Ranked-Choice Voting as Reprieve from the Court-Ordered Map
Thus far, legal debates about the rise of ranked-choice voting have centered on whether legislatures can lawfully adopt the practice. This Note turns attention to the courts and the question of remedies. It proposes that courts impose ranked-choice voting as a redistricting remedy. Ranked-choice voting allows courts to cure redistricting violations without also requiring that they draw copious numbers of districts, a process the Supreme Court has described as a “political thicket.” By keeping courts away from the fact-specific, often arbitrary judgments involved in redistricting, ranked-choice voting makes for the redistricting remedy that best protects the integrity of the judicial role.
American election law leaves the drawing of district maps—congressional districts, state legislative districts, school board districts, and so on—to legislatures.1See, e.g., Abrams v. Johnson, 521 U.S. 74, 101 (1997) (“The task of redistricting is best left to state legislatures, elected by the people and as capable as the courts, if not more so, in balancing the myriad factors and traditions in legitimate districting policies.”); Nathaniel Persily, When Judges Carve Democracies: A Primer on Court-Drawn Redistricting Plans, 73 Geo. Wash. L. Rev. 1131, 1133 (2005) (describing the “widespread recognition in the case law that redistricting is a political matter primarily and appropriately entrusted to the political branches”). Representative bodies, like independent redistricting commissions, can also redistrict; the Court regards these bodies as legislative, too, in the sense that they are the “power that makes laws.” Ariz. State Legislature v. Ariz. Indep. Redistricting Comm’n, 576 U.S. 787, 813 (2015). But sometimes the job of drawing a map falls into the hands of courts.2See infra Section I.B. This presents a dilemma. Drawing a map requires politically charged, policy-laden choices about where to place district lines, and courts are poorly suited for such a role. This Note argues that, in these circumstances, the best solution is for courts to impose a proportional voting method, such as multimember ranked-choice voting, that minimizes the need for a court to draw district lines in the first place.
Consider the case of Texas in the 2010s. During that decade’s redistricting cycle, Texas’s state legislature drew state house districts that diluted the votes of Black and Hispanic communities in violation of the Voting Rights Act (VRA).3See Perry v. Perez, 565 U.S. 388, 390–92 (2012) (per curiam); Texas v. United States, 887 F. Supp. 2d 133, 178 (D.D.C. 2012) (noting that “the incredible testimony of the lead House mapdrawer reinforc[ed] evidence suggesting mapdrawers cracked [districts] along racial lines to dilute minority voting power”), vacated, 570 U.S. 928 (2013). Federal courts barred the plan from going into effect.4Perez, 565 U.S. at 390–92. Texas, as a covered jurisdiction under section 5 of the Voting Rights Act, needed to earn preclearance before its plan could be implemented. Because it had not received preclearance, its plan could not go into effect. Id. But Texas did not respond with a valid replacement. The legislature sat on its hands, hoping the courts would give way.5See Perez v. Perry, 835 F. Supp. 2d 209, 211–212 (W.D. Tex. 2011), vacated per curiam, 565 U.S. 388 (2012). As November 2012 drew near, Texas had no map from which to hold elections. The court would have to draw districts for the State of Texas on its own.6Id.
The three-judge panel charged with redistricting regarded its job as “unwelcome,”7Id. Under 28 U.S.C. § 2284(a), three-judge panels hear challenges to the constitutionality of a redistricting plan. and it’s easy to see why. The court, in a time crunch, had to call on technical staff for assistance to parse complex demographic data in order to group Texas’s communities into more than one hundred districts.8Perez, 835 F. Supp. 2d at 212. The shape of these districts—the political constituencies they included and excluded—would determine whether incumbent politicians in the state house and Congress would keep their jobs.9See Persily, supra note 1, at 1146, 1150 (explaining that districting determines whether incumbents keep their jobs and highlighting the data required in a typical redistricting effort).
Not only were the stakes high; the legal tools at the court’s disposal were thin.10E.g., id. at 1157–64 (explaining that there are no “neutral” redistricting principles and enumerating the ambiguities of most redistricting principles); see also infra Section II.B.2. In redistricting, the available rules and norms quickly “run out.”11Justice Kagan used this formulation in the administrative-law context to underscore that sometimes a text yields no clear answers—the “law runs out”—forcing an agency to make “policy-laden” choices about how to fill in the gap. Kisor v. Wilkie, 139 S. Ct. 2400, 2415 (2019). For a description of districting principles applicable by state, see Districting Principles for 2010 and Beyond, Nat’l Conf. St. Legislatures (Jan. 4, 2021), https://www.ncsl.org/research/redistricting/districting-principles-for-2010-and-beyond.aspx [https://perma.cc/6CD3-RQ9A]. There is basic agreement among the states that districts should be contiguous, compact, and preserving of political subdivisions. About half the states explicitly instruct map drawers to preserve communities of interest. But, as Section II.B.2 suggests, these rules leave many questions unanswered. The residual decision space has to be filled by the court’s policy judgment and discretion. For instance, on the existing but invalid map, District 26 of the Texas State House was a strange-looking gerrymander in the Houston suburbs designed to suppress minority votes.12Perez, 835 F. Supp. 2d at 216. For better images of the districts, see District Viewer, Texas Gov., https://dvr.capitol.texas.gov/House/25 [https://perma.cc/AZ4J-QRU9]. The court-drawn map is found under H302. The 2000s-era plan is H100. And the enacted, invalid plan is titled H283. To cure the violation, the court set out to restore the VRA-compliant version of District 26 from the 2000s, making minor changes where necessary to correct for population changes.13Perez, 835 F. Supp. 2d at 211, 213. The court also explained that it would follow “neutral” redistricting principles: it would avoid splitting municipalities between districts and would draw a district as compactly as possible.14Id. at 212, 213.
