Two-Party Structural Countermandering


The 2010 redistricting cycle was a good decade for partisan mapmakers. Armed with ever-advancing “big data and modern technology,” they strategically manipulated electoral boundaries “with unprecedented efficiency and precision.”1 In 2012, the first election under the cycle’s new congressional maps, Democrats won 1.4 million more votes, but Republicans won the House by a seat margin of 234 to 201—the first time since 1996, and the second since World War II, that a party won a majority of seats in the House with a minority of votes.2 Several swing-state congressional maps produced lopsided results through the cycle: in Ohio, Republicans consistently won at least 12 of 16 seats, whether their vote share was 59 percent or 51 percent;3 in Pennsylvania, Republicans consistently won 13 of 18 seats, with vote share ranging between 49 percent and 55 percent;4 in North Carolina, Republicans won 9 or 10 of 13 seats with vote share ranging between 49 percent and 55 percent.5

But the most impressive cartographic feat was the 2011 Wisconsin Assembly map, adopted on a party-line vote shortly after Republicans won the trifecta (Governor’s mansion and both legislative chambers), and thus unilateral districting power, for the first time in four decades.6 As its architects intended, the map produced a decade-long Republican supermajority: in 2014, 2016, and 2020, Republicans won more than 60 percent of the seats with less than 55 percent of the votes; in 2012 and 2018, Republicans won more than 60 percent of the seats with less than half of the votes; even in the blue wave of 2018, a whopping 15-point vote swing from two years prior managed to flip just one seat, and Republicans still won 63 of 99 seats with only 47 percent of the votes.7 By strategically manipulating electoral boundaries, i.e., packing and cracking, the Wisconsin Assembly map twice inverted the vote-seat relationship, conferring a majority of seats on the party with a minority of votes. This vote-seat inversion is alternatively called minority entrenchment or a wrong-winner election.8 No matter what you call it, this perverse result violates a foundational premise of electoral democracy: if two parties compete for legislative seats, the party with more votes should win more seats.9

This perversity occurs with troubling frequency. Professor Miriam Seifter recently analyzed state legislative elections from 1968 to 2016, and found 267 cases where a party won control of a legislative chamber though its rival won more votes.10 Republicans drew the 2011 Wisconsin Assembly map, but Democrats can gerrymander with equal skill and zeal.11 Historically, both parties have benefitted when electoral maps have produced perverse outcomes. In Seifter’s study, those vote-seat inversions benefitted Republicans 111 times, but Democrats 156 times.12 There is evidence that Republicans gerrymandered more, more egregiously, and more successfully in the 2010 cycle.13 This may partly reflect greater Republican electoral success in 2010 state elections, which gave Republicans more control over the 2010 redistricting process.14 Also, residential patterns (i.e., urban clustering of Democratic voters) may make gerrymandering relatively easier for Republicans.15 But gerrymandering appeals to mapmakers across party lines.

And gerrymandering offends average Americans across the political spectrum. At a time when Americans seem to disagree on just about everything, a strong bipartisan consensus condemns gerrymandering as an affront and a threat to American democracy. We abhor the practice, and blame it for a host of democratic ills: vote-seat distortions, minority entrenchment, unresponsive legislatures, uncompetitive districts, uncontested races, low voter turnout and external efficacy, increasing polarization and gridlock.16 And we demand what I call countermandering—mechanisms to deter, limit, or mitigate gerrymandering and these associated pathologies. Hence the disappointment when partisan gerrymandering repeatedly fractured the Court, first in Bandemer in 1986, then in Vieth in 2004, and then again in LULAC in 2006.17 And hence the excitement when in fall 2016 a federal panel struck down Wisconsin’s Assembly map as a partisan gerrymander.18 It was the first time a partisan gerrymandering claim had ever succeeded in federal court—but not the last. Soon more federal courts found more partisan gerrymanders, and hopes swelled for a landmark countermandering ruling that would end the age of partisan gerrymandering just as the Warren Court had ended the age of malapportionment.

The Court dashed those hopes in the summer of 2019, with a 5-4 ruling declaring that partisan gerrymandering claims present a nonjusticiable political question.19 A supermajority of Americans favored federal judicial countermandering,20 but the Rucho majority declined the invitation. Writing for the majority, Chief Justice Roberts conceded that partisan gerrymandering is antidemocratic, but insisted that federal judicial countermandering is incompatible with the limited role of the federal courts.21 And so, the Supreme Court slammed shut the federal courthouse door it had left ajar 33 years prior.22

But Chief Justice Roberts assured Americans that their demands for countermandering need not “echo into a void.”23 Instead, he suggested other countermandering fora: state courts guided by state constitutional and statutory provisions with no federal analogue; and independent mapmaking bodies established by direct democracy or congressional mandate.24 Reformers are heeding this advice. Sixteen states now have constitutional or statutory provisions that prohibit partisan gerrymandering;25 and 15 states rely primarily on a commission to draw electoral maps for congressional and/or state legislative elections.26 Three states embraced these reforms in 2018;27 and one in 2020.28 The Census Bureau released new population figures in April, and the 2020 redistricting cycle is now kic­king into high gear, and the forces of gerrymandering and countermandering will compete once again, armed with another decade of technological advances.29

But I fear we underestimate what we’re up against, and overestimate the prospects for successful reform, because we fundamentally misapprehend the problem. We conceptualize gerrymandering, and design countermandering strategies, in terms that are narrowly performative rather than structural. The very term gerrymander forever casts Massachusetts Governor Elbridge Gerry as the personal embodiment of the hated practice,30 and frames the problem as abuse of the electoral system—dirty tricks and political sabotage by partisans with a zero-sum, scorched-earth approach to electoral democracy.31 This performative framing focuses on the partisan mapmaker rather than the underlying electoral system that invites gerrymandering and produces pathologies. The problem is how partisans discharge the mapmaking function, wielding the districting pen like a weapon.

And the only thing that stops a bad guy with a pen is a good guy with a pen. So the reform movement invests its efforts in two primary strategies: judicial countermandering, which seeks to empower courts with adequate legal standards to identify, invalidate, and remedy gerrymanders; and institutional countermandering, which seeks to transfer mapmaker power to an institutional body that will eschew gerrymandering and draw fair maps. Judicial countermandering disarms the partisan mapmaker, so she cannot wield the pen as a political weapon. Institutional countermandering gives the pen to an independent body designed to use it responsibly.

In this Article, I challenge this performative framing with a structural account that attributes present pathologies not just to partisan players, but to the game itself. Every state in the nation elects state and federal legislators using a common approach I refer to as the American Electoral Schema (“AES”): geographically partitioning the state into (usually single-seat) districts, and awarding each district seat to the individual candidate most preferred by the most district voters. (I use the word schema to describe a family of electoral systems that share a common bundle of identical or similar structural components that operate in combination to produce similar effects.) Under the structural logic of AES, a party’s success depends not only on its popularity, i.e., its statewide vote share, but also its efficiency, i.e., how it translates votes into seats. And efficiency depends critically on how the party’s supporters are distributed geographically across electoral districts. So an electoral map can interact with the underlying political geography to distort the overall vote-seat relationship.

This structural feature of AES is an open invitation to intentional partisan gerrymandering. For reasons I explain in Part I, judicial and institutional countermandering may constrain intentional partisan gerrymandering to some extent, perhaps avoiding the most egregious gerrymanders, but only at great cost, and they are unlikely to entirely extirpate partisan manipulation from the mapmaking process.

And precisely because judicial and institutional countermandering strategies focus on the performative gerrymandering of partisan mapmakers, there’s only so much they can do about unintentional gerrymandering. AES is structurally vulnerable to distorted vote-seat relationships, not only when partisan mapmakers wield the districting pen, but also when political geography conspires with ostensibly neutral districting criteria to produce unintentional gerrymandering. A politics-blind computer algorithm, guided by traditional criteria like contiguity, compactness, and preservation of local political boundaries, may unintentionally pack and crack, simply because Democrats cluster in cities, while Republicans disperse themselves in more sparsely populated rural areas.32

The problem is the game, not just the players. And the only viable solution is to fundamentally change the game through a strategy of structural countermandering, replacing AES with an alternative electoral system less vulnerable to these pathologies—a system that better serves contemporary American democracy.

But structural countermandering faces its own challenge: feasibility. It is not enough to identify some theoretical electoral system that works better on paper. Any serious effort at structural countermandering must address what Dean Gerken calls the “here to there” problem of electoral reform,33 and address “what ought to be the central question in election reform but is not: how to get change passed in this country.”34

 Most proposals for more proportional electoral systems fail the “here to there” test because they depart too radically from AES. Americans hate gerrymandering and the associated pathologies that AES produces, but they like the personal, geographic representation it facilitates: every person across the state has an individual representative who resides in, and is accountable to, a territorial community of neighbors who share common values and interests. And AES produces something else: two-party, legislative majoritarianism, where virtually every federal and state legislator is either a Republican or a Democrat, and each legislative chamber is generally controlled by a single-party majority. In stark contrast, systems of proportional representation tend to produce multiparty democracy, where significant seat share goes to three or more parties, who must form a multi-party governing coalition. The relative merits of these alternative systems are the subject of a robust academic debate which this Article makes no effort to resolve. But this Article takes seriously a two-party hypothesis: that party elites invested in the status quo will successfully defeat any proposal for an alternative electoral system that poses an immediate threat to the two-party system. At least for the near-term, American bipartyism may be entrenched, an exogenous constraint on any electoral system reform.

But we can eliminate gerrymandering while preserving a two-party system. This Article proposes an electoral schema to do precisely that. I call this proposed schema Mixed-Member Top-Two Proportional (“MM2”), because it operates much like the traditional Mixed-Member Proportional (“MMP”) system used successfully for decades in Germany and New Zealand, but replaces the broader form of proportionality used in MMP with a narrower form of two-party proportionality. Like MMP, MM2 preserves personal, geographic representation by selecting most legislators through single-seat districts; and it implements structural countermandering by allocating additional seats to political parties to compensate for any vote-seat distortion these districted elections produce. But whereas MMP allocates these seats to achieve vote-seat proportionality for every party, MM2 allocates these seats to achieve vote-seat proportionality only for the top two parties.

The Article proceeds in two parts. Part I presents the premise that we need two-party structural countermandering. It describes how our electoral schema, unlike many in Europe and elsewhere, inevitably invites gerrymandering and associated pathologies; it explains why we need structural rather than merely performative countermandering to combat these pathologies; and it argues that, as a matter of feasibility, such structural countermandering must preserve the two-party system.

Part II presents the proposal for reform, explaining how MM2 can achieve two-party structural countermandering. Using a concrete example based on the actual results of the 2018 Wisconsin Assembly, it demonstrates how even the simplest version of MM2 could have neutralized one of the decade’s most egregious partisan gerrymanders, eliminating the vote-seat inversion while promoting responsiveness, competitiveness, and voter turnout. It then delves into the details of MM2 and advances a specific version of MM2 characterized by several related design choices.

The Premise: Why We Need Two-Party Structural Countermandering

Why We Need Countermandering

This Section explains the need for countermandering. It describes the nearly uniform American approach for electing legislators at both the state and federal levels, and it contrasts this approach to one that would produce proportional representation. It then discusses three important consequences of the American system: personal, geographic representation, which is desirable; two-party legislative majoritarianism, which is normatively contested; and gerrymandering and associated anti-democratic harms, which are obviously pernicious, and which demand serious efforts at reform.

The American Electoral Schema

The states use remarkably similar methods to elect state and federal legislators. This common approach goes by several names, including First-Past-The-Post (“FPTP”), Winner-Takes-All, and Single-Member Plurality.35 In this Article, I will use the term American Electoral Schema (“AES”).

Consider a legislative body with  seats. AES combines the following elements:

  1. Geographic electoral districting. The schema geographically partitions the state into electoral districts, assigns seats to districts, and elects district representatives through district elections with district residency requirements for both candidates and voters.
  2. Low district magnitude. The number of representatives elected from a district is low, usually one. Political scientists often use the term district magnitude, denoted , to describe the number of representatives assigned to a single electoral district.36
  3. Under AES, and the electoral map simply consists of  single-member districts.
  4. Nominal choice. Voters choose people, not parties. Ballot notation may indicate party nomination or party affiliation, and voters may vote based on party, but the choice is formally structed in terms of individual candidates rather than political parties.
  5. Categorical choice. Voters select the single candidate they most prefer, rather than ranking all candidates. This makes voting easier for voters and administrators but limits the information a ballot provides about a voter’s overall preferences.
  6. Plurality election formula. The single-member district awards its seat to the candidate with more votes than any other single candidate, even if less than other candidates combined.37

Every state use AES when electing state and federal legislators, with limited intra-schema variation. In the first five decades following the founding, states used a variety of methods to elect representatives to the U.S. House,38 but Congress has required electoral districts since 1842 and single-seat electoral districts since 1967.39 Each state gets two U.S. senators with staggered terms, so the entire state serves as a single electoral district.40 Forty states rely exclusively on single-seat districts for each legislative chamber, and forty-eight states rely exclusively on single-seat districts at least for the upper chamber; of the ten states that use multi-seat districts for at least one legislative chambers, six use two-member districts, one uses three-member districts, and three use varying district magnitudes.41 Of the 7,383 seats in the 50 state legislatures, 6,301 (85.3 percent) are elected from single-seat districts.42 With two exceptions, every state uses categorical, nominal choice when electing federal and state legislators.43 Georgia and Louisiana hold a subsequent top-two run-off election if no candidate wins a majority in the general election, but every other state applies the plurality election formula to the general election for federal and state legislators.44 In sum, subject to limited intra-schema variation, AES has achieved monolithic use in contemporary American elections for federal and state legislators.45

A Stylized Foil Schema

Despite this striking uniformity, AES is not the only way to choose legislators. I discuss more alternatives in Section II.C, infra, but for now I briefly present a stylized schema of list proportional representation as a conceptual foil to AES. Here is a starkly different way a state could choose legislators:

  1. District magnitude . Treat the entire state as a single multi-member electoral district.
  2. No geographic electoral districting. With a single district, there are no electoral boundaries to draw.
  3. List choice. Voters choose parties, not people.
  4. Categorical choice. In the simplest, categorical-choice version, the voter selects the single party she most prefers.
  5. Proportional electoral formula. Award seats to parties in proportion to their vote share,46 using some specified method of seat allocation.47

Compare AES to list proportional representation: AES geographically partitions the state into multiple single-member districts while list proportional representation uses the entire state as one large multi-member district; AES asks voters to choose people, not parties, while list proportional representation asks voters to choose parties, not people; AES uses a plurality formula to award each district’s seat while list proportional representation uses a proportional formula to award all the seats in the legislative body.

Just as AES uses distinctive structural components, so too does it produce distinctive outcomes. The next three subsections explore three that will feature prominently in my argument: personal, geographic representation; two-party legislative majoritarianism; and the pathologies associated with gerrymandering.

Personal, Geographic Representation

One upside to AES is personal, geographic representation: every person across the state has an individual representative who resides in, and is accountable to, a territorial community united by shared values and interests. In contrast, list proportional representation has no local representatives because it does not use geographic electoral districting.

Political science and election law scholarship recognizes the value of personal, geographic representation.48 James Madison considered “[i]t . . . a sound and important principle that the representative ought to be acquainted with the interests and circumstances of his constituents.”49 This representative-constituent link remains significant today. The constituent can direct her grievances, policy preferences, and requests for government services to a single local representative, rather than some far-away party bureaucracy.50 The representative is well positioned to ascertain and advocate for the interests of her constituents.51 By channeling politics into “a place-based set of representational relationships,” personal, geographic representation “invites neighbors to engage one another in a debate about shared values, interests, and issues.”52

The value that voters accord to personal, geographic representation is suggested by the so-called personal vote, “that portion of a candidate’s electoral support which originates in his or her personal qualities, qualifications, activities, and record.”53 It is also suggested by the work of a district representative, which goes beyond the business of law-making and includes communicating with local constituents, providing casework services to individual constituents, and securing public resource allocations to her district.54

Two-Party Legislative Majoritarianism

A more contested feature of AES is two-party legislative majoritarianism.55 According to Duverger’s Law, named after the French sociologist Maurice Duverger, an electoral system like AES “favour[s] the two-party system”56 while a system like list “proportional representation favour[s] multi-partism.”57 The electoral system influences the party system through a mechanical effect and a psychological effect.58 The mechanical effect refers to the way AES translates votes into seats, and electoral parties (parties that run candidates) into legislative parties (parties that win seats). To win a seat under AES, a party needs a base of supporters sufficiently numerous and geographically compact to constitute a plurality in a single-seat district. The psychological effect refers to the way voters, candidates, and parties strategically modify their behavior to account for the mechanical effect. Under AES, third party candidates are often viewed as “spoilers”: if you vote for the third-party candidate you like the most, you may end up with the major party candidate you like the least.59

The United States does indeed have a strong two-party system, where virtually every federal and state legislator is either a Republican or a Democrat. This yields legislative majoritarianism, where the legislature is generally controlled by a single-party majority rather than a multi-party coalition. Every federal legislator is either a Republican or a Democratic, except for two U.S. senators, Bernie Sanders of Vermont and Angus King of Maine, who are independents but caucus with the Democrats.60 According to Ballotpedia, “[a]s of July 30, 2021, 32 state representatives in 10 states identify[] as independents or” members of a third party,61 whereas 5,366 state representatives identify as either Democrats or Republicans; “seven state senators in five states identify[] as independent or” members of a third party, whereas 1,957 senators identify as either Democrats or Republicans.62

In stark contrast, systems of proportional representation tend to produce multiparty democracy, where significant seat share goes to three or more parties, which must then form a multi-party governing coalition.63 Since 2017, the Dutch government has been a coalition of four parties.64 Eleven parties currently hold seats in the Israeli Knesset.65

There is no consensus on the desirability of this two-party, majoritarian system. Advocates of proportional, multiparty democracy make a strong case.66 But defenders of majoritarian, biparty democracy parry with a strong rejoinder, arguing that proportional, multiparty democracy may produce weak and unstable governing coalitions vulnerable to snap elections and extremist parties.67 I will return to this debate in Section II.C, infra.


