dilemma the Duverger psychological effect produces and allows third-party candidates — including, structurally, candidates of The Intelligent Party — to compete on substantive policy positions rather than against the strategic-voting suppression that the existing system imposes.

The second component is independent redistricting commissions in every state. The component eliminates the partisan-gerrymandering mechanism that the post-2010 cycles have institutionalized. The component is structurally similar to the existing commissions in California, Arizona, Michigan, and Colorado; it would extend the model to the remaining states through federal legislation conditioning federal election-administration funding on commission adoption.

The third component is automatic voter registration at age eighteen, paired with same-day registration and universal vote-by-mail. The component addresses the participation-distortion mechanism the previous section described and brings the American voter-registration system into structural alignment with the comparator-democracy systems.

The fourth component is Election Day as a federal holiday, paired with the broader expansion of voting infrastructure and the federal protections for election workers that the post-2020 environment has demonstrated to be necessary. The component reduces the participation barriers that the contemporary American work schedule imposes and provides the institutional protection for the electoral apparatus that the post-2020 threats have demonstrated to be required.

The four components, taken together, address the structural problem the previous sections identified. The political coalition required to enact them does not currently exist. The conditions under which it might assemble are the conditions the broader fourth-settlement framework requires.

The comparative record is, in this case as in the others, instructive. Every other major democracy operates an electoral system that has substantially modified at least three of the four structural features the American system maintains. None of those countries has experienced the political-economy lock-in the contemporary American system has produced. The structural alternative is operating in every comparator democracy with which the United States routinely compares itself.

The realism the previous essays have called for, applied to democracy and elections, is the realism that the structural reform is achievable on the specific architecture The Intelligent Party has proposed; that the architecture has been designed to address the political feasibility constraints that have prevented the broader reform attempts; that the principal obstacle to its enactment is the same structural feature the reforms would address — the two-party duopoly’s interest in preserving the system that produces it. The conditions under which the obstacle breaks are the conditions under which the broader fourth-settlement framework simultaneously becomes available. The democracy reform, like the campaign-finance reform, is one component of a larger settlement that operates as the enabling condition for the substantive policy reforms the prior essays have described.

The essay about the American electoral system that someone writes in 2040 will describe either the construction of the four-component framework or its refusal. The constituency that would benefit from the framework is, structurally, the substantial majority of American voters whose policy preferences the contemporary system systematically under-represents. The constituency exists. Its political assembly has not yet occurred. The structural conditions for the assembly, on the trajectory the previous essays have described, will be in place within the next decade.

The two-party trap is the trap the system has constructed and that the system continues to enforce. The structural opening of the trap is the assembly the next decade will either produce or refuse. The realism is the realism the comparator democracies have demonstrated: the structural reform is achievable on the framework the comparative record has validated, against the political coalitions that the existing system has structurally constrained. The reconstruction is the assembly the broader fourth-settlement framework requires.

---

Notes