But these principles—individually and as an ensemble—were (and almost always are) too vague to supply a decision rule.15See infra Section II.B.2.The new District 26 was reasonably compact, but the court could have shifted its boundaries east, west, or south to create a district nearly as compact and square shaped.16Perez, 835 F. Supp. 2d at 216 (displaying a picture of the court-drawn plan, H302). For a longer description of compactness, see infra Section II.B.2. The court needed to break some municipal lines to account for population growth, but the rules could not say precisely which or how many municipal lines to break.17The court broke several municipal lines: the new district indeed omitted the eastern part of Sugarland. See District Viewer, supra note 12. Under the VRA, the court would have to draw a majority–minority district—a district where voters of colors made up a majority of voters. But the VRA does not tell a court which voters of color in an area should constitute this majority.18Thornburg v. Gingles defines racial vote dilution under section 2 of the VRA. 478 U.S. 30, 49–51 (1986). It requires only that a district make a community of color a majority in that district. Id. Where there are more voters of color than are necessary to make up a majority, the district might be drawn in any number of ways while satisfying the VRA. Likewise, section 5 of the VRA requires that there be no retrogression of minority voting power. E.g., Beer v. United States, 425 U.S. 130, 149–50 (1976). But this leaves every nonretrogressive option on the table. See 52 U.S.C. § 10301 (originally enacted as Voting Rights Act of 1965, Pub. L. No. 89-110, § 2, 79 Stat. 437, 437); Id. § 10304 (originally enacted as Voting Rights Act of 1965, Pub. L. No. 89-110, § 5, 79 Stat. 437, 439).
District 26 therefore presented a challenging remedial task, but it was not unusual in this regard. Redistricting necessarily demands discretionary, policy-laden, and sometimes arbitrary trade-offs.19See infra Section II.B.2. The Supreme Court has said that map drawing always requires an unwieldy balancing of “myriad factors” that courts are not well positioned to do.20Abrams v. Johnson, 521 U.S. 74, 101 (1997).
The difficulties of court-drawn districts make it easy to understand the appeal of a world where judges could cure invalid maps without needing to draw districts, or at least not draw quite so many districts, in the first place. This is roughly the world advertised by contemporary advocates of ranked-choice voting. Ranked-choice voting (RCV) proposes an electoral system that obviates or diminishes the need to draw districts. In particular, RCV permits elections where single districts elect multiple representatives.21This discussion previews a longer analysis of multimember RCV. See infra Section II.A. Multi-winner elections usually suffer serious problems, including racial vote dilution and vote splitting. But RCV resolves them and therefore makes fair elections with fewer districts possible. Using RCV, the federal court in Texas could have constructed fifty districts of three representatives each instead of 150 districts of one member each.22For a sample map of Texas under multimember RCV, consider Fair Representation in Texas, FairVote (2017), https://fairvote.app.box.com/v/FairRepTexas [https://perma.cc/NP62-SVZW].
In light of the way that RCV lessens the burden on courts of implementing redistricting remedies, this Note argues for the use of multimember RCV as a redistricting remedy. Part I describes RCV and connects it to longstanding debates about the legitimacy of court-ordered remedies. Part II explores the contemporary law on remedial redistricting and critiques the open-ended, quasi-legislative choices that this approach forces on the courts. Part III makes the case that multimember RCV is the remedy that most simplifies the judicial task and best comports with contemporary understandings of the separation of powers.
I. Multimember RCV as a Solution to the Problem of the Court-Ordered Map
Multimember RCV has reentered the national conversation after several decades of dormancy. It is uniquely well suited to mitigate the burdens that remedial redistricting places on the judiciary. This Part explains the mechanics of RCV, discusses the basic difficulties of court-ordered redistricting, and points to a natural affinity between the advantages of multimember RCV and the problems of court-ordered maps.