There is one feature of AES that few would dare to defend: its vulnerability to gerrymandering and associated pathologies, including vote-seat distortions, minority entrenchment, unresponsive legislatures, uncompetitive districts, uncontested races, low voter turnout and external efficacy, and increasing polarization and gridlock. The fundamental cause of all these pathologies is not the partisan mapmaker, but rather AES itself, which facilitates both intentional and unintentional gerrymandering, and more generally ill serves contemporary American democracy.

Under AES, electoral fortunes turn discontinuously on whether a candidate’s support exceeds a threshold: With below-threshold support, the candidate loses and each vote she gets is lost; with above-threshold support, the candidate wins, but each vote she gets beyond the threshold is surplus; and half of all votes cast are lost or surplus, wasted in the sense that they have no impact on the outcome.68 For this reason, a party’s success depends not only on its popularity, i.e., its statewide vote share, but also its efficiency, i.e., how it translates votes into seats. And efficiency depends critically on how the party’s supporters are distributed geographically across electoral districts. So an electoral map can interact with the underlying political geography to distort the overall vote-seat relationship. By strategically manipulating electoral boundaries, i.e., packing and cracking, the partisan mapmaker can actually invert the vote-seat relationship, conferring a majority of seats on the party with a minority of votes. Another form of vote-seat inversion, which I call majority dominance, occurs when a party earns supermajority seat-share without supermajority vote-share.69

This power to invert the vote-seat relationship is awesome—in the biblical sense—and power corrupts. But AES is structurally vulnerable to distorted vote-seat relationships, not only when partisan mapmakers wield the districting pen, but also when political geography conspires with ostensibly neutral districting criteria to produce unintentional gerrymandering. A politics-blind computer algorithm, guided by traditional criteria like contiguity, compactness, and preservation of local political boundaries, may unintentionally pack and crack, simply because Democrats cluster in cities, while Republicans disperse themselves in more sparsely populated rural areas.

And vote-seat distortion is just the final product, the snapshot at the end of the electoral process. Just as important are the dynamics of the general election cycle, as incumbents, challengers, parties, donors, and voters make strategic choices based on anticipated electoral results. Some key aspects of these dynamics include: (1) vote-seat responsiveness;70 (2) district competitiveness;71 (3) uncontested races;72 (4) incumbency rate;73 (5) voter turnout;74 (6) external efficacy.75 These concepts all interrelate. Seat share is responsive to vote share if and only if there are competitive districts with narrow anticipated victory margins that could plausibly flip parties with enough of a vote swing. The least competitive district is obviously an uncontested district, and a district is more likely to be uncontested if prospective challengers think it is uncompetitive. Incumbency rates may be high for good reasons—an incumbent delivers for her constituents and they reward her with another term—or for reasons independent of the electoral map—for example, name recognition or campaign finance rules. But uncontested or uncompetitive district races necessarily give incumbency rates an artificial boost. While competitiveness is negatively correlated with incumbency rate, it is positively correlated with voter turnout and external efficacy, as rational choice theory predicts and the empirical literature confirms.76

The 2018 Wisconsin Assembly election is most infamous for the vote-seat inversion it produced—Republicans won 63 of 99 seats with 47 percent of the vote. But the map’s pathologies extend beyond this ultimate result to the dynamics that produced it. The 2018 election was a blue wave, featuring a 7.5 percent vote swing from Republicans to Democrats compared to 2016.77 Republicans won the 2016 vote share by 9 points, but lost the 2018 vote share by 7 points.78 This represents a decisive repudiation of the party in power. But that repudiation had virtually no effect, because it managed to flip only one out of 99 seats.79 The Republican supermajority in the Wisconsin Assembly plummeted from 64 out of 99 seats to 63 out of 99 seats. Consider this from the perspective of the party in power: Win the popular vote by 9 points, and you get a filibuster proof supermajority in the legislature, but lose the popular vote by 7 points . . . and you get a filibuster proof supermajority in the legislature. This non-responsiveness subverts elementary principles of democratic theory.

This non-responsiveness was associated with uncompetitive districts. In the 2018 Assembly election, the average margin of victory in a district race was 28.1 percent.80 Of the 99 district races, only 11 had a margin of victory under 10 percent.81 The other 88 races weren’t even close.82 Indeed, in 33 races (precisely one third of all races) the winner ran unopposed, because nobody else even tried to compete.83 Under AES, your vote really matters if and only if it helps your preferred candidate win your district. Otherwise, your vote is wasted. So if the outcome of the district race is a foregone conclusion—and it often is—why bother voting? Or running? Safe districts yield a vicious circle of voter apathy and party atrophy because the only way to achieve statewide representation under AES is to win a district race.84

While the Wisconsin Assembly was impervious to the blue wave of 2018, Democratic candidates won elections for governor and attorney general. But consider how the Wisconsin legislature responded behind the electoral shield of partisan gerrymandering.

[T]he outgoing Republican-controlled legislature enacted, and the lame-duck Republican governor signed, legislation that deprived the governor of control over significant public programs and transferred from the attorney general to the legislature authority to withdraw from various kinds of litigation. The legislation was apparently intended in great measure to cripple the ability of the incoming Democrats to deliver on their campaign pledges—the very basis, presumably, of their election to office—thereby subverting the only reason to hold elections in the first place.85

AES is vulnerable not just to intentional partisan gerrymandering, but to a set of pathologies, associated with both intentional and unintentional gerrymandering, so severe that countermandering is imperative.

Why We Need Structural Countermandering

The pathologies just described are well known, but they are often attributed specifically to intentional partisan gerrymandering rather than to AES itself. Under this performative framing, it is implicitly assumed that these pathologies could be adequately constrained if only electoral boundaries were drawn by the right mapmaker or subject to judicial scrutiny under the right legal standard. In this section, I challenge this theory of constraint. I argue that structural countermandering is the only viable strategy because of the inherent limits of judicial and institutional countermandering under AES.

The U.S. Supreme Court recently dealt a significant blow to the judicial countermandering strategy when it declared that partisan gerrymandering claims present a non-justiciable political question that federal courts cannot adjudicate.86 I think Rucho was wrongly decided, and I have argued elsewhere that it should be read narrowly to foreclose only “allocative” claims of partisan gerrymandering that frame liability and remedy in terms of the vote-seat relationship, leaving open “non-allocative” claims like those predicated on an electoral map’s unfair geographic impact.87 But the Court is unlikely to revisit or distinguish Rucho in the near-term, so the federal courts cannot constrain partisan gerrymandering directly.88 And even if the Rucho majority was mistaken about the authority and competence of the federal courts to adjudicate claims of partisan gerrymandering, it identified legitimate prudential concerns implicated whenever any court rules on such claims.89

While Rucho closed the federal courthouse door, the state courts remain open, and some have more specific state constitutional provisions to guide adjudication of partisan gerrymandering claims.90 But even if state judges have more to work with textually, they face similar challenges prudentially and doctrinally. Assuming that such claims are framed in terms of vote-seat relationships, they still ask state courts to reallocate power between the two political parties. This raises the concerns identified by the Rucho majority: Embroiling judges in high-stakes political contests and perceptions of judicial bias.91 And state judges may not have the life tenure, salary protections, and insulation from political pressures that federal judges enjoy.

Litigation is slow and expensive. Gerrymandering litigation is always a moving target because electoral maps have limited shelf life. The state must draw a new map every ten years to ensure population equality based on the most recent census data.92 With elections every two years, a map is good for five elections—at most, because states may engage in mid-cycle redistricting, a practice the Supreme Court has explicitly approved.93

Perhaps redistricting commissions can eschew what AES invites, if states can establish these bodies and endow them with the requisite independence, competence, resources, composition, procedures, and substantive criteria. But partisans don’t pass the pen willingly. They resist. They try to keep the redistricting initiative off the ballot, and if that fails, they mobilize against its passage.94 If it passes, they try to obstruct, or coopt, the commission, through appointment and removal of commissioners, or by starving the commission of resources, or through lawsuits that challenge the commission or its maps.95 And then they try to get another initiative on the ballot, this one repealing, or capturing, the commission.96

And if redistricting commissions are to keep drawing congressional maps, the Court itself must eschew a strong independent state legislature theory of the Elections Clause, which provides that “[t]he Times, Places, and Manner of . . . [federal] [e]lections . . . shall be prescribed in each [s]tate by the [l]egislature thereof.”97 When Arizonans first established an independent redistricting commission, the Arizona state legislature insisted that the Elections Clause grants it alone the power to draw congressional maps. The Court ultimately rejected that argument, but in a 5-4 decision with a strident dissent authored by Chief Justice Roberts, who accused the majority of “gerrymander[ing] the Constitution” with an atextual reading of the Elections Clause.98 Now that Justices Kavanaugh and Barrett have replaced Justices Kennedy and Ginsberg, some fear (or hope) that the Court will reverse that decision and return the power to draw congressional maps to state legislatures.99

Judicial and institutional countermandering may constrain intentional partisan gerrymandering to some extent, perhaps avoiding the most egregious gerrymanders, but only at great cost. And even if they could entirely extirpate partisan manipulation from the mapmaking process, they cannot end unintentional gerrymandering. AES is structurally vulnerable to distorted vote-seat relationships, not only when partisan mapmakers wield the districting pen, but also when political geography conspires with ostensibly neutral districting criteria to produce unintentional gerrymandering. Massachusetts has not sent a single Republican to the House of Representatives since 1994, even though Republican candidates regularly win about 30 percent of votes cast in the nine Massachusetts congressional districts. A recent analysis concluded:

[T]he underperformance of Republicans in Massachusetts is not attributable to gerrymandering, nor to the failure of Republicans to field House candidates, but is a structural mathematical feature of the actual distribution of votes observable . . . . Republican votes clear 30%, but are distributed so uniformly that they are locked out of the possibility of representation. Though there are more ways of building a valid districting plan than there are particles in the galaxy, every single one of them would produce a 9–0 Democratic delegation.100

As Justice Scalia put it in the Vieth plurality opinion:

[W]e would find it impossible to ensure that [the majority] party wins a majority of seats—unless we radically revise the States’ traditional structure for elections. In any winner-take-all district system, there can be no guarantee, no matter how the district lines are drawn, that a majority of party votes statewide will produce a majority of seats for that party.101 

Recall that Seifter recently found 267 cases between 1968 and 2016 where a party captured a state legislative chamber with minority vote share.102 Performative countermandering could have prevented each of these reversals only if intentional partisan gerrymandering producer each one, and only if performative countermandering could have prevented each such instance of intentional partisan gerrymandering.

More generally, AES itself produces the pathologies that a performative theory attributes to bad actors. Intentional gerrymandering may exacerbate them, but they can happen even without intentional gerrymandering. There is robust debate about what role intentional gerrymandering plays exactly, but there are strong indications that these pathologies are at least partially attributable to the system itself.103

Even if judicial or institutional countermandering succeeds in preventing both intentional and unintentional vote-seat distortions, the results may be a bipartisan gerrymander that carves up the state into the desired number of safe red and safe blue districts, sacrificing principles of geographic representation to produce partisan fairness. Such bipartisan gerrymanders diminish responsiveness by creating many uncompetitive districts, and thereby producing many races that are uncontested, only lightly contested, or otherwise a foregone conclusion. And when the electoral map pre-ordains the outcome, participation in the general election may seem pointless, the vain ritual of a faux democracy. The result is low voter turnout and reduced external efficacy. The action naturally shifts from the general election to the primary election, so the ideology of the district representative shifts from that of the median general election voter to that of the median primary election voter.104 The result may be a legislature more ideologically polarized than the electorate.

Performative countermandering could affirmatively demand fair partisan outcomes, rather than just the absence of discriminatory partisan intent.105 But to achieve partisan fairness, mapmakers may have to subordinate other important districting goals, like capturing communities of interest or drawing competitive districts. AES itself produces the pathologies often attributed exclusively to intentional partisan gerrymandering. If we can overcome all the barriers to successful judicial and institutional countermandering, we may curb the worst vote-seat distortions, but only at great cost, and the other pathologies will persist, if not worsen.

Why We Need Two-Party Structural Countermandering

The prior Section argued that structural countermandering presents the only viable response to gerrymandering because AES will confound any non-structural countermandering strategy. This section argues that the only viable way to achieve structural countermandering is to preserve the two-party system, because the two-party system will resist any reform that threatens it.

Note that I make a predictive claim about the two-party system’s entrenchment, not a normative claim about its superiority. My personal view is that there are strong arguments for and against the two-party system, and the question merits vigorous debate.106 But in this Article, I make no attempt to resolve this debate. I only argue that any reform effort that threatens the two-party system will face steep, and possibly insurmountable, resistance. One may not like the two-party system, and for good reason. But the two-party system likes the two-party system quite a bit. It is hard to take on the two-party system, especially when there is no consensus against it. Serious efforts to address gerrymandering must take this reality into account.

For example, the Dutch list proportional representation system is logically immune from gerrymandering because there are no electoral boundaries to manipulate. Theoretically, the United States could eliminate gerrymandering entirely by simply replacing the prevailing electoral system with the Dutch system, much like the Dutch did themselves after World War I. But as a practical matter, this is a non-starter. This intuition can be translated into a more general hypothesis: the two-party system will defeat any proposal for an alternative electoral schema that immediately threatens the two-party system. This hypothesis consists of a few related claims:

  1. Most proposals to reform AES would shift the United States from a two-party system with legislative majoritarianism to a multiparty system with inter-party governing coalitions, or at least people so predict; and
  2. Whatever its benefits, multiparty governing coalitions would involve drawbacks—fringe parties, extremists, unstable coalitions, snap elections, gridlock, lack of accountability, and so forth, or at least people so predict; and
  3. Given (1) and (2), party elites empowered by the status quo will perceive these proposals as a threat to the status quo that empowers them, no matter what’s best for the polity, or what effects these proposals would actually have; and
  4. Given (1) and (2), party elites can effectively attack and successfully defeat these proposals.

One alternative schema that has garnered considerable attention and support is one based on multi-member districts, with three to five representatives per district, and seat allocation based on a single-transferrable vote. This is a multi-seat version of the sort of ranked-choice voting (“RCV”) approach that has enjoyed so much success recently in the single-seat context. The national advocacy group FairVote has endorsed this approach. Representative Don Beyer (D-VA) introduced a bill, called the Fair Representation Act (“FRA”), to adopt this system for congressional elections.107 Recently, legal scholar Paul Diller has proposed mini-FRAs for state legislatures.108

Another option is the Open List Proportional Representation variation endorsed by political scientists Jack Santucci and Matthew Shugart for the U.S. House.109 Like the FRA, Open List Proportional Representation would use multi-member districts with three to five members per district. But voters would choose a single candidate, rather than ranking all candidates. That vote would be treated as a vote both for the candidate and for her party. Seats would be allocated to parties based on some allocation formula to achieve proportionality. Each party’s seats would then go to the most popular candidates affiliated with that party.110 Santucci and Shugart favor Open List Proportional Representation because it is “minimally disruptive” in an administrative sense—just like they now do under AES, each voter would simply pick her favorite candidate and election officials would simply count up the votes and report the totals.111

Another option is the MMP system used successfully for decades in Germany and New Zealand.112 Each country implements its own variation, but the essential features are the same: the legislative body is divided into two tiers: a nominal (i.e., person) tier and a list (i.e., party) tier. Each voter gets two votes, one for each tier. An electoral map partitions the state into single-seat districts for the nominal tier, and each district seat goes to the individual candidate most popular with her district’s voters. But MMP allocates the list-tier seats statewide to political parties to achieve overall proportionality between the vote share and seat share of each party. MMP is “mixed-member” because it uses different election methods for different members of the legislative body. MMP is “proportional” because list seat allocation compensates for nominal seat allocation to produce overall proportionality for each party. This system retains the personal, geographic representation of the U.S. system but avoids the problem of gerrymandering and associated pathologies. Its success in Germany and New Zealand has understandably impressed academics and reformers worldwide.113 Despite its popularity elsewhere, MMP has received very little attention in the United States.

The FPA, Open List Proportional Representation, and MMP are all serious proposals that offer compelling advantages over AES. But each one would shift the United States from a two-party system with legislative majoritarianism (i.e., a single party controls a majority of seats in the legislative body) to a multiparty system with inter-party governing coalitions, or at least people so predict.