A. RCV, Wasted Votes, and Multimember Districts
Between 1915 and 1948, twenty-four cities in the United States used RCV, including New York, Cincinnati, and Cleveland.23Lee Drutman, Breaking the Two-Party Doom Loop: The Case for Multiparty Democracy in America 186 (2020). By the 1960s, RCV had nearly gone extinct.24Only Cambridge, Massachusetts continued to use it. Id. Some scholars attribute RCV’s decline to the opportunities that the method appeared to offer usually unsuccessful political coalitions: African Americans in Toledo, for instance, and communists in New York City, won seats for the first time under multimember RCV. Id. For a list of jurisdictions that currently use RCV, see Where Ranked Choice Voting Is Used, FairVote, https://www.fairvote.org/where_is_ranked_choice_voting_used [https://perma.cc/78SW-QS6X]. But today, RCV has resurfaced as a much-discussed, New York Times–approved reform.25Editorial, The Primaries Are Just Dumb, N.Y. Times (Feb. 26, 2020), https://www.nytimes.com/2020/02/26/opinion/democrats-primary-south-carolina.html [https://perma.cc/HXN8-KLHY] (advocating for the use of RCV during the chaotic 2020 Democratic primaries). Editorials in leading newspapers and magazines endorsed it.26See, e.g., id.; Anne-Marie Slaughter, Francis Fukuyama & Larry Diamond, Ranked-Choice Voting, Politico, https://www.politico.com/interactives/2019/how-to-fix-politics-in-america/polarization/ranked-choice-voting [https://perma.cc/5YQF-KZRJ]; Sean McMorris, Opinion, There’s a Better Way to Vote: Choose More than One Candidate and Rank Them, L.A. Times (Mar. 18, 2020, 3:00 AM), https://www.latimes.com/opinion/story/2020-03-18/ranked-choice-vote-early-voting-primary-elections-wasted-votes [https://perma.cc/4FLG-GTJ5]; Editorial, It’s Time for Maryland and Virginia to Embrace Ranked-Choice Voting, Wash. Post (Jan. 24, 2019, 7:41 PM), https://www.washingtonpost.com/opinions/its-time-for-maryland-and-virginia-to-embrace-ranked-choice-voting/2019/01/24/df485890-2020-11e9-9145-3f74070bbdb9 [https://perma.cc/SQP4-CR9X]; Editorial, Give Ranked Voting A Shot. Our Democracy Would Benefit., Wash. Post (June 14, 2018, 7:03 PM), https://www.washingtonpost.com/opinions/give-ranked-voting-a-shot-our-democracy-would-benefit/2018/06/14/73c59356-6835-11e8-9e38-24e693b38637 [https://perma.cc/2QBE-PGJB]; Lee Drutman, If We’re Abolishing the Electoral College, Let’s Also Have Ranked-Choice Voting for President, Vox (Mar. 21, 2019, 11:26 AM), https://www.vox.com/polyarchy/2019/3/21/18275785/electoral-college-ranked-choice-voting-president-democracy [https://perma.cc/2J3Z-GBM5]. The American Academy of Arts and Sciences featured multimember RCV on its short list of proposals to improve American democracy.27Am. Acad. of Arts & Scis., Our Common Purpose: Reinventing American Democracy for the 21st Century 24 (2020).Maine adopted RCV in 2016.28Katharine Q. Seelye, Maine Adopts Ranked-Choice Voting. What Is It, and How Will It Work?, N.Y. Times (Dec. 3, 2016), https://www.nytimes.com/2016/12/03/us/maine-ranked-choice-voting.html [https://perma.cc/UVH3-YBDK]. In 2020, Alaska did the same,29Kelsey Piper, Alaska Voters Adopt Ranked-Choice Voting in Ballot Initiative, Vox (Nov. 19, 2020, 6:20 PM), https://www.vox.com/2020/11/19/21537126/alaska-measure-2-ranked-choice-voting-results [https://perma.cc/ZA7J-88SN]. several Democratic presidential candidates indicated support,30Nik DeCosta-Klipa, Elizabeth Warren Sounds Very Open to Ranked Choice Voting, Boston.com (June 11, 2019), https://www.boston.com/news/politics/2019/06/11/elizabeth-warren-ranked-choice-voting [https://perma.cc/A7T4-P2UH]; Danielle Wallace, AOC, Andrew Yang Get Wish as NYC Voters Adopt Ranked-Choice Voting System, Fox News (Nov. 6, 2019), https://www.foxnews.com/politics/nyc-new-york-city-ranked-choice-voting-system-andrew-yang-aoc-ocasio-cortez [https://perma.cc/FJL6-5R9X]. and, for the first time, RCV was used in presidential primaries.31See RCV for Presidential Nominations, FairVote, https://www.fairvote.org/rcv_for_presidential_nominations#experience_ranked_choice_voting [https://perma.cc/RM4M-8USC].
The basic premise of RCV is as follows: an RCV ballot collects voters’ first choices, but also their second, third, and fourth choices—and often more.32See Richard H. Pildes & G. Michael Parsons, The Legality of Ranked-Choice Voting, 109 Calif. L. Rev. (forthcoming 2021) (manuscript at 11–12), https://doi.org/10.2139/ssrn.3563257 [https://perma.cc/QJ4A-JSYH] (providing a basic overview of the RCV tabulation process); How RCV Works, FairVote, https://www.fairvote.org/how_rcv_works [https://perma.cc/K2DP-38QA]. When a voter registers a first preference for a candidate who has too few votes to win, her vote is reallocated from the losing candidate, on whom the vote is said to be “wasted,” to a candidate who remains viable. If, for instance, you most preferred Ralph Nader in 2000, an RCV ballot would allow you to also indicate your second-place preference for another candidate, either George W. Bush or Al Gore. When Nader ultimately failed to receive a majority of first-place votes, your ballot would not have been “wasted.”33In an election with one winner, such as a presidential contest, a wasted vote is any vote for a losing candidate. Pildes & Parsons, supra note 32 (manuscript at 14). For an explanation of wasted votes in the multiwinner context, see infra text accompanying notes 77–83. For a refresher on the 2000 campaign and the debate over whether Nader split votes, see Evelyn Nieves, Conversation/Ralph Nader; A Party Crasher’s Lone Regret: That He Didn’t Get More Votes, N.Y. Times (Feb. 18, 2001), https://www.nytimes.com/2001/02/18/weekinreview/conversation-ralph-nader-party-crasher-s-lone-regret-that-he-didn-t-get-more.html [https://perma.cc/3N3B-T3AA]. It would have been reallocated to your top choice among the still viable candidates.
Most American jurisdictions, of course, vote in a different way: they use the “single-choice vote” (SCV), where a ballot asks for one preference, not a ranking, and where the winner is the candidate who earns the most votes.34Pildes & Parsons, supra note 32 (manuscript at 7–8). The Nader scenario illustrates two basic advantages RCV has over SCV. First, RCV prevents vote splitting, in which two candidates divide support (e.g., Nader and Gore) and allow a third candidate to prevail.35Id. Second, RCV makes it less important to vote strategically. Because a voter need not worry about vote splitting, she can support the candidates she likes best, instead of the candidates she thinks have the best chance of winning.36Id. (manuscript at 9).
In proposing RCV as a remedy for redistricting violations, this Note stresses a particular impact of RCV on the geography of American elections.37In considering RCV as a remedy, this Note does not consider threshold issues about the legality of RCV as such. If state or federal law prohibited multimember RCV, it could not be adopted as a remedy any more than it could be adopted from scratch by a legislature. See, e.g., Harper v. City of Chicago Heights, 223 F.3d 593, 601 (7th Cir. 2000). Federal law does not allow multimember RCV in elections for Congress, because 2 U.S.C. § 2c requires single-member districts in congressional elections. There is also continuing debate about the compatibility of RCV and some state constitutional provisions. Pildes & Parsons, supra note 32. Scholars and courts have already developed these issues, and this Note does not elaborate on them further. Under RCV, a jurisdiction can elect multiple candidates from a single district, instead of requiring that each district elect only one candidate.38See infra Section II.A. RCV would therefore ease the burden on the map drawer of producing the redistricting remedy.
As a remedy, RCV raises a simple legal question and a difficult remedial question. The simple question is whether a legislature, if it so desires, can cure its redistricting violation by implementing RCV. Courts have said, essentially, yes.39No published opinions have dealt with RCV as a remedy. VRA Remedy Database, FairVote, https://www.fairvote.org/vra_remedy_database [https://perma.cc/R6LM-CMDD]. But several opinions deal with closely related methods, cumulative voting and limited voting, and have approved their use so long as the size of the minority community exceeds the threshold of exclusion. See infra Section II.A. For example, in United States v. Village of Port Chester, the at-large voting system for the city’s Board of Trustees violated section 2 of the VRA. 704 F. Supp. 2d 411 (S.D.N.Y. 2010). The court approved the defendant’s proposed remedy, a cumulative voting system, because the Port Chester Hispanic community’s citizen voting-age population (21.9%) exceeded the threshold of exclusion (14.3%). Port Chester, 704 F. Supp. 2d at 450–51. In United States v. Euclid City School Board, the court upheld the limited voting method on grounds that the Black voting age population was 27% and the threshold of exclusion 25%. 632 F. Supp. 2d 740, 761, 770 (N.D. Ohio 2009). In Dillard v. Chilton County Board of Education, the threshold of exclusion was 12.51% and the minority voting population 11.86%. 699 F. Supp. 870, 874–75 (M.D. Ala. 1988). A legislature may decide how to cure its redistricting violation, so long as it offers an effective remedy.40E.g., Euclid City Sch. Bd., 632 F. Supp. 2d at 750 (“[A] court is not to inquire whether the defendants have proposed the very best available remedy, or even whether the defendants have proposed an appealing one.”). And, as Part II demonstrates, RCV is indeed a valid remedy. In June 2019, for instance, the Department of Justice settled a Voting Rights Act lawsuit with the City of Eastpointe, Michigan when Eastpointe’s City Council adopted multimember RCV instead of the usual single-member districting map.41Press Release, Dep’t of Just., Justice Department Reaches Agreement with City of Eastpointe, Michigan, Under the Voting Rights Act (June 5, 2019), https://www.justice.gov/opa/pr/justice-department-reaches-agreement-city-eastpointe-michigan-under-voting-rights-act [https://perma.cc/35R7-L3CB].
RCV also raises a second and more difficult remedial question, which is the subject of this Note. Courts must sometimes design a remedy themselves, as happened in Texas in 2012.42See supra Introduction (discussing the case of Texas in 2012). This situation raises the issue of whether a court might order a jurisdiction to adopt RCV as a redistricting remedy. Several district courts have contemplated this kind of remedy.43See infra Section II.B.1. Courts have contemplated ordering, not multimember RCV, but closely related proportional election methods. As RCV gains momentum and as this decade’s redistricting cycle commences, we can expect the question to surface more often. This Note analyzes the current doctrine and argues in favor of multimember RCV as a court-ordered redistricting remedy.