FairVote published a report assessing the likely impact of the FPA.114 This report noted that increasing district magnitude decreases the threshold vote share needed to win a seat: “the threshold would be just over a third of the vote in two-winner districts, just over a fourth of the vote in three-winner districts, just over a fifth of the vote in four-winner districts, and just over a sixth of the vote in five-winner districts.”115 FairVote predicted that this lower threshold would help third party candidates win seats.

[I]n nearly all districts, any faction of voters that make up more than 17 or 25% of the electorate will be able to elect a candidate of their choosing. As less doctrinaire voters are as numerous as partisan Democrats and Republicans in the general election electorate, ranked choice voting would allow them to come together to elect more representatives able to act as bridge builders in districts across the country, and open the door for independent and third party candidates that may better represent their views.116

Some favor the FPA in part precisely because they predict it will foster multiparty democracy.117 FairVote asked 14 political scientists to assess 37 structural reforms based on 16 different criteria.118 Two of the highest-rated reforms, called RCV-3 and RCV-5, were essentially variations of the FPA with different district magnitudes. FairVote reported, “a consensus that using RCV in three- and five- seat districts would provide serious potential for more third parties . . . to hold traditional major party candidates accountable and potentially win seats.”119 Open List Proportional Representation similarly threatens the two-party system, because it also increases district magnitude and thus decreases the vote share threshold needed to win a seat.120 MMP also promotes multiparty democracy: in New Zealand, five parties win seats; in Germany, six parties win seats.121 To be sure, some think these proposals would have little impact on the two-party system.122 But the question is what party elites will fear, not whether those fears are justified. Even if their fears are overblown, party elites are likely to perceive proposals such as these as threats to the two-party system.

Threatened by these proposals, party elites will likely mobilize against them. They will follow a familiar script, warning voters that any step towards a more proportional multiparty democracy is a slippery slope to the worst possible political dysfunction.123 The script will conflate different electoral systems and offer a simplistic institutional determinism, blaming proportional representation for: the fall of the Weimar Republic and the rise of the Nazi party;124 Ukraine’s failure to stop Crimea’s annexation;125 Israel’s recent parliamentary election, its fourth snap election in two years;126 and other cautionary tales of political dysfunction.127 These attacks may resonate in a political environment that has long evinced a deep hostility to proportional representation.128

For these reasons, I fear these proposals fail the “here to there” test. AES produces intolerable pathologies, but any proposal for an alternative schema will fail if it immediately threatens the two-party system. This dilemma begs the question: is there an electoral schema that avoids gerrymandering and its associated pathologies, but retains both personal, geographic representation and biparty legislative majoritarianism? The next Part proposes an electoral schema designed to do just that.

The Proposal: How to Achieve Two-Party
Structural Countermandering

In this Part, I develop a proposal for an electoral schema that addresses gerrymandering structurally within the confines of the two-party system. I call this proposed schema Mixed-Member Top-Two Proportional (“MM2”), because it operates much like the traditional Mixed-Member Proportional (“MMP”) described above129 but replaces the broader form of proportionality used in MMP with a narrower form of two-party proportionality.

Like MMP, MM2 preserves personal, geographic representation by selecting most legislators through single-seat districts and it implements structural countermandering by allocating additional seats to political parties to compensate for any vote-seat distortion these districted elections produce. But whereas MMP allocates these seats to achieve vote-seat proportionality for every party, MM2 allocates these seats to achieve vote-seat proportionality only for the top two parties.

For example, suppose parties A, B, and C respectively win 48 percent, 46 percent, and 6 percent of the statewide popular vote in all single-seat district elections. Under the current schema, given partisan gerrymandering or other factors, party B might win a majority of these districts even though party A candidates earned more votes. Under traditional MMP, additional seats would be allocated to the political parties so that parties A, B, and C respectively win roughly 48 percent, 46 percent, and 6 percent of overall seats. By producing vote-seat proportionality for each party, traditional MMP countermands the effect of partisan gerrymandering, but it produces a three-party legislature with no legislative majority. Under MM2, additional seats would be allocated to the political parties so that parties A and B respectively win seat shares roughly proportional to their relative two-party vote shares. So party A seat share would be  and party B seat share would be. By producing vote-seat proportionality only for the top two parties, MM2 also countermands the effect of partisan gerrymandering, but it produces a two-party legislature with a single-party legislative majority.

Just like AES, MM2 is not a single electoral system but rather an electoral schema, a family of electoral systems that share common motivations, structural components, and effects. The common structural components include: dividing the legislative body into two classes, a nominal class and a list class; partitioning the state into electoral districts, assigning nominal seats to districts, and awarding those seats to the candidates district voters prefer to represent them; and the most important component of all, a structural countermandering principle that automatically allocates list seats to the top two political parties based on their relative popularity.

But there is more than one way to do MM2, and implementation entails important design questions. These questions include: the overall size of the legislative body and the relative size of the two classes; whether the ballot should ask voters only which candidates they prefer to represent their districts, or also which political parties they prefer overall; for each question, whether the voter should select their top choice or rank all the options; and how precisely election administrators should aggregate voter responses to determine election results. Section B canvasses these design questions in detail.

But before Section B’s deep dive into design details, Section A presents the case for MM2 using a simplified version to demonstrate its operation and illustrate its primary benefits.

The Concept

The Proposal in Brief

To see how MM2 works, let’s consider a real-world example. The Wisconsin Assembly has 99 seats, each assigned to a single-seat district. After gaining power in 2010 and passing an egregious partisan gerrymander,130 Republicans consistently earned disproportionate seat share throughout the 2010 redistricting cycle, twice securing majority seat share with minority vote share:

Table 1:131 Wisconsin State Assembly Election Results

Vote Share Versus Seat Share Under the Post-2010 Map 

The most extreme election occurred in 2018, when Republicans won 63 of 99 seats with just 47 percent of the votes, while Democrats won just 36 of 99 seats with 53 percent of the votes. 

Let’s consider a simplified analysis of what would have happened in 2018 with MM2. This analysis is simplified in two respects. First, I will use the most stripped-down version of MM2, where each voters gets one vote, and the countermandering principle is implemented in the simplest way. Second, for the purpose of this example, I use the real election data from 2018. Of course, voters, candidates, parties, and other electoral actors (like mapmakers) behave strategically based on the rules of the game; if we change the rules, everyone will change their behavior accordingly. So if Wisconsin had actually adopted MM2 shortly before the 2018 election, we would not expect the same election results. Candidates may have made different decisions about whether and how to run. Voters may have made different decisions about whether and how to vote. Below, I will discuss these possible behavioral changes, and argue that MM2 would promote desirable behavior.132 But for the present example I necessarily assume away all these dynamic effects and simply work with real election data from 2018.

With these caveats in mind, consider a counterfactual 2018 Wisconsin Assembly election under MM2. In the real 2018 election, the Wisconsin Assembly has 99 seats and the Wisconsin Assembly map consisted of 99 single-seat districts. Suppose Wisconsin had implemented MM2 by keeping those 99 seats as nominal seats but adding 40 list seats for a total of 139 seats overall. (Note that an odd number of seats facilitates legislative majoritarianism by excluding the possibility of an even partisan split.) We now have a legislative body with two classes—one nominal, one list.

Since a primary objective of MM2 is to preserve and enhance personal, geographic representation, the nominal class would operate much like the prevailing approach today. A mapmaker partitions the state into single-seat districts guided by geographic criteria designed to establish meaningful place-based relationships between each district representative and the geographic community she represents. Each district chooses its representative through a traditional election with district residency requirements for voters and candidates. As we shall see, countermandering can neutralize gerrymandering independent of the electoral map, so MM2 does not specify who draws the map or how they draw it.133 To keep things simple, let us suppose that in our counterfactual 2018 election, Wisconsin used the same Assembly map actually used in the real 2018 election. And let us assume that all actors behave in this counterfactual 2018 election just as they actually did in the real 2018 election. This means we get precisely the same lopsided results in the districted elections: Republicans win 63 of 99 nominal seats with just 47 percent of the votes; Democrats win just 36 of 99 nominal seats with 53 percent of the votes. Partisan gerrymandering produces severe vote-seat distortion in the nominal class.

After election officials determine the results of all the single-seat district elections, the countermandering provision kicks in and automatically allocates list seats to achieve overall biproportionality. Here, the system would automatically neutralize the vote-seat distortion in the nominal class through compensatory allocation of the 40 list seats, awarding 37 list seats to the Democratic party and the remaining 3 list seats to the Republican party. This compensatory allocation would result in 73 seats for the Democrats (36 nominal plus 37 list) and 66 seats for the Republicans (63 nominal plus 3 list). Overall, Democrats would have earned 52.5 percent seat share (73 out of 139 seats) with 53 percent vote share, while Republicans would have earned 47.5 percent seat share (66 out of 139 seats) with 47 percent. Thus, with a list class less than half the size of the nominal class, MM2 can achieve near-perfect overall vote-seat bi-proportionality even with one of the most egregious partisan gerrymanders of the 2010 redistricting cycle.

Table 2: Counterfactual 2018 Wisconsin Assembly Election with MM2



Can a state replace AES with MM2? Absolutely. MM2 is the natural response to Rucho, and the proposal’s legality is the logical implication of Rucho. The decision is sometimes characterized as an endorsement of the status quo, of partisan gerrymandering, and of AES. But, on its own terms, Rucho endorses state autonomy and judicial restraint, not partisan gerrymandering or the prevailing electoral schema. The Rucho majority acknowledged, “[e]xcessive partisanship in districting leads to results that reasonably seem unjust. . . . [S]uch gerrymandering is ‘incompatible with democratic principles.’”134 More generally, a central premise of the Rucho decision is that the choice of an electoral system, the definition of political fairness, and the determination of a normative baseline for seat allocation presents “questions that are political, not legal.”135

There are no legal standards discernible in the Constitution for making such judgments, let alone limited and precise standards that are clear, manageable, and politically neutral. Any judicial decision on what is “fair” in this context would be an “unmoored determination” of the sort characteristic of a political question beyond the competence of the federal courts.136

Indeed, the U.S. Constitution neither mandates nor proscribes any electoral schema. For congressional elections, the Election Clause, Article I, Section 4, leaves it to each State to prescribe their “Time, Place, and Manner,” subject to congressional regulation.137 The Constitution says nothing about state elections, except that “[t]he United States shall guarantee to every [s]tate in this Union a Republican Form of Government.”138 According to Rucho, with this limited textual guidance, the federal courts cannot determine fairness and proscribe partisan gerrymandering. But states can. “Our conclusion does not condone excessive partisan gerrymandering. Nor does our conclusion condemn complaints about districting to echo into a void. The States, for example, are actively addressing the issue on a number of fronts.”139

And states are not limited to judicial or institutional countermandering. Unlike the federal courts, the states can engage in structural countermandering, mandate biproportionality, establish the “norm” that the top two parties “should enjoy [power] commensurate” with their popularity, and adopt a new “electoral system” based on this norm.140

Of course, all state action is constrained by the First and Fourteenth Amendments, but MM2 accords symmetric treatment to the top two parties. Indeed, MM2 embodies the principle of partisan symmetry that featured so prominently in efforts to establish a federal constitutional standard to adjudicate partisan gerrymandering.141 MM2 accords parties disparate treatment based on their popularity, but this disparate treatment is justified by the state’s legitimate interest in preserving bipartyism and legislative majoritarianism. American courts have repeatedly sustained electoral laws that accord disparate treatment based on a party’s popularity, such as ballot access rules that distinguish between major and minor parties, as permissible efforts to promote an electoral system that is stable, efficient, and administrable.142

Finally, adoption of MM2 would not violate the Voting Rights Act, even though it may affect the number of nominal seats or the size of electoral districts. The Court has held that the size of a governmental body is not subject to a claim of racial vote dilution under Section 2 of the Voting Rights Act because “there is no objective and workable standard for choosing a reasonable benchmark.”143


Structural Countermandering

The Wisconsin Assembly map used in 2018 is widely considered an extreme partisan gerrymander, an outlier across states and over time. Litigants spent years in federal court trying to undo it, and successfully persuaded a three-judge panel of its unconstitutionality, but the Supreme Court in Rucho foreclosed federal judicial intervention. MM2 could have undone this partisan gerrymander in an instant, without establishing a commission or instituting litigation, not by changing the map, but by countermandering any distortion it produces. Note that the allocation of list seats would happen automatically according to a simple formula codified in state law. The “countermandering” provision would say something like: “The list seats shall be allocated to the top two parties so as to produce the closest possible correspondence between the statewide vote share and overall seat share of each of the top two parties.” No person or body would exercise discretionary judgment ex post regarding the appropriate allocation of list seats. The state’s chief election officer would simply apply the provision after all ballot tallies had been certified. And everyone would know ex ante how the provision will operate.

This system renders partisan gerrymandering structurally impossible. As long as there are enough list seats to countermand any seats-votes distortion in the nominal seats, the system will produce the desired relationship between vote share and overall seat share. The countermanding effect occurs automatically by the ex-ante rules of seat allocation. It requires no litigation and no legal standard for partisan gerrymandering. It does not depend on the good faith of the mapmaker. The mapmaker can try as she pleases: draw the most grotesque districts, pack and crack to your heart’s content. You may succeed in distorting the seats-votes relationship for the nominal class. But then the system will allocate list seats accordingly. And the result will be a legislative chamber with a partisan composition that reflects the top two parties’ overall popularity in the electorate. The system is simply impervious to gerrymandering. Any attempt to gerrymander would obviously be an exercise in futility.

While the system stops the sort of intentional gerrymandering courts could potentially address, it also countermands the sort of unintentional gerrymandering beyond the reach of any intent-based legal standard. Democrats may cluster in cities, while Republicans spread out in more sparsely populated rural areas.144 The mapmaker may apply geographic districting criteria that tend to produce seats-votes distortions, like compactness or preservation of local political subdivisions.145 If these conditions combine to favor one party in the allocation of nominal seats, the system will automatically favor the other party in the allocation of list seats to achieve overall two-party fairness in the partisan composition of the legislative chamber.

The system may do more than eliminate the possibility of intentional or unintentional gerrymandering. It may deter intentional gerrymandering. As discussed supra, a partisan gerrymander can favor one party in the allocation of nominal seats, but the system will automatically allocate list seats accordingly to produce a fair overall allocation. For this reason, the partisan mapmaker cannot use gerrymandering techniques to help her party overall. As an initial matter, gerrymandering is politically unpopular and entails some reputational costs. Partisans gerrymander despite these costs because the political benefits outweigh them. But if gerrymandering has no benefits, why entail the reputational costs? Moreover, such techniques may actually hurt her party overall if a list seat is more desirable than a nominal seat. By favoring her party in the allocation of nominal seats, the partisan mapmaker’s gerrymandering techniques necessarily disfavor her party in the allocation of list seats. Gerrymandering does not win your party more seats. It just swaps your list seats for nominal seats.

There are good reasons to think that a party would prefer a list seat to a nominal one. By design, a nominal legislator cannot focus exclusively on furthering her party’s legislative agenda. She is accountable to a geographic constituency whose interests may sometimes be opposed to, or orthogonal to, her party’s legislative agenda. She must perform constituent services. She must dedicate resources (including time and focus) to ascertaining local needs and representing those needs in the capital. In contrast, a list legislator can focus exclusively on her party’s legislative agenda. Her mandate comes from the statewide electorate, not a geographic community with particular place-based needs. She does not have to consider how her work impacts her district because she has no district. She has a party platform she was elected to enact, not a local constituency she was elected to represent. For similar reasons, we may expect list legislators to assume leadership roles in the legislative assembly, such as the speaker and majority whip. Given the reputational costs of gerrymandering and the likelihood that list legislators are better situated to lead the party and further its agenda, a partisan mapmaker under this new system has strong incentives to avoid partisan gerrymandering.

Enhancing Geographic Representation

MM2 not only preserves, but actually enhances, personal geographic representation—for at least two reasons, and possibly three. First, without the temptation of partisan gerrymandering, the mapmaker may draw a map that makes more sense geographically and thus better aligns with principles of personal, geographic representation. Second, MM2 enables a division of labor between nominal and list legislators. List legislators, with no local constituency, can focus more on policy, committee work, legislative oversight, gubernatorial nominations, and so forth. Nominal legislators can focus more on the local aspect of their work—constituent services and representing local interests. Third, in the two-vote version of MM2, each voter answers two questions: (1) which candidate do you prefer to represent your district; and (2) which party do you prefer overall? This ballot structure facilitates ticket splitting by class, i.e., a voter can support one party overall, but support an individual candidate from another party. This means that an individual candidate can win support from voters who might prefer the other party overall. “You may not agree with me on every issue, but you trust me to represent our community, because I understand local needs and values, and I deliver for my constituents.” This dynamic may further enhance the benefits associated with personal, geographic representation.

General Election Dynamics

AES aggregates ballots for one purpose: to choose each district’s representative. Under this schema, the only way to influence the overall partisan composition of the legislative body is to flip a seat in a competitive district. But most voters live in an uncompetitive district, if the race is contested at all. And everyone knows this when they decide whether to vote, to donate to a campaign, to run for office, or how to allocate a political party’s scare resources.

MM2 alters this logic. No matter how it’s implemented, MM2 aggregates ballots for two distinct purposes: to choose each district’s representative and to allocate list seats. Even if a voter’s preferred candidate loses the district election, her ballot is accorded equal weight when determining the relative popularity of the top two parties. If you vote for the Republican, you meaningfully support the Republican party even if a Democrat wins the district. A one-point swing in statewide party support produces a one-point swing in overall seat share, and a 15-point vote swing produces a 15-point seat swing—even if the state is carved up into blue and red districts so safe that a not a single nominal seat flips. This gives a powerful incentive to vote, and to run for district representative, even in the safest district.

The incentive is even stronger when voters are given two votes, one for a representative, one for a party. In the two-vote versions of MM2, a voter can support a party even if she prefers not to support that party’s candidate in the district race, or even if that party has no candidate running in the district. And the incentive is strongest when voters can rank their preferences over candidates and parties. In the two-vote, ranked preferences version of MM2, even voters who support third parties, third party candidates, or independent candidates can also express a preference between the top two parties, and that preference will be accorded equal weight in determining the overall partisan composition of the legislative body.

This altered logic may turn a vicious cycle into a virtuous one, synergistically promoting vote-seat responsiveness and district competitiveness, voter turnout and external efficacy, party mobilization and donor investment, strong challengers and robust campaigns. MM2 can promote all these desirable effects even with a highly uncompetitive map.

But MM2 may also promote a more competitive map. As discussed in Section III.A.3.i, structural countermandering renders partisan gerrymandering futile, if not affirmatively counterproductive. While it eliminates the incentives for partisan gerrymandering, it also obviates the need for bipartisan gerrymandering. And while bipartisan gerrymandering can only achieve partisan fairness through safe districts, MM2 can achieve partisan fairness with competitive districts.

Imagine a perfect swing state where each party enjoys half the vote share. Suppose the mapmaker draws perfectly competitive districts such that each district is predicted to result in a tie. Now suppose that one voter in each district switches their party allegiance from the Democrats to the Republicans. The result would be a Republican sweep, where the Republican candidate wins every district by a single vote. With slightly more than 50 percent vote share, the Republican party would win all the seats. This extreme hypothetical reflects a more general trend: competitive maps produce election results that are highly sensitive to small voter swings. For this reason, there is an inherent tension under the traditional system between drawing competitive districts and achieving a certain partisan outcome. Partisan gerrymanders pack opponents into a smaller number of super-safe districts that they win by a landslide, and then crack the remaining opponents into a larger number of moderately safe districts that they lose by a small-but-still-decisive margin.146 Bipartisan gerrymanders carve the state up into an appropriate number of safe blue districts and safe red districts. To maintain partisan fairness over shifts in voter preferences, a mapmaker must strategically vary the competitiveness of districts, and most of them must be uncompetitive.147 In contrast, MM2 can tolerate more competitive districts, because it can countermand any seats-votes distortion that competitiveness may produce. Free from the fear of the huge seat bonus, the mapmaker committed to partisan fairness can confidently draw more competitive districts. The result may be a more competitive map.

Third Parties and Independents

By construction, MM2 shuts out third parties and protects the two-party system. It may be hard for a proponent of multipartyism to support such a proposal. But there are several reasons to do so. First, third parties are not thriving under the present system.148 If the choice is the status quo or MM2, why not support the alternative without gerrymandering and its pathologies?

MM2 also affirmatively supports third parties in a meaningful sense, when it uses a single-transferable vote for both tiers and reports the results of each round in the transfer process. This means a voter can express their preferences for a third party or a third-party candidate without the spoiler effect. The election results will reflect how many voters gave their top ranking to a third party or a third-party candidate. The Green (or Libertarian) party could use this to mobilize support, demonstrate its strength, and push the Democrats (or Republicans) to shift to the left (or right). In this way, MM2 facilitates third party mobilization much more than the prevailing system.

MM2 also facilitates mobilization for multipartyism. MM2 converts the two-party structure from an inexorable result of the system to a specific, alterable design choice. The implementing statute would literally codify bipartyism in language that qualified the word “party” by the word “major.” Specifically, the relevant statutory provision would read something like: “The list seats shall be allocated so that, for each major party, seats share corresponds to vote share.” To shift from MM2 to MMP, simply strike the word “major.” In this way, MM2 represents a significant intermediate step between the prevailing system and MMP. To put it in Dean Gerken’s terms, there may be no path directly from Here to MMP, but there may be a path from Here to MM2 and a path from MM2 to MMP.149 Indeed, with the New Zealand experience in mind, MM2 could be adopted as part of a package that includes a formal study of multipartyism (blue-ribbon committee and/or citizen commission) with a referendum following its report.

The Details

This Section canvasses the key design questions a state must address when implementing MM2.

Chamber Size and Class Ratio

MM2 requires a two-class legislative body, with some nominal seats awarded to individual candidates and some list seats awarded to political parties. This immediately presents some numerical questions: How many seats should there be overall? How many nominal seats versus list seats? When adopting MM2, should a state convert nominal seats into list seats, or just add more list seats, thereby increasing the overall size of the legislature? I suggest the state maintain the size of the nominal class, and thus the number of single-seat districts, but add a list class roughly half the size of the nominal class, thereby increasing the overall size of the legislature by roughly 50 percent.

There may be some drawbacks to increasing overall size. The size of the legislature may be codified in the state constitution, which may be hard to amend.150 Also, people are generally skeptical of any proposal to improve politics by creating more politicians or expanding legislatures,151 though I suspect people are more concerned about gerrymandering than a larger legislature.

Ultimately, I prefer to increase the size of the legislative body rather than decrease the size of the nominal class—for several reasons. A decrease in the number of single-seat districts would necessarily pit incumbents against each other. If adopted, this would disrupt pre-existing relationships between representative and their constituents. And pre-adoption, this prospect would make the proposal a harder sell with incumbent legislators. For this reason, I suggest it is particularly important to use chamber expansion rather than seat conversion in states that cannot adopt MM2 without legislative approval.

As explained above, the adoption of MM2 would not violate the Voting Rights Act.152 Moreover, the list tier would offer a new way to protect minority voting rights that does not depend on Section 2 litigation to establish a majority-minority district.153 However, a decrease in the number of single-seat districts would necessarily entail an increase in the (area and population) size of each district, which may disrupt pre-existing majority-minority districts. When deciding whether to implement MM2 through chamber expansion or seat conversion, all else being equal, I prefer the approach that preserves pre-existing majority-minority districts and consider this an additional argument in favor of chamber expansion.

As for the relative size of the two classes, a range of values are possible. In 2017, New Zealand’s parliament consisted of 120 seats, 71 in the nominal class, 49 in the list class.154 The essential idea is to allocate list class seats to compensate for any distortions in the nominal class. The more list class seats, the greater the capacity to compensate, and the larger the distortions the system can handle.

In general, let  denote the number of list seats and  the number of nominal seats, and define the class size ratio  as the relative size of the two classes. Let  denote the statewide vote share for the party disfavored by the electoral map, and let  denote the share of nominal seats won by that disfavored party. The minimum class size ratio needed to achieve overall vote-seat biproportionality, which I denote , can be expressed in terms of  and .


In the previous example, the 2018 Wisconsin Assembly map so disfavored the Democrats that they won only  percent (i.e., 36 of 99) of the nominal seats with  of statewide votes. In this case, the minimum necessary class size ratio is just over 35 percent.


If we add 40 list seats to the 99 nominal seats for a total of 139 seats, the class size ratio  is just over 40 percent, a little more than the minimum ratio of 35.4 percent, and thus enough to achieve perfect disproportionality.

Two-Votes and Vote Linkage

While the nominal class allocates each district a local representative, the list class allocates seats to the top two political parties to align votes and seats, popularity and power. But here an important design question arises: how do we determine each party’s popularity? And more specifically: should we give voters one vote or two?

In the one-vote version, the ballot ask voters a single question: which individual do you prefer for district representative? Election administrators aggregate voter responses to this single question within each district to determine which candidate wins that district’s seat. Election administrators then aggregate voter responses to this same question statewide to determine the relative popularity of the top two political parties, treating a vote for a district candidate as support for that candidate’s party. Election administrators then allocate list seats based on the countermandering principle, using the share of votes cast statewide for candidates of each of the top two parties.

Specifically, let  denote a set of  single-seat districts in an ordered sequence with index  such that  denotes the th district in the sequence. Let  denote the set of  political parties in an ordered sequence with index  such that  denotes the th district in the sequence and the top two parties are  and . Without loss of generality, I will follow the notational convention that the Democrats are  and the Republicans are . Assume each district features one candidate from each of the top two parties. Let  denote the number of ballots cast for the  candidate in district . Without loss of generality, define the two-party statewide district vote share  as:


The Wisconsin Assembly example supra implicitly uses this approach.

But, of course, this is not the only, or necessarily the best, way to ascertain the relative popularity of the top two parties. Instead of estimating party preferences based on expressed candidate preferences, we can simply ask voters for their party preferences. In the two-vote version of MM2, the ballot asks voters two questions:

(1) Which individual do you prefer for district representative?

(2) Which party do you prefer overall?

Election administrators then aggregate voter responses to the second question to determine the relative popularity of the top two political parties. Specifically, election administrators allocate list seats based on the countermandering principle, using relative share of votes cast statewide for each of the top two parties. Both Germany and New Zealand use this two-vote approach to traditional MMP.

One argument in favor of the one-vote version of MM2 is that it operates just like the prevailing electoral schema from the perspective of voters. The ballot would look identical and the voter would fill it out in just the same way, by indicating their preference for district representative.

But the two-vote version offers important advantages that justify the shift to a new ballot structure and voter experience. First, the two-vote version may provide better incentives for all involved. As discussed above, it gives every voter a good reason to participate, and every candidate and party a good reason to compete hard for every vote in every district, even in the safest district, even in an uncontested district.155

Relatedly, MM2 may more accurately ascertain the relative popularity of the top two parties if implemented with two votes instead of just one. In the one-vote version, voters cannot directly express their preferences for political parties. Party support can be measured only indirectly by aggregating voters’ candidate preferences. Obviously, there is a strong correlation between support for a candidate in a district race and support for that candidate’s political party. But the two are not identical, and there are significant problems with using district outcomes to measure statewide party support. With only one vote, a voter can only express support for a political party by voting for a candidate from that party. This creates a serious problem in uncontested districts, where one of the two major parties runs no candidate, and the district vote totals thus provide no information about the party preferences of that district’s voters. Recall that the 2018 Wisconsin Assembly map featured uncontested races in one third of all districts. Even when a district race is contested, the votes each candidate earns provides an imperfect signal of party preferences—for several reasons. One is the well-documented tendency of incumbent legislators to outperform their party. Partly because of these incumbency and other candidate-specific effects, political scientists often prefer to estimate party affiliation based on statewide elections rather than legislative elections.156 Another is turnout effects. Consider an eligible voter in an uncompetitive district whose preferred candidate is destined to lose. Suppose this voter would turn out on election day if the ballot features two questions, but stay home if the ballot features only one. The two-vote version will capture this voter’s party preference, but the one-vote version will miss it.

For these reasons, I prefer the two-vote version of MM2. But yet another question remains: if voters directly express preferences for political parties, should we ignore candidate preferences and rely exclusively on party preferences to determine relative party popularity? In other words, should the allocation of list seats depend only on the list vote, or should it also take into account the nominal vote? In their traditional MMP systems, both Germany and New Zealand rely exclusively on the list vote, and ignore the nominal vote, when allocating list seats.157 In his study of mixed-member systems, Matthew Shugart used the terms “seat linkage” and “vote linkage.”158 Seat linkage means list seat allocation is partially based on nominal seat share.159 A mixed-member proportional system is defined by “compensatory” seat linkage, which is equivalent to the countermandering principle, i.e., awarding list seats to compensate for misallocation of nominal seats.160 Vote linkage means list seat allocation is partially based on nominal vote share.161 In his canvass of mixed-member systems around the world, Shugart found no system that combined seat linkage and vote linkage but considered the approach plausible.162

There are three arguments in favor of vote linkage. First, vote linkage may provide the most accurate way to measure party support. For example, consider a voter who leaves blank the list portion of the ballot but votes for the Republican candidate on the nominal portion of the ballot. Is it more accurate to treat this ballot as providing some indication of Republican support or no indication of Republican support? Second, vote linkage encourages each political party to recruit the strongest possible candidates and campaign vigorously in each district. Third, as I explore in greater detail in the next subsection, mixed-member systems are vulnerable to strategic circumvention when implemented with two votes and seat linkage, but vote linkage addresses this vulnerability.

Strategic Circumvention

With two votes, voters can split their ballot, casting their list vote for one party while casting their nominal vote for a candidate not affiliated with that party—either a candidate affiliated with another party or an independent candidate. Candidates can strategically choose their party affiliation and voters can strategically split their ballots. When implemented with seat linkage but no vote linkage, the allocation of list seats favors those parties who earn many list votes but few nominal seats, and the number of nominal votes is irrelevant. To maximize its share of list seats, a party must minimize its share of nominal seats but maximize its share of list votes.

These considerations suggest two circumvention strategies, which I will call (1) the dummy-party strategy; and (2) the faux-independent strategy. In the traditional ballot-split dummy-party strategy, a single political coalition traditionally organized under a single party could establish a second “dummy” party to game the system. With two parties, the coalition can essentially earn double the seats attainable as a single party. One party competes only for nominal seats, while the other party competes only for list seats. Supporters of the coalition strategically split their votes, awarding their nominal votes to one party and their list votes to the other party. The nominal party wins many nominal seats. The list party earns many list votes but no nominal seats, so the system automatically allocates this party many list seats. The two parties can then work together in the legislature and pursue the common policy agenda of their shared political coalition with greater influence due to their artificially doubled collective seat share. No political coalition has attempted a dummy-party strategy in Germany or New Zealand, but such abuse has been observed in Albania, Lesotho, Italy, and Venezuela.163

The faux-independents strategy is less brazen. Instead of establishing a dummy party, a political coalition organizes as a single party competing for both nominal and list seats. But some of the party’s nominal candidates
—perhaps the incumbents in the safest districts—strategically register as independents. These incumbents win reelection, but the system does not attribute these nominal seats to the political party because the incumbents were formally independent candidates with no party affiliation. This artificially reduces the party’s nominal seat share, which may cause the system to allocate it more list seats. Once they’ve secured reelection, these pseudo-independents can then caucus with their former party, or even change back their party affiliation.

With the right implementation, MM2 can minimize the risk of either circumvention strategy. Note that MM2 already makes the dummy-party strategy far more difficult because it allocates no list seats to third parties. To pursue the dummy-party strategy under MM2, a single political coalition would have to run two parties, one that competes only for nominal seats, and one that competes only for list seats, and the second party would have to earn list seats, presumably by earning the most, or the second-most, list votes. But this assumes the “top two” parties are the two parties with the most list votes. Instead, I propose an alternative definition: the “top two” parties eligible for list seat allocation are the two parties whose candidates earn the most nominal votes. With this nominal-vote definition of the “top two” parties eligible for list seat allocation, the dummy-party strategy fails.

But what about the faux-independent strategy? Recall that Wisconsin Republicans won 63 of 99 Assembly seats with 47 percent of the 2018 vote. To see the faux-independent strategy in action, make the following simplifying assumptions: 25,000 voters cast ballots in each of the 99 districts; each district was a head-to-head race between one Republican and one Democrat (no independent or third party candidates); under the two-voter version of MM2, 47 percent of voters would have cast their nominal vote for the Republican candidate and their list vote for the Republican party, while 53 percent of voters would have cast their nominal vote for the Democrat candidate and their list vote for the Democratic party. If Republicans had won 63 of the 99 nominal seats, Democrats would have received 36 of the 40 list seats to align each party’s overall seat share with its list vote share: Of the 139 seats, Democrats get about 53 percent, i.e., 73, and Republicans get about 47 percent, i.e., 67.

Table 3: Counterfactual 2018 Wisconsin Assembly Election with MM2

But now consider what would have happened if 13 of those 63 successful Republican candidates had run as independents, and won an average 60 percent vote share. (Note that I am assuming that Republicans willing to run as faux-independents represent safer districts.) Suppose that each of those 13-districts featured a head-to-head contest between one Democrat and one faux independent. In these 13 districts, 60 percent of voters cast a nominal vote for the faux independent candidate and a list vote for the Republican party while the other 40 percent of voters cast a nominal vote for the Democrat candidate and a list vote for the Democratic party. 

Since 13 Republicans won seats as faux independents, formal Republicans won only 50 of 99 seats. This artificially reduces the apparent vote-seat distortion that partisan gerrymandering produces in the nominal class. If the list seats are now allocated to align overall two-party seat share and two-party list vote share for each of the top two parties, Democrats will now get only 31 of 40 list seats, while Republicans would get the remaining nine. Overall, Democrats would have 67 seats, while Republicans would have only 59 seats. But the remaining 13 seats would be held by faux independents who would caucus with the Republicans, producing a 72-seat majority “coalition” of 59 Republicans and 13 “independents.” Democrats would still be much better off with this outcome than the real outcome. But the faux independent strategy would essentially flip six seats from Democrats to Republicans, substitute a 72-68 Republican legislative majority for a 73-66 Democratic legislative majority, and snatch minority rule from the jaws of structural countermandering.