B. The Problem of the Court-Ordered Map
American election law assigns responsibility for designing elections and drawing districts to legislatures.44See supra note 1 and accompanying text. But in the course of the many redistricting battles that follow each census, courts often have to assume this responsibility themselves.45Persily, supra note 1, at 1131 (“In the many redistricting struggles that now follow each census, plaintiffs routinely turn to the courts, not only to strike down plans as illegal, but also to draw remedial plans to take their place.”). In the 2000s, courts reviewed congressional redistricting plans in eighteen states and drew maps in nine; courts reviewed state legislative plans in thirty-five and drew maps in thirteen. In the 2010s, the courts reviewed congressional plans in twenty-eight states and drew maps in twelve; for state legislatures, the numbers were, respectively, thirty-five and nine. Justin Levitt, Who Draws the Lines?, All About Redistricting, https://redistricting.lls.edu/who-courtfed10.php (data on file with the Michigan Law Review). This poses a problem. Redistricting is a complex, multifaceted task, whether legislatures or courts draw the maps. A plurality of the Supreme Court once regarded redistricting as such a morass—a “political thicket”—that it refused to entertain plaintiffs’ claims about redistricting violations, let alone draw maps to correct them.46Colegrove v. Green, 328 U.S. 549, 556 (1946) (plurality opinion). About courts drawing districts, Justice Frankfurter wrote “[o]f course no court can affirmatively re-map . . . districts so as to bring them more in conformity with the standards of fairness for a representative system.” Id. at 553.
Today, the courts hear hundreds of redistricting claims and often draw remedial districts themselves.47For a catalogue of court-initiated remedies, see Levitt, supra note 45. For one measure of the volume of redistricting litigation, see Ellen Katz with Margaret Aisenbrey, Anna Baldwin, Emma Cheuse & Anna Weisbrodt, Documenting Discrimination in Voting: Judicial Findings Under Section 2 of the Voting Rights Act Since 1982, 39 U. Mich. J.L. Reform 643, 654 (2006) (documenting hundreds of VRA section 2 claims). This is not because the Supreme Court has come to celebrate the judiciary’s role in redistricting. On the contrary, the Court continues to emphasize that there are few “neutral legal principles” available when drawing districts48Perry v. Perez, 565 U.S. 388, 393 (2012) (per curiam). and that courts “possess no distinctive mandate to compromise sometimes conflicting state [redistricting] policies in the people’s name.”49Connor v. Finch, 431 U.S. 407, 415 (1977). But the courts’ involvement in remedial redistricting necessarily followed from their involvement in redistricting at the liability stage.
Beginning in the 1960s, as the federal government confronted Jim Crow, the Supreme Court and Congress brought the Fourteenth Amendment to bear on abuses of redistricting. The Court required population equality among districts in Reynolds v. Sims and took aim at racial vote dilution in White v. Regester.50377 U.S. 533 (1964) (articulating one-person, one-vote principle); 412 U.S. 755 (1973). Congress relied on the latter decision to create a statutory cause of action against racial vote dilution in the Voting Rights Act of 1982.51Voting Rights Act Amendments of 1982, Pub. L. 97–205, sec. 3, § 2, 96 Stat. 131, 134 (codified at 52 U.S.C. § 10301); see Laughlin McDonald, Voting Rights and the Court: Drawing the Lines, 15 S. Changes, no. 3, 1993, at 1, 2.
A typical redistricting lawsuit is brought under section 2 of the Voting Rights Act and focuses on racial vote dilution.52For evidence that section 2 is a “typical” or common redistricting lawsuit, consider the most comprehensive survey of section 2 litigation to date, published in 2006, which counted 331 section 2 lawsuits between 1982 and 2006 with published opinions. Katz et al., supra note 47, at 654. About section 2, note that while the Court recognized a cause of action under the Fourteenth Amendment for racial vote dilution in White, 412 U.S. 755, the Court subsequently made this cause of action much less useful for plaintiffs by requiring that plaintiffs demonstrate discriminatory intent. City of Mobile v. Bolden, 446 U.S. 55 (1980), superseded by statute, Voting Rights Act. Plaintiffs in such cases argue that a jurisdiction has denied a community of color a chance to translate its votes into political power.53The Court articulated the standard for racial vote dilution claims under section 2 in Thornburg v. Gingles, 478 U.S. 30, 50–51 (1986). Plaintiffs must demonstrate, at minimum, that (1) plaintiff belongs to a minority community “that . . . is sufficiently large and geographically compact to constitute a majority in a single-member district”; (2) that the minority group is politically cohesive and has distinctive interests; and (3) that whites vote enough as a bloc so that white voters usually will defeat the minority’s preferred candidate. Id. A court may then consider the “Senate Factors,” qualitative indicators of dilution, such as whether political campaigns have featured overt racist appeals. Id. at 36–37. A state government, for instance, may have gerrymandered its congressional districts so that voters of color are packed into one district rather than spread efficiently across several districts.54Imagine a district where Black voters are 90 percent of the electorate, and a neighboring district where Black voters are 10 percent of the electorate. If there was racially polarized voting, and if the districts could have been drawn so Black voters constituted a majority in each district, plaintiffs could probably make out a section 2 violation. In effect, the map drawer would be guilty of denying Black voters a fair opportunity to elect candidates of choice by packing their influence into one district, rather than spreading it out across several. See id. at 49–51 (defining the elements of a vote dilution claim as geographic compactness, cohesion of the minority group, and cohesion of the larger majority). As a result, voters of color might receive less representation in Congress than their share of the electorate would suggest they deserve. When a court finds that a state map impermissibly dilutes minority voters or otherwise violates the VRA, it may strike the map down and impose a remedy for the violation.