Table 4: Counterfactual 2018 Wisconsin Assembly Election with MM2

13 Republicans Run as Faux Independents List Seats Allocated
Based on List Vote Share


This example demonstrates how the faux-independent strategy can circumvent MM2—when list seats are allocated exclusively on the basis of the list vote, i.e., without vote linkage. But in the preceding section, I identified powerful arguments in favor of voter linkage: (1) to get a more accurate measure of relative party support; and (2) to incentivize each political party to recruit the strongest possible candidates and campaign vigorously in each district. And here’s the third argument for vote linkage: it mitigates the impact of the faux-independent strategy. To see how, consider the same example, but this time with vote linkage. Specifically, the system will now allocate list seats to align, for each of the top two parties, overall seat share with overall vote share, based on the sum of a party’s list votes and its candidate’s nominal votes.

Under this approach, the faux-independent strategy still reduces the number of nominal seats attributed to the Republican party. But it also reduces the number of nominal votes attributed to the Republican party, which reduces the Republican party’s overall vote share and its corresponding allocation of list seats. So while Republicans net three seats with this strategy, the Democrats still win a majority of seats overall, and MM2 still achieves majority rule.

Table 5: Counterfactual 2018 Wisconsin Assembly Election with MM2

13 Republicans Run as Faux Independents List Seats Allocated
Based on Overall Vote Share


Of course, the faux-independent strategy has a greater effect when more partisans run as independents. But there are costs, risks, and coordination problems that may limit the number of partisans willing to run as independents. And electoral rules could further limit this strategy. For example, consider a rule that a party must challenge at least 85 percent of district races in order to qualify as a top-two party eligible for list seat allocation.

In light of the foregoing analysis, I recommend the two-vote version of MM2 with a nominal-vote definition of top-two and vote linkage, where list seat allocation is based on overall vote share rather than list vote share alone. I am hopeful this system can withstand both dummy-party and faux-independent circumvention strategies. But I recommend further study of the circumvention question, including other possible mechanisms to deter and thwart circumvention strategies.

Ranked Choice and Vote Transfer

Up to this point, I have assumed that each voter, for each question, indicates their single top choice, and election administrators simply add up top-choices, awarding nominal seats based on the plurality voting formula and awarding list seats to align overall two-party seat share and overall two-party vote share, computed again by simply adding up top-choices. But MM2 could also be implemented with ranked choices, both for the nominal vote and the list vote, and there are strong reasons to consider this approach.

As discussed above, AES uses single-choice voting and the plurality formula, supplemented in only two states with a run-off if no candidate wins a majority in the general election.164 This approach has the virtues of familiarity and simplicity, easy for voters to understand and for election officials to administer. But it also has well-known, significant drawbacks. With more than two candidates, single-choice voting and the plurality formula can elect a candidate opposed by most voters.165 This may occur due to “vote-spoiling,” where the less popular major party candidate prevails because a third-party candidate peels off enough votes from the more popular major party candidate,166 or “vote-splitting,” where a candidate with minority support wins because the opposition splits their votes among two or more alternatives.167 In the context of legislative elections, support for third party or independent candidates could produce a plurality-majority vote-seat inversion, where a party could win a majority of seats with a plurality, but not a majority, of votes. From 1968 to 2016, plurality-majority vote-seat inversion has occurred 68 times in state legislative elections, 35 for state upper chambers and 33 for state lower chambers.168 Georgia and Louisiana avoid this possibility by holding a subsequent run-off election whenever no candidate wins an absolute 50 percent +1 majority in the general election.169 This approach has its own disadvantages: The cost and inconvenience of another election, which will likely have lower voter turnout than the first, and an extended period of uncertainty before the winner is finally determined.170

Under the psychological Duverger effect, third party and independent voters and candidates anticipate the risk of vote-spoiling and behave accordingly. Instead of expressing her true preference, a voter may act strategically, hold her nose, and vote for the lesser of two evils, to avoid vote-spoiling, and to have her say on which of the two major party candidates prevail. Or she might not turn out at all. Similarly, a prospective third party or independent candidate might simply decide against running because he has no realistic prospect of success, only a realistic prospect of playing the spoiler. For this reason, single-choice voting fails to accurately measure latent demand for third party candidates.

Given the disadvantages of single-choice voting, state and local governments across the United States are increasingly embracing ranked-choice voting.171 Under ranked-choice voting, the voter ranks the options, marking 1 by her top choice, 2 by her second choice, 3 by her third choice, and so forth. These preferences are then aggregated through an iterative process of candidate elimination and vote transfer, taking into account a voter’s second-choice if her first-choice candidate loses (and her third-choice if her second-choice candidate loses). Through this reallocation mechanism, ranked-choice voting avoids vote-spoiling, ensures the winner enjoys majority support, and permits voters to support third-party or independent candidates without inadvertently helping their least-preferred candidates. Ranked-choice voting may have additional salutary effects: less negative campaigning and more civility;172 more candidates and a more diverse group of elected officials.173 However, ranked-choice voting may have other disadvantages, related to voter confusion, ballot exhaustion, and administrative burdens.174

The motivating objective of MM2 is to accomplish two-party structural countermandering: eliminate gerrymandering and associated pathologies while maintaining two-party legislative majoritarianism and personal, geographic representation via single-seat electoral districts. MM2 can perform this function without any ranked-choice voting, relying solely on single-choice voting in both the nominal and list tiers. MM2’s nominal tier single-seat district elections can use single-choice voting with a plurality (or majority) electoral formula, just like states do now under AES. The two-vote version of MM2 can similarly use single-choice voting for the list vote, asking each voter to indicate the single political party she most prefers. Election officials can then simply ignore ballots cast for third parties, and compute the vote totals for the top two parties. With the vote linkage approach I advocate for supra, election officials can simply ignore ballots cast for third party or independent candidates, and compute the vote totals for candidates affiliated with the top two parties.

However, ranked-choice voting may enhance MM2 by avoiding vote-spoiling and vote-splitting, allowing voters to express their true preferences, boosting voter turnout and external efficacy, reducing negative campaigning and polarization, and promoting more civility, candidate entry, and chamber diversity. I suspect these benefits outweigh ranked-choice voting’s costs. If so, I support implementing MM2 with ranked-choice voting.

For MM2’s nominal tier single-seat district elections, ranked-choice voting would operate just as described above. For the allocation of MM2’s list tiers, ranked-choice voting could be used to account for every voter’s preference between the top two parties, even those voters who most prefer third parties, third-party candidates, or independent candidates. As explained above, I propose a two-vote version of MM2 with vote linkage. Under this approach, list seats are allocated to align, for each of the top two parties, overall seat share with overall vote share, and overall vote share takes into account both the list vote for each party and the nominal vote for each party’s candidates. In the ranked-choice vote version of MM2, use the eliminate-and-transfer process to aggregate over voter’s preferences. To determine the list vote for each of the top two parties, keep eliminating the party with the fewest list votes and transferring that party’s ballots until only two parties remain with list votes. To determine the nominal vote for each of the top two parties, perform an analogous process of elimination and transfer, ignoring votes for independent candidates and otherwise treating a vote for a candidate as a vote for that candidate’s party, until only two parties remain with nominal votes.

If MM2 is implemented with ranked-choice voting, I suggest election officials make public the vote tallies at each stage of the process. This promotes transparency, but also provides an objective measure of the latent demand for third parties and third-party candidates.


Electoral system reform is a big ask, and may prove to be a Sisyphean effort. For this reason, anyone proposing to change a long-used electoral system bears the burdens of production and persuasion, and the standard of proof should be commensurate with the scope of the relief requested. The alternative system must be presented with sufficient detail to facilitate rigorous scrutiny, and the system must withstand that scrutiny. It is not enough to show that the current system has flaws or that the proposed system might offer advantages. The current system’s flaws must be so profound, and the proposed system’s advantages so compelling, that it is worth investing in the reform effort, and pushing the boulder up the hill. That is precisely the claim I mean to make. The current system profoundly disserves American democracy. A strategy of constraint is insufficient. MM2 would be a game-changer. And the game must change. I hope this Article has succeeded in persuading the reader of these claims, or at least providing enough support to survive a motion to dismiss.

If so, there is much more to consider, which I hope to address in future work. In which state might a push for MM2 prove most successful? Michigan comes to mind. It has recently experienced egregious partisan gerrymanders that produced vote-seat inversions.175 And Michiganders can use the initiative process to change the electoral system directly, bypassing the legislature.176 Indeed, they used this process in 2018 to adopt an independent redistricting commission.177 Another possibility is Idaho, a state so deep red that MM2 would not threaten Republican control, and one of the few states where the legislature itself established a bipartisan redistricting commission.178

This Article has focused on state legislative chambers. In a subsequent piece, I plan to explore prospects for MM2 in congressional elections. In this context, the number of seats is much smaller, exhibits far greater inter-state variation, and is quite likely fixed—unless Congress decides to expand the size of the House of Representatives for the first time in almost a century.179 This means that a state can only implement MM2 for its congressional delegation by converting some nominal seats into list seats, thereby expanding the (population and area) size of each single-seat district, potentially disrupting pre-existing majority-minority districts, and likely pitting incumbents against each other.180 And MM2 needs some minimum number of seats overall, so it can convert enough nominal seats into list seats to accomplish structural countermandering through compensatory list seat allocation. However, MM2 generally requires fewer seats than MMP, because it only seeks vote-seat proportionality for the top two parties rather than vote-seat proportionality for every party. And something like ten seats might do the trick. For example, in the 2010 cycle, North Carolina had 13 seats in the House of Representatives.181 Imagine North Carolina had implemented MM2 for congressional elections by converting three of its 13 seats into list seats. This would have prevented majority-minority vote-seat inversion unless the mapmaker had somehow drawn a partisan gerrymander so egregious that it gave the party with minority vote share seven out of ten nominal seats.182 This suggests that MM2 could work, at least for more populous states with larger congressional delegations. Note that the 17 most populous states have at least nine House Representatives, and they collectively account for 305 of the chamber’s 435 seats, i.e., over 70 percent.183

MM2 reflects normative and predictive intuitions about the U.S. electoral and party systems: electoral system reform is imperative, because AES is so pathological and so resistant to intra-schema constraint; but electoral system reform is only possible within the confines of the prevailing party system, because elites invested in the two-party system will successfully defeat any reform that threatens it. If these intuitions are correct, the only way forward is to maintain the party system but change the electoral system—to depathologize the two-party system. MM2 does just that. It maintains geographic representation and two-party, legislative majoritarianism. But it uses structural countermandering to achieve a better two-party electoral system. Under MM2, the partisan composition of the legislative body is ultimately determined by the relative popularity of the two major parties, no matter how the electoral map is drawn, no matter how a party’s supporters are distributed geographically. Since each party’s power actually reflects its popularity, every vote matters equally in determining the legislative body’s partisan composition. By giving every voter an equal say on the relative power of the two major parties, MM2 promotes voter efficacy and voter turnout, competitiveness and responsiveness. If we are going to have a two-party system, MM2 is a far better way to run elections.

The 2010 redistricting cycle was a good decade for partisan mapmakers, and a bad one for American democracy. As we begin the 2020 redistricting cycle, the time is ripe for exploring new approaches. Structural countermandering through MM2 warrants serious consideration.

  1. [1]. Rucho v. Common Cause, 139 S. Ct. 2484, 2512–13 (2019) (Kagan, J., dissenting) (“[B]ig data and modern technology ... make today’s gerrymandering altogether different from the crude linedrawing of the past.... Mapmakers now have access to more granular data about party preference and voting behavior than ever before.... [A]dvancements in computing technology have enabled mapmakers to put that information to use with unprecedented efficiency and precision.”).

  2. [2]. Sam Wang, The Great Gerrymander of 2012, N.Y. Times (Feb. 2, 2013), https:// [https://perma.

  3. [3]. Ohio A. Philip Randolph Inst. v. Householder, 373 F. Supp. 3d 978, 1074 (S.D. Ohio 2019).

  4. [4]. See League of Women Voters of Pa. v. Commonwealth, 178 A.3d 737, 764–65 (Pa. 2018).

  5. [5]. Rucho, 139 S. Ct. at 2509–10 (Kagan, J., dissenting).

  6. [6]. Whitford v. Gill, 218 F. Supp. 3d 837, 846 (W.D. Wis. 2016), vacated, 138 S. Ct. 1916 (2018).

  7. [7]. Mitchell Schmidt, 2020 Election Again Shows Lopsided Republican Legislative Maps, Wis. St. J. (Nov. 12, 2020), [https://].

  8. [8]. See Vieth v. Jubelirer, 541 U.S. 267, 360 (2004) (Breyer, J., dissenting) (“By entrenchment I mean a situation in which a party that enjoys only minority support among the populace has nonetheless contrived to take, and hold, legislative power.”); Nicholas Stephanopoulos, Reforming Redistricting: Why Popular Initiatives to Establish Redistricting Commissions Succeed or Fail, 23 J.L. & Pol. 331, 381 n.282 (2007) (using the term “minority entrenchment” to describe cases where “the majority party in the state legislature only enjoys the support of a minority of the electorate”); Mitchell N. Berman, Managing Gerrymandering, 83 Tex. L. Rev. 781, 814 n.207 (2005) (“minority-to-majority entrenchment”); Jack Santucci, Multiparty America?, 82 J. Pol.34, 34 (2020) (“wrong-winner elections (e.g., when the party with the most votes does not win public office)”); Thomas Quinn, Throwing the Rascals Out? Problems of Accountability in TwoParty Systems,55 European J. Pol. Rsch.120, 121 (2016) (“‘wrong-winner’ elections in two-party systems, where the party winning the most votes does not win the most seats”); Alan Renwick, Do ‘Wrong Winner’ Elections Trigger Electoral Reform? Lessons From New Zealand, 45Representation 357, 357 (2009).

  9. [9]. Miriam Seifter, Countermajoritarian Legislatures, Colum. L. Rev. (forthcoming 2021) (manuscript at 4–5), [https://] (“[M]ost minimally, we should expect that the candidate or party that receives the most votes will win.”); Nicholas O. Stephanopoulos, Elections and Alignment, 114 Colum. L. Rev. 283, 287 (2014) (proposing a partisan alignment ideal, whereby “if a majority of voters wish to be represented by a candidate from a certain party, this in fact is who represents them”); Steven Levitsky & Daniel Ziblatt, End Minority Rule: Either We Become a Truly Multiracial Democracy or We Cease to Be a Democracy at All, N.Y. Times (Oct. 23, 2020), https://www. [] (“Democracy is supposed to be a game of numbers: The party with the most votes wins.”); Daryl R. DeFord, Nicholas Eubank & Jonathan Rodden, Partisan Dislocation: A Precinct-Level Measure of Representation and Gerrymandering, Pol. Analysis (forthcoming 2021) (manuscript at 4), [
    F65Q-6SHW] (“In the most obvious normative failure, a party with less than half of the statewide votes can receive more than half of the seats, which happens routinely in U.S. state legislatures.”); Douglas J. Amy, Real Choices/New Voices: How Proportional Representation Elections Could Revitalize American Democracy 38 (2002) (Minority entrenchment “violate[s] one of the most sacred tenets of democratic politics: majority rule”).

  10. [10]. Seifter, supra note 9 (manuscript at 28). Seifter found an additional 78 cases where a party won a majority of seats with a plurality, but not a majority, of votes, which occurs when the votes cast for third party and independent candidates exceed the vote difference between the top two parties. Id.

  11. [11]. Federal courts have identified multiple congressional maps as intentional pro-Democratic gerrymanders, including those drawn by New Jersey post-1980, by Texas post-1990, and by Georgia post-2000. See Karcher v. Daggett, 462 U.S. 725, 764 (1983) (Stevens, J., concurring) (The congressional map New Jersey adopted post-1980 “was designed to increase the number of Democrats, and to decrease the number of Republicans, that New Jersey’s voters would send to Congress in future years.”); League of United Latin Am. Citizens v. Perry, 548 U.S. 399, 410–11 (2006) (The congressional map Texas adopted post-1990 was “designed to favor Democratic candidates ... [u]sing then-emerging computer technology to draw ... lines with artful precision ... [and was] later described as the ‘shrewdest gerrymander of the 1990s.’” (citations omitted)); Larios v. Cox, 300 F. Supp. 2d 1320, 1329 (N.D. Ga.), aff’d, 542 U.S. 947 (2004) (The congressional map Georgia adopted post-2000 “was an intentional effort to allow incumbent Democrats to maintain or increase their delegation, primarily by systematically underpopulating the districts held by incumbent Democrats, by overpopulating those of Republicans, and by deliberately pairing numerous Republican incumbents against one another.”).

  12. [12]. Seifter, supra note 9 (manuscript at 28).

  13. [13]. Simon Jackman, Assessing the Current North Carolina Congressional Districting Plan 8 (2017), [] (“Recent decades show more pro-Republican gerrymandering, as measured by the efficiency gap.”).

  14. [14]. Of the 88 legislative chambers that held elections in 2010, Republicans won 53 of them, flipping 20 previously controlled by Democrats. State Legislative Elections Results, 2010: State Legislature, Ballotpedia,,_2010 [] (last updated Aug. 9, 2021). After the 2010 elections, Republicans enjoyed trifectas (i.e., the Governor’s mansion plus majorities in both legislative chambers) in 20 states. State Legislative Elections Results, 2010: Trifectas, Ballotpedia, https://,_2010#Trifectas [] (last updated Aug. 9, 2021).

  15. [15]. Jowei Chen & Jonathan Rodden, Unintentional Gerrymandering: Political Geography and Electoral Bias in Legislatures, 8 Q. J. Pol. Sci. 239, 257 (2013) (“[H]uman geography makes the task of a Democratic cartographer far more difficult than that facing a Republican-favoring cartographer ... .”); Gretchen Helmke, Mary Kroeger & Jack Paine, Democracy by Deterrence: Norms, Constitutions, and Electoral Tilting, Am. J. Pol. Sci. (forthcoming) (manuscript at 17), [] (“[T]he most extreme pro-Republican gerrymander nets more seats for Republicans than does the most extreme pro-Democratic gerrymander for Democrats.”); Davis v. Bandemer, 478 U.S. 109, 159 (1986) (O’Connor, J., concurring) (“[T]he voting strength of less evenly distributed groups will invariably be diminished by districting as compared to at-large proportional systems for electing representatives.”); Edward B. Foley, Due Process, Fair Play, and Excessive Partisanship: A New Principle for Judicial Review of Election Laws, 84 U. Chi. L. Rev. 655, 668 (2017) (“Given the way Democrats and Republicans sort themselves geographically in many states, with Democrats concentrated in urban areas and Republicans spread out in rural or exurban ones, a map reflecting this differential population density would make it more difficult for Democrats to translate votes into seats than for Republicans to do so.”); Nicholas Goedert, The Case of the Disappearing Bias: A 2014 Update to the “Gerrymandering or Geography” Debate, 2015Rsch. & Pol. 1, 3 (2015). (“[T]he absence of bias in 2014, just like the presence of bias in 2012, is explainable by a combination of intentional gerrymandering and the asymmetric distribution of partisans.”).

  16. [16]. See infra Section II.A.5.

  17. [17]. See generally Davis, 478 U.S. 109 (holding that plaintiffs did not prove a violation of the Equal Protection Clause); Vieth v. Jubelirer, 541 U.S. 267 (2004) (upholding a District Court’s decision to dismiss a gerrymandering claim); League of United Latin Am. Citizens v. Perry, 548 U.S. 399 (2006) (dividing the Court on a partisan gerrymandering claim).

  18. [18]. Whitford v. Gill, 218 F. Supp. 3d 837, 843 (W.D. Wis. 2016), vacated, 138 S. Ct. 1916 (2018).

  19. [19]. Rucho v. Common Cause, 139 S. Ct. 2484, 2506–08 (2019).

  20. [20]. Kylee Groft, The Results are In: Most Americans Want Limits on Gerrymandering, Campaign Legal Ctr. (Sept. 11, 2017), []; Bipartisan Poll Shows Strong Support for Redistricting Reform, Campaign Legal Ctr. (Jan. 28, 2019),
    /update/bipartisan-poll-shows-strong-support-redistricting-reform [].

  21. [21]. Rucho, 139 S. Ct. at 2506–07 (“Excessive partisanship in districting leads to results that reasonably seem unjust. But the fact that such gerrymandering is ‘incompatible with democratic principles,’ does not mean that the solution lies with the federal judiciary. We conclude that partisan gerrymandering claims present political questions beyond the reach of the federal courts.” (quoting Ariz. State Legislature v. Ariz. Indep. Redistricting Comm’n, 135 S. Ct. 2652, 2658 (2015)).

  22. [22]. See Davis, 478 U.S. at 118–25, abrogated by Rucho, 139 S. Ct. 2484 (2019).

  23. [23]. Rucho, 139 S. Ct. at 2507.

  24. [24]. Id. at 2507–08.

  25. [25]. The states are Arizona, California, Colorado, Delaware, Florida, Hawaii, Iowa, Michigan, Minnesota, Montana, Ohio, Nebraska, New York, Oregon, Utah, and Washington. See Redistricting Criteria, Nat’l Conf. State Legislatures (July 16, 2021),
    redistricting/redistricting-criteria.aspx [].

  26. [26]. The states are Alaska, Arizona, Arkansas, California, Colorado, Hawaii, Idaho, Michigan, Missouri, Montana, New Jersey, Ohio, Pennsylvania, Virginia, Washington. Creation of Redistricting Commissions, Nat’l Conf. State Legislatures (June 14, 2021),
    research/redistricting/creation-of-redistricting-commissions.aspx#_ftn1 [
    A3-MVXC]. Other states use advisory or backup commissions. See Redistricting Commissions: State Legislative Plans, Nat’l Conf. State Legislatures (July 12, 2021),
    research/redistricting/2009-redistricting-commissions-table.aspx [].

  27. [27]. The states are Michigan, Colorado, and Missouri. Creation of Redistricting Commissions, supra note 26. Michigan did so by citizens’ initiative. Id.

  28. [28]. Virginia became the 15th state to adopt an independent redistricting commission. See Creation of Redistricting Commissions, supra note 26.

  29. [29]. Rucho, 139 S. Ct. at 2513 (Kagan, J., dissenting) (“And gerrymanders will only get worse (or depending on your perspective, better) as time goes on—as data becomes ever more fine-grained and data analysis techniques continue to improve. What was possible with paper and pen—or even with Windows 95—doesn’t hold a candle (or an LED bulb?) to what will become possible with developments like machine learning.”); Vieth v. Jubelirer, 541 U.S. 267, 312–13 (2004) (Kennedy, J., concurring) (“Technology is both a threat and a promise. On the one hand, if courts refuse to entertain any claims of partisan gerrymandering, the temptation to use partisan favoritism in districting in an unconstitutional manner will grow. On the other hand, these new technologies may produce new methods of analysis that make more evident the precise nature of the burdens gerrymanders impose on the representational rights of voters and parties.”).

  30. [30]. Reformers strategically use this personalization. One campaign has created a parody social media presence for the late Elbridge Gerry: @GovGerry’s twitter bio reads: “I was a Founding Father, governor & vice president. But today people only remember me for ‘Gerrymandering.’ Not cool. Let’s #EndGerrymandering!” Elbridge Gerry (@GovGerry), Twitter, [] (last updated Aug. 9, 2021). Another is the “F*ck Gerry(mandering) Fund,” launched by Crooked Media to raise money for redistricting initiatives. Lucy Diavolo, The Team Behind Pod Save America Launches the F*ck Gerry(mandering) Fund, TeenVogue (May 17, 2019),
    pod-save-america-crooked-media-fuck-gerrymandering-fund [].

  31. [31]. See generally David Daley, Ratf**ked: The True Story Behind The Secret Plan To Steal America’s Democracy (2016) (explaining how politicians used tricks and redistricting to return to power).

  32. [32]. See generally Chen & Rodden, supra note 15 (showing “substantial bias can ... emerge from patterns of human geography”).

  33. [33]. See Heather K. Gerken, Getting from Here to There in Election Reform, 34 Okla. City U. L. Rev. 33, 33 (2009); Heather K. Gerken, Getting from Here to There in Redistricting Reform, 5 Duke J. Const. L. & Pub. Pol’y 1, 3 (2010).

  34. [34]. Gerken, Getting from Here to There in Election Reform, supra note 33, at 33.

  35. [35]. See, e.g., Denis K. Kadima, Choosing an Electoral System: Alternatives for the Post-War Democratic Republic of Congo, 2 J. African Elections 33, 36 (2006) (“The SMP is also known as ‘first-past-the-post’, ‘winner-takes-all’, ‘simple majority’ or ‘relative majority’.”).

  36. [36]. See, e.g., Kenneth Benoit, District Magnitude, Electoral Formula, and the Number of Parties, 39 Eur. J. Pol. Rsch. 203, 203 (2001).

  37. [37]. Richard H. Pildes & G. Michael Parsons, The Legality of Ranked-Choice Voting, 109 Calif. L. Rev. (forthcoming Oct. 2021) (manuscript at 114–15),
    papers.cfm?abstract_id=3563257 [].

  38. [38]. Stephen Calabrese, Multimember District Congressional Elections, 25 Legis. Stud. Q. 611, 612 (2000) (distinguishing between four methods: “(1) Single-Member District Elections”; “(2) General-Ticket Elections”; “(3) Plural-District Elections”; and “(4) At-Large Elections”).

  39. [39]. 2 U.S.C. § 2c (2018). When a state is apportioned only one representative, the entire state serves as a single-member district. The states that received one representative in the 2010 apportionment are: Alaska, Delaware, Montana, North Dakota, South Dakota, Vermont, and Wyoming. 2010 Census Apportionment Results, at tbl. 1, U.S. Census Bureau (Dec. 2010), https:// []. The Elections Clause gives states primary authority, but Congress ultimate authority, to regulate “[t]he Times, Place, and Manner of” federal legislative elections. U.S. Const. art. I, § 4. For a helpful summary of historical changes in federal statutory requirements for House elections, see Micah Altman, Traditional Districting Principles: Judicial Myths vs. Reality, 22 Soc. Sci. Hist. 159, 167 tbl. 2 (1998).

  40. [40]. The original Federal Constitution provided for appointment of U.S. senators by state legislatures. But state legislatures increasingly made these appointments on the basis of a state preference election, and the Seventeenth Amendment, ratified in 1913, mandated direct election of U.S. senators. U.S. Const. art. I, § 3, amended by U.S. Const. amend. XVII.

  41. [41]. In Vermont the House uses ; the Senate uses . Vt. Const. ch. II, § 13; Id. ch. II, § 18; Vt. Stat. Ann. tit. 17, § 1881 (2021). In West Virginia the House uses ; the Senate uses . W. Va. Const. art VI, § 4; Id. art. VI, § 8; W. Va. Code § 1-2-2 (2011). In New Hampshire the House uses ; the Senate uses . N.H. Const. pt. II, art. 9; Id. pt. II, art. 26; N.H. Rev. Stat. Ann. § 662:5 (2012). In Maryland the House uses ; the Senate uses . Md. Const. art. III, § 3. In Arizona, Idaho, New Jersey, North Dakota, South Dakota, and Washington the House uses ; the Senate uses . Ariz. Const. art. IV, pt. 2, § 1; N.J. Const. art. IV, § 2; N.D. Const. art. IV, § 2; Idaho Const. art. III, § 5; S.D. Const. art. III, § 5; Wash. Const. art II, §43; Wash Rev. Code Ann. § 44.05.090 (2019).

  42. [42]. See State Legislative Chambers That Use Multi-Member Districts, Ballotpedia, https:// [
    64-WPRT] (last updated Aug. 10, 2021).

  43. [43]. Maine and Alaska have recently adopted ranked choice voting, which I discuss in greater detail infra Section II.C.

  44. [44]. See Runoff Election, Ballotpedia, [https://] (last updated Aug. 10, 2021). Different voting rules are possible with multi-member districts: (1) under bloc MMP each voter gets one vote for each open seat, she can only vote once for a particular candidate, and she must use all her votes; (2) under bloc with partial abstention (“BPA”) MMD each voter gets one vote for each open seat, she can only vote once for a particular candidate, and she chooses whether or not to use all her votes; (3) under staggered MMD, two legislators represent a single district but elections take place in different years; and (4) in post MMD, each seat is assigned a numbered post and candidates run for a specific post. For lower chamber elections, each of the ten states that use MMDs use BPA, post, or a combination of the two; for upper chamber elections, the two states that use MMDs use bloc or staggered. For lower chambers: Arizona, New Hampshire, New Jersey, North Dakota, Vermont, and West Virginia use BPA; Idaho and Washington use post; and Maryland and South Dakota use a combination of BPA and post. For upper chambers: Vermont uses bloc and West Virginia uses staggered. See State Legislative Chambers That Use Multi-Member Districts, supra note 42.

  45. [45]. As I discuss in Section II.B, infra, while all states draw electoral maps, states vary significantly in who draws the map, by what process, and according to what substantive criteria. Another area of significant intra-schema interstate variation is ballot access rules. Generally, states run primary elections using the same electoral schema, but there are important differences in terms of who participates. See Ballot Access for Major and Minor Party Candidates, Ballotpedia, [
    AU6N-8YZT] (last updated Aug. 10, 2021); State Primary Election Types, Nat’l Conf. State Legislatures (Jan. 5, 2021), [].

  46. [46]. Once the system determines how many seats each party gets, each party must determine which members get its seats. There are many ways to do this, but the simplest is a so-called closed-list system where each party predetermines a ranked list of members, and the party’s seats are allocated to members according to that list until all the party’s seats are filled. See, e.g., Dominik Hangartner, Nelson A. Ruiz & Janne Tukiainen, Open or Closed? How List Type Affects Electoral Performance, Candidate Selection, and Campaign Effort 2 (VATT Inst. for Econ. Rsch., Working Paper No. 120, 2019), [

  47. [47]. Generally, it may not be possible to achieve perfect vote-seat proportionality for every party because there are a finite number of seats which must be allocated in integer, rather than fractional, form. For this reason, it is necessary to choose an allocation method that approximates proportionality for every party. There are multiple allocation methods to choose from, including “highest average” methods (like D’Hondt and Sainte-Lagu) and “largest remainder” methods (like Hare and Droop), and a robust literature comparing these alternatives. See, e.g., Michael Gallagher, Comparing Proportional Representation Electoral Systems: Quotas, Thresholds, Paradoxes and Majorities, 22 British J. Pol. Sci. 469, 470 (1992); Juraj Medzihorsky, Rethinking the D’Hondt Method, 1 Pol. Rsch. Exch. 1, 1–2 (2019); Friedrich Pukelsheim, Proportional Representation: Apportionment Methods and Their Applications 71–103 (2017). Note that Congress uses such an allocation method when dividing 435 seats among 50 states proportional to state population. Computing Apportionment, U.S. Census Bureau (Mar. 1, 2021), https:// [https://].

  48. [48]. Benjamin Plener Cover & David Niven, Geographic Gerrymandering, 15 Harv. L & Pol’y Rev. (forthcoming 2021) (on file with author); Nicholas O. Stephanopoulos, Redistricting and the Territorial Community, 160 U. Pa. L. Rev. 1379, 1390–97 (2012); James A. Gardner, What is “Fair” Partisan Representation, and How Can It Be Constitutionalized? The Case for a Return to Fixed Election Districts, 90 Marq. L. Rev. 555, 576–82 (2007); Hilary Pearse, Geographic Representation and Electoral Reform, 28 Canadian Parliamentary Rev. 26, 28 (2005); Josep M. Colomer, Personal and Party Representation, in Personal Representation: The Neglected Dimension of Electoral Systems 1–20 (Josep M. Colomer ed., 2011).

  49. [49]. The Federalist No. 56, at 379 (James Madison) (Benjamin Fletcher Wright ed., 1961).

  50. [50]. David Niven, Benjamin Plener Cover & Michael Solimine, Are Individuals Harmed by Gerrymandering? Examining Access to Congressional District Offices, 102 Soc. Sci. Q. 29, 40–43 (2021).

  51. [51]. Stephanopoulos, supra note 48, at 1392–93 (under “theory of communal representation,” when electoral districts align with meaningful territorial communities, “[v]oters should be less confused and more politically engaged” and “it should be relatively straight-forward for elected officials to identify and advance their districts’ interests.”).

  52. [52]. Cover & Niven, supra note 48; see also Matthew J. Parlow, Civil Republicanism, Public Choice Theory, and Neighborhood Councils: A New Model for Civic Engagement, 79 U. Colo. L. Rev. 137, 154 (2008) (“Community stakeholders are given the opportunity to confront one another ... and transform themselves, their preferences, their intentions, and the community by searching for commonly held values, generating those public values, and agreeing upon the common good.”).

  53. [53]. Bruce Cain, John Ferejohn & Morris Fiorina, The Personal Vote: Constituency Service and Electoral Independence 9 (1987); see also Tom W. Rice & Alisa A. Macht, Friends and Neighbors Voting in Statewide General Elections, 31 Am. J. Pol. Sci. 448, 448 (1987) (“Gubernatorial and senatorial general election candidates from 46 states over the 1976 to 1982 period were examined and results indicate the average candidate polled 3.66 percentage points more of the general election vote in his or her ‘home county’ than another candidate from the same party but different county could have expected to garner.”).

  54. [54]. Mark C. Ellickson & Donald E. Whistler, Explaining State Legislators’ Casework and Public Resource Allocations, 54 Pol. Rsch. Q. 553, 560–61 (2001).

  55. [55]. Samuel Issacharoff & Richard H. Pildes, Politics as Markets: Partisan Lockups of the Democratic Process,50 Stan. L. Rev. 643, 680 (1998) (“The FPTP system virtually ensures continuing two-party dominance.”).

  56. [56]. Maurice Duverger, Political Parties: Their Organization and Activity in the Modern State 217 (Barbara North & Robert North trans., Science ed. 1963) (1954) (emphasis omitted).

  57. [57]. Id. at 239 (emphasis omitted).

  58. [58]. See generally William Roberts Clark & Matt Golder, Rehabilitating Duverger’s Theory: Testing the Mechanical and Strategic Modifying Effects of Electoral Laws, 39 Compar. Pol. Stud. 679 (2006) (reviewing Duverger’s theory to find that social forces impact the number of political parties a country will have).