This Note focuses on court-ordered remedies: the instances where courts must “affirmatively re-map” to resolve a redistricting violation by redrawing districts and reconfiguring elections. Courts, to be clear, do not always need to create a remedy on their own.55Colegrove v. Green, 328 U.S. 549, 553 (1946). After a map is struck down, the first step is to offer the offending legislature a chance to design a remedy itself. If the legislature’s new map cures the redistricting violation, the court has no more remedial work to do.56See Persily, supra note 1, at 1133 (explaining that courts only draw districts as a “last resort”).
But courts must design their own remedies when the offending legislature does not come forward with a remedy.57Id. If, for instance, legislative deliberations about a new redistricting map break down, if an election is approaching too quickly, or if the legislature continues to produce invalid maps, a court must construct a remedy of its own. This scenario saddles the court with an uncomfortable remedial task.
Court-ordered redistricting raises a difficult problem because, as scholars and judges have long argued, remedies that restructure public institutions push the limits of the judicial role.58See, e.g., Daryl J. Levinson, Rights Essentialism and Remedial Equilibration, 99 Colum. L. Rev. 857, 861 (1999) (“[C]ourts have no special license or ability to make the types of policy decisions that remedies require, and because the political branches possess not only democratic legitimacy but also superior fact-finding and interest-balancing capacities, courts should defer to the political branches about issues of implementing or enforcing rights.”); Abram Chayes, The Supreme Court, 1981 Term—Foreword: Public Law Litigation and the Burger Court, 96 Harv. L. Rev. 4, 45–47 (1982). When ordering such a remedy, a judge must do more than answer a binary question about liability.59See William A. Fletcher, The Discretionary Constitution: Institutional Remedies and Judicial Legitimacy, 91 Yale L.J. 635, 644–45 (1982). A judge must, in quasi-legislative fashion, craft a new institution—whether a reformed prison, desegregated school, or a new redistricting scheme.60See Levinson, supra note 58, at 865 (describing the view that legislatures, not courts, traditionally take responsibility for “find[ing] facts, mak[ing] fact-specific compromises, and efficiently trad[ing] off costs and benefits in the service of effectuating constitutional values in the real world”); Fletcher, supra note 59 (describing the difficult responsibilities a judge takes on when redistricting, desegregating schools, or addressing abusive prison conditions). In so doing, Judge William Fletcher has written, a judge moves “beyond the normal competence and authority of a judicial officer, into an arena where legal aspirations, bureaucratic possibilities, and political constraints converge, and where ordinary legal rules frequently are inapplicable.”61Fletcher, supra note 59, at 641. Redistricting remedies thus raise basic questions about institutional legitimacy, the scope of the judicial role, and the separation of powers.
At present, courts remedy redistricting violations by drawing single-member districts.62See Chapman v. Meier, 420 U.S. 1, 19 (1975) (“Absent particularly pressing features calling for multimember districts, a United States district court should refrain from imposing them upon a State.”). But single-member remedies exacerbate rather than resolve the difficulties of court-ordered maps. Single-member remedies require that courts make fact-specific, value-laden efforts to define political communities by drawing districts around them.63See infra Section II.B.2. Instead, courts should use multimember RCV as a redistricting remedy. RCV permits elections with multiple winners from single districts. It permits, in other words, elections with larger—and fewer—districts than the status quo. As districts grow in size and scope, map drawers make fewer decisions about which voters belong together in a political community. Multimember RCV thus simplifies the judicial task and better protects the integrity of the judicial role.
To suggest “multimember ranked-choice voting” is really to suggest two changes: first is a move from the “single-choice vote” to “ranked-choice voting”; second is a shift from “single-member districts” to “multimember districts.” We have already encountered the distinction between single-choice voting and ranked-choice voting. SCV asks voters for a single preference and makes a winner of the candidate who receives the most votes. RCV, by contrast, collects a full menu of preferences from voters and determines the winner after reassigning votes wasted on nonviable candidates to those who are still in contention.64See supra Section I.A.
SCV and RCV therefore differ in the information they collect from voters and in the way they aggregate this information to choose the winner of an election. By contrast, the terms “single-member districts” and “multimember districts” describe a difference of geography. Single-member districts divide a jurisdiction into a piece of territory for each representative elected. The United States House of Representatives provides an example. Americans elect 435 members of Congress. Each member of Congress hails from a distinct district, called “single-member” because each district elects one representative.65E.g., Am. Acad. of Arts & Scis., supra note 27, at 25.
In a multimember election, a district instead elects multiple representatives.66Id. Consider Ann Arbor, Michigan, which has ten city council members but only five districts.67City Council, City Ann Arbor, https://www.a2gov.org/departments/city-council/Pages/Home.aspx [https://perma.cc/59KD-Y46V]. These districts are “multimember” because they each elect two representatives. Multimember elections might feature several multimember districts, as in Ann Arbor, where there are five such districts. A multimember election might also have no districts at all. Imagine that a city has a five-member city council and elects all five candidates from the entire city. This “no districts” election, a special kind of multimember election, is sometimes called an “at-large” election.68E.g., City of Mobile v. Bolden, 446 U.S. 55, 70 & n.15 (1980) (opinion of Stewart, J.) (using the phrase “at-large election” to describe this type of election), superseded by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97-205, sec. 3, § 2, 96 Stat. 131, 134.
To propose multimember RCV as a redistricting remedy is therefore to propose that courts create electoral maps with no districts or, alternatively, with districts that elect multiple representatives at a time. Winners would be chosen using RCV. Courts currently take a different approach.69See infra Part II. For reasons explored in Part II, they use single-member districts with SCV. The preference for single-member districts means that courts draw one district for each representative elected. These districts choose their winners with SCV.70Unless otherwise noted, “single-member districts,” when used in this Note, refers to single-member districts with SCV.
This discussion brings to light the basic appeal of multimember RCV as a redistricting remedy: it does not require that courts draw a district for every representative elected. Some versions of multimember RCV would not require that courts draw any districts at all. While single-member districts maximize courts’ exposure to the “political thicket” of drawing districts, the use of multimember RCV would minimize it.
II. Multimember RCV as a Corrective to Two Mistakes in Redistricting Doctrine
Multimember RCV allows courts to avoid drawing districts for every representative elected. It keeps courts at arm’s length from one aspect of the political thicket. American election law, however, advises against multimember remedies and in favor of single-member districts.71Chapman v. Meier, 420 U.S. 1, 2, 15–16 (1975). In Chapman, the Supreme Court provided three reasons for presuming that single-member districts are the remedy of choice. These reasons are as follows: First, multimember districts tend to cause vote dilution. Second, multimember districts impose more difficult choices on voters than single-member elections and make ballots “unwieldy, confusing, and too lengthy to allow thoughtful consideration.” Third, multimember districts may deprive voters of the feeling that they have a representative especially responsive to them. Id. at 15–16. Section II.A directly addresses the vote dilution question. The second concern, about confusion, relates to the discussion of voter confusion and ballot exhaustion, also in Section II.A. The third concern is not addressed directly. But it’s one of the many reasons that the imposition of multimember RCV is intrusive. Part III addresses how to balance the different intrusions at stake in remedial redistricting.This Part explores those reasons and offers a critique. Multimember RCV remedies vote dilution, just as single-member districts do. Multimember RCV remedies also ensure that a court avoids acting like a legislature when redistricting.
Two reasons loom large in the courts’ preference for single-member districts. First, courts have generally assumed that only single-member districts can solve racial vote dilution.72See Connor v. Finch, 431 U.S. 407, 415 (1977) (explaining that multimember districts “tend to submerge electoral minorities and overrepresent electoral majorities”). If this were true, courts could not issue multimember remedies, because the remedy itself might cause a legal violation or fail to resolve the existing one. This concern has some truth. Multimember districts with SCV cause vote dilution. Many jurisdictions historically employed it for precisely that reason.73See infra Section II.A. But RCV makes the crucial difference. Section II.A explains that multimember districts with RCV redress vote dilution as effectively as single-member districts do.
Second, courts have ordered single-member districts in an effort to respect the separation of powers. Courts agree that legislatures, not judges, should decide how to conduct elections. Most legislatures use single-member districts, not multimember RCV. Courts have decided, therefore, that when they must redistrict on behalf of a legislature, they should order the remedy that the legislature might have chosen itself. Section II.B argues that courts should not draw single-member districts on behalf of a legislature since doing so requires that courts act too much like a legislature.
A. Multimember RCV as a Solution to Vote Dilution
Courts continue to draw remedial single-member district lines in part because of the assumption that multimember elections without districts cause vote dilution.74Connor, 431 U.S. at 413–14. This assumption forms one basis of the Supreme Court’s presumption in favor of single-member remedies.75Chapman, 420 U.S. at 15–16 (citing vote dilution as one of three reasons for preferring single-member remedies).
As it turns out, courts have misunderstood the connection between multimember elections and vote dilution. Certain kinds of multimember elections, described below, do cause vote dilution. But multimember elections can occur without vote dilution so long as they collect enough information from voters. An RCV ballot collects a full menu of voter preferences. The combination of RCV and multimember districts permits multimember elections without vote dilution.
This Section uses Mobile, Alabama, a subject of vote-dilution litigation in the 1980s, to discuss the available remedies.76City of Mobile v. Bolden, 446 U.S. 55 (1980), superseded by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97-205, sec. 3, § 2, 96 Stat. 131, 134, as recognized in Thornburg v. Gingles, 478 U.S. 30, 35 (1986). Note that the Court ultimately found no racial vote dilution under the Fourteenth Amendment. Id. at 78. The Voting Rights Act Amendments of 1982 would likely have changed the outcome. See infra text accompanying notes 87–89 (describing the application of Gingles factors to Mobile). When the Supreme Court heard a challenge about racial vote dilution in Mobile in the 1980s, Black voters made up 35 percent of Mobile’s electorate and white voters the remaining 65 percent.77See Bolden, 446 U.S. at 97 (White, J., dissenting). Despite this multiracial electorate, no Black candidate in Mobile had ever been elected to a seat on Mobile’s three-member city commission.78Id. at 55 (opinion of Stewart, J.); id. at 97–98 (White, J., dissenting). By the 1980s, the problem was not that Black voters were barred from casting ballots. They could access the voting booth. But their votes, once cast, were diluted and lost in a sea of white votes because of the way Mobile held elections.79Id. at 97–98.