  59. [59]. Perhaps the most famous American example occurred in 2000, when George W. Bush won Florida—and the presidency—by 537 votes, while 97,488 Floridians voted for Ralph Nader. See Michael C. Herron & Jeffrey B. Lewis, Did Ralph Nader Spoil Al Gore’s Presidential Bid?: A Ballot-Level Study of Green and Reform Party Voters in the 2000 Presidential Election, 2 Q.J. Pol. Sci. 205, 206 (2007). In the 2016 presidential election, Jill Stein (Green Party) and Gary Johnson (Libertarian) played similar roles in some swing states. Pildes & Parsons, supra note 37 (manuscript at 116).

  60. [60]. Party Division, U.S. Senate, [https://] (last updated Aug. 10, 2021); Party Breakdown, U.S. House Representatives Press Gallery, [] (last updated Aug. 10, 2021).

  61. [61]. Partisan Composition of State Legislatures, Ballotpedia,
    Partisan_composition_of_state_legislatures [] (last updated Aug. 10, 2021). Specifically, state lower chambers have 11 third party members: the Vermont house has 7 members of the Vermont Progressive Party, Maine and Wyoming each have one Libertarian, New York has a single house member from the Independence Party and Maine has one member of the Independent for Maine Party. Id. State lower chambers have 20 independents: Vermont has 5; Maine and Alaska each have 3; Louisiana has 3; Mississippi has 2; California, Massachusetts, New Mexico, and Wyoming each have 1. Id. Alaska has one nonpartisan member. Id.

  62. [62]. Id. The only upper chamber with third-party members is the Vermont senate, which has two members of the Vermont Progressive Party. Id. The Minnesota senate has two independents, while the Arkansas senate, the Oregon senate, and the Pennsylvania senate each have one independent. Id.

  63. [63]. Lanny W. Martin & Georg Vanberg, Parties and Policymaking in Multiparty Governments: The Legislative Median, Ministerial Autonomy, and the Coalition Compromise, 58 Am. J. Pol. Sci. 979, 980 (2014) (“Coalition governments, which are the norm in parliamentary systems operating under proportional representation electoral rules, must confront a wrinkle in policymaking that is absent under single-party government: Policy is made jointly by parties that are separately accountable at election time.” (emphasis omitted)); Dimiter Toshkov, Lars Mäder & Anne Rasmussen, Party Government and Policy Responsiveness. Evidence from Three Parliamentary Democracies, 40 J. Pub. Pol’y 329, 335 (2020) (“While single-party majority cabinets are common in the UK, multiparty coalitions are typical in Germany and in Denmark, where one also observes the phenomenon of multi-party minority coalition cabinets.” (emphasis omitted)).

  64. [64]. A Guide to Dutch Political Parties, Expatica (June 9, 2021),
    nl/living/gov-law-admin/dutch-political-parties-108098 [].

  65. [65]. Government: Political Parties in Israel, Isr. Sci. & Tech. Directory, [] (last updated Aug. 10, 2021).

  66. [66]. Lee Drutman,Breaking the Two-Party Doom Loop: The Case for Multiparty Democracy in America 206–39 (2020); Christopher Ingraham, How to Fix Democracy: Move Beyond the Two-Party System, Experts Say, Wash. Post (Mar. 1, 2021, 3:14 PM), https://www.washington []; Santucci, supra note 8, at 35–36; Richard L. Hasen, Entrenching the Duopoly: Why the Supreme Court Should Not Allow the States to Protect the Democrats and Republicans from Political Competition, 1997 Sup. Ct. Rev. 331, 333–35.

  67. [67]. Rivka Weill, On the Nexus of Eternity Clauses, Proportional Representation, and Banned Political Parties,16 Election L.J.237, 240 (2017) (“[PR] elections encourage extremist—and even secessionist—political parties to compete against weak and unstable governments, and foster them with political and economic resources to advance their agendas.”); Barry Eichengreen & David Leblang, Exchange Rates and Cohesion: Historical Perspectives and Political‐Economy Considerations,41 J. Common Mkt. Stud.797, 805 (2003) (“Proportional representation (PR) may lead to fragmented party systems and unstable governing coalitions.”); André Blais & Marc André Bodet, Does Proportional Representation Foster Closer Congruence Between Citizens and Policy Makers?,39 Compar. Pol. Stud.1243, 1246 (2006) (“[In PR systems], extremist parties can more easily win votes ... [because] parties or voters may fail to coordinate[] and these coordination failures may allow an extremist government to form.”); Pippa Norris, Radical Right: Voters and Parties in the Electoral Market 114 (2005) (“[D]espite having roughly the same share of the vote, radical right parties were more than twice as successful in gaining seats under PR as under majoritarian elections.” (emphasis omitted)).

  68. [68]. Benjamin Plener Cover, Quantifying Partisan Gerrymandering: An Evaluation of the Efficiency Gap Proposal,70 Stan. L. Rev.1131, 1222–32 (2018).

  69. [69]. Legislative bodies generally impose supermajority requirements for significant actions, such as overriding a gubernatorial veto, amending the state constitution, passing the budget, raises taxes. These supermajority requirements reflect the principle that a bare majority confers a more limited governing mandate than a substantial majority. Majority dominance subverts this principle by conferring a supermajority on a party that lacks supermajority support.

  70. [70]. How sensitive are changes in a party’s seat-share to changes in that party’s vote-share? In other words, if a party boosts its vote share by a given increment, will its seat share increase a lot, a little, or none at all? With high responsiveness, seat share increases a lot; with low responsiveness, seat share increases a little; with no responsiveness, seat share doesn’t budge. Technically, responsiveness is denoted and defined as the slope of the vote-seat curve at the point of the electoral outcome, i.e., .

  71. . How competitive is each district race, i.e., how uncertain is its outcome? District competitiveness is generally measured by the margin of victory (the vote difference between the top two candidates) and the district is classified as competitive or safe depending on whether that margin falls below or above some specified threshold (like five or ten percent).

  72. . How many districts feature a single candidate running unopposed, or a single major-party candidate opposed only by independent or third-party candidates with no realistic prospect of winning?

  73. . What percentage of incumbents win reelection?

  74. . What percentage of a district’s eligible voters actually participate in the district election?

  75. [75]. Do people believe that politicians respond to constituents, that elected officials care about average people, that they can meaningfully participate in the democratic process?

  76. [76]. André Blais & Ignacio Lago, A General Measure of District Competitiveness, 28 Electoral Stud. 94, 94 (2009) (“[P]eople are more likely to feel that their vote counts when an election is close, and such feeling fosters turnout.”).

  77. [77]. Wisconsin State Assembly Elections, 2018, Ballotpedia [hereinafter Ballotpedia, Wisconsin State Assembly Elections],,_2018 [] (last updated Aug. 10, 2021); Schmidt, supra note 7.

  78. [78]. Ballotpedia, Wisconsin State Assembly Elections, supra note 77.

  79. [79]. Id.

  80. [80]. Id.

  81. [81]. Id.

  82. [82]. Id.

  83. [83]. Id.

  84. [84]. See generally Nicholas O. Stephanopoulos & Christopher Warshaw, The Impact of Partisan Gerrymandering on Political Parties, 45 Legis. Stud. Q. 609 (2020) (presenting empirical evidence of partisan gerrymandering’s adverse effects: candidates are less likely to contest districts; those that do are weaker; donors are less willing to give money; and ordinary voters are less likely to support the disfavored party); Gill v. Whitford, 138 S. Ct. 1916, 1938 (2018) (Kagan, J., concurring) (“Members of the ‘disfavored party’ in the State, deprived of their natural political strength by a partisan gerrymander, may face difficulties fundraising, registering voters, attracting volunteers, generating support from independents, and recruiting candidates to run for office ... .” (citations omitted)).

  85. [85]. James A. Gardner, Illiberalism and Authoritarianism in the American States, 70 Am. U. L. Rev. 829, 883 (2021) (citations omitted); see also Carolyn Shapiro, Democracy, Federalism, and the Guarantee Clause, 62 Ariz. L. Rev. 183, 187 (2020) (pointing out that similar conduct occurred in Michigan and North Carolina).

  86. [86]. Rucho v. Common Cause, 139 S. Ct. 2484, 2498–508 (2019).

  87. [87]. Benjamin Plener Cover, Rucho for Minimalists, 71 Mercer L. Rev. 695, 704–18 (2020); Cover & Niven, supra note 48.

  88. [88]. A partisan gerrymander could still be challenged in federal court based on claims of minority vote dilution, malapportionment, or racial gerrymandering, but not based on claims of partisan gerrymandering. See, e.g., Perez v. Abbott, 250 F. Supp. 3d 123, 143–80 (W.D. Tex. 2017) (minority vote dilution); Davis v. Abbott, 781 F.3d 207, 215–18 (5th Cir. 2015) (malapportionment); Ala. Legis. Black Caucus v. Alabama, 135 S. Ct. 1257, 1268–74 (2015) (racial gerrymandering).

  89. [89]. Rucho, 139 S. Ct. at 2515–16 (Kagan, J., dissenting) (“I’ll give the majority this one
    —and important—thing: It identifies some dangers everyone should want to avoid. Judges should not be apportioning political power based on their own vision of electoral fairness, whether proportional representation or any other. And judges should not be striking down maps left, right, and center, on the view that every smidgen of politics is a smidgen too much. Respect for state legislative processes—and restraint in the exercise of judicial authority—counsels intervention in only egregious cases.”).

  90. [90]. Common Cause v. Lewis, No. 18 CVS 014001, 2019 WL 4569584, at *108–12 (N.C. Super. Ct. Sept. 3, 2019); see also League of Women Voters of Pa. v. Commonwealth, 178 A.3d 737, 803, 814–21 (Pa. 2018) (striking down Pennsylvania’s congressional map for violating guarantees in Pennsylvania Constitution Article I, Section 5 that “[e]lections shall be free and equal; and no power ... shall at any time interfere to prevent the free exercise of the right of suffrage” (quoting Pa. Const. art. 1, § 5)); League of Women Voters of Fla. v. Detzner, 172 So. 3d 363, 375, 401–13 (Fla. 2015) (striking down Florida’s congressional map for violating the state’s “Free Districts Amendment,” which provides that no districting plan “shall be drawn with the intent to favor or disfavor a political party,” (quoting Fla. Const. art. III, § 20(a))).

  91. [91]. Rucho, 139 S. Ct. at 2507 (“The expansion of judicial authority would not be into just any area of controversy, but into one of the most intensely partisan aspects of American political life.”); id. at 2498–99 (“‘With uncertain limits, intervening courts ... would risk assuming political, not legal, responsibility for a process that often produces ill will and distrust.’” (quoting Vieth v. Jubelirer, 541 U.S. 267, 307 (2003) (Kennedy, J., concurring))).

  92. [92]. Justin Levitt & Michael P. McDonald, Taking the “Re” Out of Redistricting: State Constitutional Provisions on Redistricting Timing, 95 Geo. L.J. 1247, 1247 (2007) (“Supreme Court rulings of the 1960s required legislative districts to be of equal population, and thus redistricting became a decennial obligation for the states following the release of new federal census population data.”).

  93. [93]. League of United Lat. Am. Citizens v. Perry, 548 U.S. 399, 415 (2006) (“With respect to a mid-decade redistricting to change [congressional] districts drawn earlier in conformance with a decennial census, the Constitution ... state[s] no explicit prohibition.”); id. at 457 (Stevens, J., concurring in part and dissenting in part) (“[T]he Constitution places no per se ban on mid-cycle redistricting ... .” ).

  94. [94]. Stephanopoulos, supra note 8, at 338.

  95. [95]. Matt Vasilogambros, The Tumultuous Life of an Independent Redistricting Commissioner, Pew (Nov. 26, 2019), []; Andrew C. Maxfield, Comment, Litigating the Line Drawers: Why Courts Should Apply Anderson-Burdick to Redistricting Commissions, 87 U. Chi. L. Rev. 1845, 1847 (2020); Harris v. Ariz. Indep. Redistricting Comm’n, 136 S. Ct. 1301, 1305–06 (2016).

  96. [96]. Rudi Keller, Voters Repeal Clean Missouri Redistricting Plan They Enacted in 2018, Columbia Daily Trib. (Nov. 4, 2020, 4:05 PM),
    58002 [].

  97. [97]. U.S. Const. art. I, § 4.

  98. [98]. Ariz. State Legis. v. Ariz. Indep. Redistricting Comm’n, 576 U.S. 787, 826 (2015) (Roberts, C.J., dissenting).

  99. [99]. See Michael T. Morley, The Independent State Legislature Doctrine, Federal Elections, and State Constitutions,55 Ga. L. Rev.1, 90–93 (2020); Nicholas Stephanopoulos, The Anti-Carolene Court, 2019 Sup. Ct. Rev. 111, 152–53.

  100. [100]. Moon Duchin et al., Locating the Representational Baseline: Republicans in Massachusetts, 18 Election L.J. 388, 388 (2019).

  101. [101]. Vieth v. Jubelirer, 541 U.S. 267, 289 (2004).

  102. [102]. Seifter, supra note 9 (manuscript at 26–32). This statistic refers only to minority-majority vote-seat inversions, where a party wins a majority of seats with a strict minority (i.e., not even a plurality) of votes. But support for third party or independent candidates could produce a plurality-majority vote-seat inversion, where a party wins a majority of seats with a plurality, but not a majority, of votes. Over the same period, plurality-majority vote-seat inversion has occurred 68 times, 35 for state upper chambers and 33 for state lower chambers. Thus, vote-seat inversion of either the minority-majority or plurality-majority variety has occurred 335 times, at least once in 40 states, 181 for state upper chambers and 154 for state lower chambers. Id.

  103. [103]. See, e.g., Peter Selb, A Deeper Look at the Proportionality-Turnout Nexus, 42 Compar. Pol. Stud. 527, 527 (2009) (“Evidence that turnout is higher under proportional representation (PR) than in majoritarian elections is overwhelming ... it is argued that majoritarian electoral systems tend to produce safe seats and that voters have little incentive to turn out there. Thus, uneven turnout over electoral districts due to variable intensities of local competition is made responsible for the lower overall turnout... . This article scrutinizes the relationship between electoral rules, competition, and turnout with district-level data from 31 national elections. Results from a heteroscedastic model indicate that lower net turnout in majoritarian systems is indeed a consequence of uneven turnout over districts due to variable levels of local competitiveness.”).

  104. [104]. Jonathan S. Gould, The Law of Legislative Representation, 107 Va. L. Rev. 765, 795 (2021) (“The need to win primaries leads legislative candidates to position themselves closer to more extreme primary voters, rather than to more moderate general-election voters. The result is that Republican legislators are well to their median constituent’s right and Democratic legislators are well to their median constituent’s left.”footnotes omitted); David W. Brady, Hahrie Han & Jeremy C. Pope, Primary Elections and Candidate Ideology: Out of Step With the Primary Electorate?, 32 Legis. Stud. Q. 79, 82 (2007); Richard H. Pildes, Why the Center Does Not Hold: The Causes of Hyperpolarized Democracy in America, 99 Calif. L. Rev. 273, 308 (2011).

  105. [105]. John F. Nagle & Alec Ramsay, On Measuring Two-Party Partisan Bias in Unbalanced States,20 Election L.J. 116, 117 (2021) (“If there is such systemic bias, there may still be plans that are outliers in the ensemble that are nonetheless fair, and we believe a redistricting commission should adopt such a plan rather than an average plan.”).

  106. [106]. See supra Section II.A.4.

  107. [107]. Fair Representation Act, H.R. 4000, 116th Cong. (2019).

  108. [108]. Paul Diller, Making State Democracy More Representative (draft manuscript on file with author).

  109. [109]. Jack Santucci, A Modest and Timely Proposal, (Dec. 9, 2020), https:// []; Matthew S. Shugart, Emergency Electoral Reform: OLPR for the US House, Fruits & Votes (Jan. 19, 2021),

    the-us-house [ ].
  110. [110]. For example, assume there are five district seats, and five candidates run from each major party. Suppose Republican candidates collectively earn 60 percent of the votes cast while Democratic candidates collectively earn the remaining 40 percent. The five seats would be allocated among the parties proportional to their relative vote share, so Republicans would get three seats and Democrats would get the other two. The three Republican seats would go to the three Republican candidates who earned the most votes out of all the Republican candidates. The two Democratic seats would go to the two Democratic candidates who earned the most votes out of all the Democratic candidates.

  111. [111]. Shugart, supra note 109; see Santucci, supra note 109.

  112. [112]. See generally Matthew Soberg Shugart & Martin P. Wattenberg, Mixed-Member Electoral Systems: The Best of Both Worlds? (2003) (analyzing major countries that have adopted mixed-member systems).

  113. [113]. See generally Shaun Bowler, David M. Farrell & Robin T. Pettitt, Expert Opinion on Electoral Systems: So Which Electoral System is “Best”?, 15J. Elections, Pub. Op. & Parties3 (2005) (discussing responses to an electoral system survey).

  114. [114]. See generally FairVote, The Fair Representation Act: Remedying Dysfunction in the U.S. House of Representatives (2017) [hereinafter FairVote, The Fair Representation Act], [] (analyzing the Fair Representation Act’s impact on competition and partisan fairness).

  115. [115]. Id. at 7.

  116. [116]. Id. at 20.

  117. [117]. Drutman, supra note 66, at 206–39.

  118. [118]. FairVote, Comparative Structural Reform: Assessing the Impact of 37 Structural Reforms 4 (2016), [https://].

  119. [119]. Id. at 13.

  120. [120]. Shugart, supra note 109 (“Basically, the point is that there are (at least)two ‘rights’ and two ‘lefts’but currently only one party on the right and one on the left... . The need for PR is to let the free-market small-d democrats in the currently existing parties act independently of their more extreme wings. This is precisely what PR systems permit–each side’s extreme can be its own party rather than a wing of one majority-seeking party, without raising concerns over ‘spoilers’ that arise under plurality elections... . I also am not going to go into the details of how actual coalitions would work under this stylized latent four-party system that PR would allow to break forth.” (footnotes omitted)); Ingraham, supra note 66 (“But an interesting thing about multi-member districts, according to political scientists, is that they could usher in the end of two-party politics... . Multi-member districts would, almost by definition, fracture the Democratic and Republican parties.”).

  121. [121]. What About “Fringe Parties”?, Fair Vote Can.,
    fringeparties [] (last updated Aug. 17, 2021).

  122. [122]. Santucci, supra note 109 (“Get over the allergy to third-party politics. Take a serious look at what states and districts are likely to produce third-party winners. Others may disagree with me, but Isuspectgains would be isolated and modest.”).

  123. [123]. In some cases, reform opponents may use a literal script. To take but one illustrative example, consider the media campaign of the No BC Proportional Representation Society, the official opponent group for the electoral reform proposal on the 2018 British Columbia referendum ballot. About Us, Vote No To Pro Rep, [https://] (last updated Aug. 17, 2021). According to the group’s founder, “[t]he most prominent ad showed scenes from Europe talking about the problems of extremism under proportional representation, ending with marching soldiers and the sound of their boots – and it hit a major nerve.” Bill Tieleman, How the No Side Surged From Behind to Defeat Proportional Representation, Vancouver Sun (Feb. 21, 2019), []. The ad is available at: Vancouver Sun, Proportional Representation: Pros and Cons of Reforming B.C.’s Electoral System, YouTube (Oct. 22, 2018), [].

  124. [124]. See generally F.A. Hermens, Democracy or Anarchy? A Study of Proportional Representation (1941) (discussing how the proportional representation helped the Nazi party gain power).

  125. [125]. Serhij Vasylchenko, The Negative Consequences of Proportional Representation in Ukraine, 21 Demokratizatsiya 425, 447–48 (2013).

  126. [126]. Jeffrey Heller, Netanyahu Alleges Israeli Election Fraud, Accuses Rival of Duplicity, Reuters (June 6, 2021, 8:02 AM), [].

  127. [127]. Quentin L. Quade, Debate—Proportional Representation: PR and Democratic Statecraft,2 J. Democracy 36, 36–38 (1991) (“[T]he obvious and dramatic cases in which PR clearly contributed to governmental weakness and systemic collapse[:] Pre-Mussolini Italy, with its splintered parties and political gridlock, would be a worthy example. France’s Fourth Republic (1945-1958), chronically crippled and finally made suicidal by its inability to deal with colonial and domestic problems, would be another. Finally, the Weimar Republic, where coalition was endemic and weakness perpetual, might be the best of all test cases for PR.”); Bill Tieleman, No BC Proportional Representation Society Brings Together Political Opponents to Fight Pro-Rep in BC Fall Referendum, (Jan. 14, 2018, 5:41 PM),
    2018/01/no-bc-proportional-representation.html [] (“We have our work cut out for us to defeat proportional representation and not end up like Italy, Israel, Austria or other countries where the electoral system promotes extremists and damages democracy.”).

  128. [128]. After withering attacks by Senate Republicans, President Bill Clinton withdrew his nomination of Lani Guinier to lead the civil rights division of the Department of Justice, citing her scholarship critiquing the two-party system and advocating proportional representation. Text of President Clinton’s Comments on Withdrawal of Guinier Nomination, Wash. Post (June 4, 1993), [https://]; see also 52 U.S.C. § 10301(b) (2018) (“Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.” (emphasis omitted)). See generally Sanford Levinson, Gerrymandering and the Brooding Omnipresence of Proportional Representation: Why Won’t It Go Away,33 UCLA L. Rev.257 (1985) (analyzing the merits and demerits of proportional representation).

  129. [129]. See supra Section II.C.

  130. [130]. A three-judge federal panel struck down this map as a partisan gerrymander, but the Supreme Court reversed on standing grounds in Gill v. Whitford, 138 S. Ct. 1916, 1934 (2018).

  131. [131]. Schmidt, supra note 7.

  132. [132]. See infra Section III.A.3.

  133. [133]. If the state decreases the size of the nominal class, it must draw a new electoral map with a smaller number of larger single-seat electoral districts. If the state preserves the size of the nominal class and expands the body, it also preserves the number of single-seat electoral districts. With this latter approach, the state could even adopt MM2 while keeping the same electoral map it used pre-adoption.

  134. [134]. Rucho v. Common Cause, 139 S. Ct. 2484, 2506 (2019) (quoting Ariz. State Legis. v. Ariz. Indep. Redistricting Comm’n, 576 U.S. 787, 791 (2015)).

  135. [135]. Id. at 2500.

  136. [136]. Id. (citing Zivotofsky v. Clinton, 566 U.S. 189, 196 (2012)).

  137. [137]. U.S. Const. art. 1, § 4.

  138. [138]. Id. art. 4, § 4. There is a serious argument that the Republican guarantee prohibits AES, with its vulnerability to partisan gerrymandering, and its tendency to invert the relationship between votes and seats (and between voters and politicians). There is no serious argument that the Republican guarantee requires AES.

  139. [139]. Rucho, 139 S. Ct. at 2507; see also Voinovich v. Quilter, 507 U.S. 146, 1156 (1993) (“[T]he federal courts may not order the creation of majority-minority districts unless necessary to remedy a violation of federal law. But that does not mean that the State’s powers are similarly limited.” (citation omitted)).

  140. [140]. Rucho, 139 S. Ct. at 2499 (“Partisan gerrymandering claims rest on an instinct that groups with a certain level of political support should enjoy a commensurate level of political power and influence... . But such a claim is based on a ‘norm that does not exist’ in our electoral system—‘statewide elections for representatives along party lines.’” (quoting Davis v. Bandemer, 478 U.S. 109, 159 (1986) (O’Connor, J., concurring))).

  141. [141]. See, e.g., Bernard Grofman & Gary King, The Future of Partisan Symmetry as a Judicial Test for Partisan Gerrymandering After LULAC v. Perry,6 Election L.J.2, 3 (2007) (offering a “social science perspective” on partisan gerrymandering).

  142. [142]. Timmons v. Twin Cities Area New Party, 520 U.S. 351, 369–70 (1997).

  143. [143]. Holder v. Hall, 512 U.S. 874, 881 (1994); see Richard Briffault, Race and Representation After Miller v. Johnson, 1995 U. Chi. Legal F. 23, 65 (“[T]he Supreme Court has held that the size of an elected body cannot be treated as a standard, practice, or procedure dilutive of voting rights under Section 2.”); see also Gardner, supra note 48, at 590 (expressing doubt “that Section 2 would even apply to a state’s choice to adopt a representation system of fixed districts where” the size of the legislature varies each election cycle).

  144. [144]. Chen & Rodden, supra note 15, at 241.

  145. [145]. Rucho, 139 S. Ct. at 2500.

  146. [146]. Samuel Issacharoff & Pamela S. Karlan, Where to Draw the Line?: Judicial Review of Political Gerrymanders, 153 U. Pa. L. Rev. 541, 551 (2004); Eric McGhee, The Role of Partisan Gerrymandering in U.S. Elections 3 (Aug. 25, 2017) (unpublished manuscript),
    sol3/papers.cfm?abstract_id=3341696 [] (“The logic of packing and cracking dictates that a party should win its seats by the narrowest acceptable margins to avoid wasting voters that could be used to win another seat elsewhere. But incumbents would naturally prefer to win by broader margins as protection against adverse partisan tides.”).

  147. [147]. See Cover, supra note 68, at 1212–13.

  148. [148]. See supra Section II.A.4.

  149. [149]. See generally Gerken, Getting from Here to There in Election Reform, supra note 33 (discussing the problem of election reform and how to fix it as a “here-to-there” problem).

  150. [150]. See, e.g., Idaho Const. art. III, § 2 (fixing the size of the state senate at 35 seats).

  151. [151]. For example, Illinois voters ended a century-old system of cumulative voting in the state House when it adopted the so-called “‘Cutback Amendment’ which was largely advertised as an effort to reduce the size of the legislature by one-third.” Rob Richie & Andrew Spencer, The Right Choice for Elections: How Choice Voting Will End Gerrymandering and Expand Minority Voting Rights, from City Councils to Congress, 47 U. Rich. L. Rev. 959, 1004 n.294 (2013).

  152. [152]. See supra Section III.A.2.

  153. [153]. Thornburg v. Gingles, 478 U.S. 30, 48–51 (1986); Bartlett v. Strickland, 556 U.S. 1, 12 (2009). The list tier may also promote racial and gender diversity in the legislative chamber. See, e.g., Fiona Barker & Hilde Coffé, Representing Diversity in Mixed Electoral Systems: The Case of New Zealand, 71 Parliamentary Affs. 603, 603 (2018); Sona N. Golder, Laura B. Stephenson, Karine Van der Straeten, André Blais, Damien Bol, Philipp Harfst & Jean-François Laslier, Votes for Women: Electoral Systems and Support for Female Candidates, 13Pol. & Gender107, 108 (2017).

  154. [154]. MPs and Electorates, N.Z. Parliament, [] (last updated Sept. 12, 2021).

  155. [155]. See supra Section III.A.3.iii.

  156. [156]. Nagle & Ramsay, supra note 105, at 119.

  157. [157]. Shugart & Wattenberg, supra note 112, at 15 tbl.1.1.

  158. [158]. Id. at 14–17.

  159. [159]. Id. at 14 (“If seats are linked, the number of seats a party takes from its list is determined partially by the number of seats it has won from the nominal tier... . [A] formula is applied to the two tiers combined ... . The formula establishes a party’s total seat allocation; then list seats are allocated to bring the party’s representation up to that total.”).

  160. [160]. Id. (“The form of seat linkage that will most concern us here is the compensatory type, which produces a mixed-member proportional (MMP) system. For example, in Germany a party wins a percentage of seats nationally that is determined by applying a PR formula to the total number of parliamentary seats. Each party then fulfills is entitlement of seats by taking whatever SSDs [(single-seat districts)] it has won and augmenting them with the number of candidates off the list that it requires to reach its overall entitlement.” (footnotes omitted)).

  161. [161]. Id. at 15 (“[I]f votes are linked, then the votes that are used to allocate list-tier seats are not solely the votes that are cast for party lists, because those votes are adjusted by the transfer of votes from the nominal tier.” (footnotes omitted)).

  162. [162]. Id. at 16–17 (“[N]o system employs vote linkage along with seat linkage. Such a system is plausible, however, and might works as follows... . No country has yet adopted such a system, but it would be both logical and feasible ... .”).

  163. [163]. Arolda Elbasani, Mixed Member Electoral Systems in Transition Contexts: How Has the System Worked in Albania?, 3 CEU Pol. Sci. J. 72, 86–88 (2008); Jorgen Elklit, The 2007 General Election in Lesotho: Abuse of the MMP System?, 7 J. African Elections 10, 16 (2009); Daniel Bochsler, A Quasi-Proportional Electoral System ‘Only for Honest Men’? The Hidden Potential for Manipulating Mixed Compensatory Electoral Systems, 33 Int’l Pol. Sci. Rev. 401, 405 (2012). See generally Eric Linhart, Johannes Raabe & Patrick Statsch, Mixed-Member Proportional Electoral Systems—The Best of Both Worlds?, 29 J. Elections, Pub. Op. & Parties 21 (2019) (discussing the MMP system in Albania, Bolivia, Germany, Lesotho, New Zealand and Venezuela).

  164. [164]. See supra Section II.A.1.

  165. [165]. Pildes & Parsons, supra note 37 (manuscript at 115–16).

  166. [166]. See note 59 and accompanying text.

  167. [167]. Like Bill Clinton winning the 1992 presidential elections when conservatives split their votes between Republican incumbent George H. W. Bush and independent challenger Ross Perot. See Christoffer Dunstan,The Systematic Exclusion of Third Parties in American Politics, 4L. &Soc’y J. U.C. Santa Barbara41, 55 (2004).

  168. [168]. Seifter, supra note 9 (manuscript at 28–29).

  169. [169]. Runoff Election, supra note 44.

  170. [170]. Pildes & Parsons, supra note 37 (manuscript at 118).

  171. [171]. As of June 2021, FairVote projects that 53 jurisdictions will use ranked-choice voting in their next elections. Where Ranked Choice Voting is Used, FairVote,
    where_is_ranked_choice_voting_used [] (last updated Aug. 19, 2021). This count includes two states (Maine and Alaska), one county, 23 cities and towns participating in a municipal pilot program authorized by the Utah legislature, and 26 cities outside of Utah. Id.

  172. [172]. Martha Kropf, Using Campaign Communications to Analyze Civility in Ranked Choice Voting Elections, 9 Pol. & Governance 280, 290 (2021); Eamon McGinn, Rating Rankings: Effect of Instant Run-Off Voting on Participation and Civility 14–16 (Oct. 8, 2020) (unpublished manuscript),
    _in_Minneapolis.pdf [].

  173. [173]. Sarah John, Haley Smith & Elizabeth Zack, The Alternative Vote: Do Changes in Single‐Member Voting Systems Affect Descriptive Representation of Women and Minorities?, 54 Electoral Stud. 90, 98–99 (2018).

  174. [174]. Joseph Anthony, Amy Fried, Robert Glover & David C. Kimball, Ranked Choice Voting in Maine from the Perspective of Local Election Officials,Election L.J. (forthcoming 2021); Craig M. Burnett & Vladimir Kogan, Ballot (and Voter) “Exhaustion” Under Instant Runoff Voting: An Examination of Four Ranked-Choice Elections,37 Electoral Stud.41, 47 (2015).

  175. [175]. See generally League of Women Voters of Mich. v. Benson, 373 F. Supp. 3d 867 (E.D. Mich. 2019), vacated sub nom.; Chatfield v. League of Women Voters of Mich., 140 S. Ct. 429 (2019) (providing an example of a case addressing partisan gerrymandering in Michigan). In 2018, Republicans won the Michigan House and Senate even though Democrats got more votes and a Democrat won the governorship. Jonathan Oosting, Why Democrats Won More Votes, but GOP Won More Legislative Seats in Michigan, Detroit News (Nov. 20, 2018, 5:30 AM), https://www. [].

  176. [176]. Mich. Const. art. II, § 9; id. art. XII, §§ 1–4.

  177. [177]. Paul Egan, Michigan Voters Approve Anti-Gerrymandering Proposal 2, Detroit Free Press (Nov. 7, 2018, 5:23 AM),
    proposal-2-michigan-gerrymandering/1847078002 [].

  178. [178]. Danielle M. Strollo, Note, Fair Reforms for Idaho’s Reapportionment Commission, 56Idaho L. Rev.295, 297 (2020).

  179. [179]. See generally Caroline Kane, Gianni Mascioli, Michael McGarry & Meira Nagel, Fordham U. Sch. L., Why the House of Representatives Must Be Expanded and How Today’s Congress Can Make It Happen (2020),
    downloads/id/14402/Why_the_House_Must_Be_Expanded___Democracy_Clinic.pdf [https://] (providing nine reasons as to why the House of Representatives must be expanded).

  180. [180]. See supra Section III.B.1.

  181. [181]. For the 2020 cycle, North Carolina will have 14 seats. See 2020 Census: Apportionment of the U.S. House of Representatives, U.S. Census Bureau (Apr. 26, 2021),
    library/visualizations/2021/dec/2020-apportionment-map.html [].

  182. [182]. North Carolina Representative David Lewis explicitly supported “draw[ing] the map[s] to give a partisan advantage to 10 Republicans and 3 Democrats because he [did] not believe it would be possible to draw a map with 11 Republicans and 2 Democrats.” Common Cause v. Rucho, 318 F. Supp. 3d 777, 808 (M.D.N.C. 2018) (alterations omitted), vacated, 139 S. Ct. 2484 (2019).

  183. [183]. Directory of Representatives, U.S House of Representatives,
    representatives [] (last updated Aug. 20, 2021) (Arizona (9); California (53); Florida (27); Georgia (14); Illinois (18); Indiana (9); Massachusetts (9); Michigan (14); New Jersey (12); New York (27); North Carolina (13); Ohio (16); Pennsylvania (18); Tennessee (9); Texas (36); Virginia (11); Washington (10)).


Associate Professor of Law at the University of Idaho College of Law (J.D., Yale Law School; M.Sc., London School of Economics).

I am grateful to the organizers of and participants in the 2021 AALS Election Law Works-in-Progress Workshop, including Gene Mazo, Travis Crum, Nicholas Stephanopoulos, Guy Uriel-Charles, Ned Foley, Lisa Manheim, and Rob Yablon. Thanks to Jack Santucci, Jack Nagel, and Matthew Shugart for their helpful feedback. Special thanks to Trevor Buller and Alex Siebert for excellent research assistance. As always, I am grateful to Aliza Plener Cover for her invaluable support.