Mobile held separate, citywide contests for the three seats on the city commission.80Id. at 59 (opinion of Stewart, J.). The candidate with the most votes in each contest earned the seat.81Id. at 59–60. In other words, Mobile held multimember elections with no districts (e.g., an at-large election) and selected winners in each contest via single-choice vote. The important implication of these rules was that the contest for each seat involved the same voters and could involve the same coalitions. Any coalition big enough to outvote opponents in the race for one seat was able to do the same in the races for the other two. This was a recipe for vote dilution.
In Mobile, white voters were a majority of the electorate and Black voters the minority. It followed that if white and Black voters lined up behind opposing candidates in all three contests, as usually happened,82Id. at 71 (indicating that Mobile experienced racially polarized voting). the Black community’s candidates would lose every time. Such a system, as Professor Lani Guiner has described, allowed “[f]ifty-one percent of the population [to] consistently decide one hundred percent of the elections.”83Lani Guinier, The Triumph of Tokenism: The Voting Rights Act and the Theory of Black Electoral Success, 89 Mich. L. Rev. 1077, 1094 (1991).
Beginning in the early twentieth century, jurisdictions across the country used this election format to dilute the votes of Black communities and other minorities.84Samuel Issacharoff, Polarized Voting and the Political Process: The Transformation of Voting Rights Jurisprudence, 90 Mich. L. Rev. 1833, 1839–40 (1992) (tracing the history of multimember districts to an alliance of northern Progressives and southerners supportive of Jim Crow, who saw multimember districts as a way to curtail the power of smaller ethnic voting blocs). Courts came to solve this brand of vote dilution by jettisoning multimember elections and incorporating single-member districts.85See Samuel Issacharoff, Pamela S. Karlan, Richard H. Pildes & Nathaniel Persily, The Law of Democracy: Legal Structure of the Political Process 846 (5th ed. 2016) (describing the wave of lawsuits after Thornburg v. Gingles, 478 U.S. 30 (1986), that converted multimember districts to single-member districts); see also Guinier, supra note 83, at 1094. Single-member districts solve dilution by subdividing territory. A minority at the jurisdiction level can turn into a majority at the district level. In Mobile, if there had been single-member districts, Black voters could constitute a majority in one of three smaller subdivisions of territory. Consequently, when the community voted as a bloc, it could earn one of the three seats up for election.86Because there were three city commission seats in Mobile, an election with single-member districts would imply three districts. Each district could need to have a third of Mobile’s population. Reynolds v. Sims, 377 U.S. 533 (1964). Black voters were 35% of Mobile’s electorate, Bolden, 446 U.S. at 97 (White, J., dissenting), so Black voters were more than numerous enough to constitute a majority in one district.
Single-member remedies require subdividing territory and drawing districts, however. When a court administers a single-member remedy, a court has to decide where to draw lines. A court would have to decide which neighborhoods belong together in a district. In Mobile, the court would have to decide which of the city’s Black voters belonged in a majority-minority district.87Each of Mobile’s three districts would have 33% of the city’s population. A coalition could command a majority in a district if it was 50% + 1 of the vote in a district, or (50% ⋅ 33%) = 16.5% + 1 of Mobile’s population. Since Black voters were 35% of Mobile’s population, not all of Mobile’s Black community would need to inhabit one district in order for that district to produce a Black candidate of choice. And so the map drawer would need to decide how to allocate Black votes around the city. See infra Section III.B (elaborating on the idea that districting requires judges to decide which minority voters belong in a district together). A legislature or democratically responsive body might credibly make decisions about whether, for instance, a city block should go in district A or B.88Robert A. Burt, Miranda and Title II: A Morganatic Marriage, 1969 Sup. Ct. Rev. 81, 113–14. (“Congress can make distinctions among classes that the Court would itself be hard put to explain on principled grounds . . . because the institutional legitimacy of a legislative act depends not so much on the rational persuasiveness of its decisions as on the simple fact that a majority of ‘responsible’ elected officials were willing to vote for the proposition.”). But, as the Supreme Court has emphasized, this kind of decisionmaking is a strange fit for the courts.89See infra Section II.B.2.
Courts have nonetheless accepted responsibility for drawing districts, in part because of the assumption that single-member districts are the only viable remedy.90See supra text accompanying notes 77–85. But multimember RCV is a remedy that also cures vote dilution. If Mobile had used multimember RCV, the city would still hold one election.91How RCV Works, supra note 32. All candidates would run in the same contest rather than for a particular seat (e.g., for seat one or seat two). RCV would set the “threshold of exclusion,” or the threshold needed for victory, at 25% plus one vote.92Steven J. Mulroy, The Way Out: A Legal Standard for Imposing Alternative Electoral Systems as Voting Rights Remedies, 33 Harv. C.R.-C.L. L. Rev. 333, 363 (1998). The threshold of exclusion reflects the following: If one candidate achieves more than 25% of the vote, it is guaranteed that, at most, only two other candidates can also earn more than 25% of the vote. (It would be impossible for four total candidates to earn more than 25% of the vote.) Therefore, any candidate in Mobile who earned at least one vote above 25% of the vote is a guaranteed winner since they are guaranteed to be among the top three finishers. As a general matter, the threshold of exclusion is calculated as one vote greater than 1 ∕ (1+s), where s is the number of seats needing to be filled. Id. at 342. In Cambridge, Massachusetts, for instance, where the entire city elects nine city council members, the threshold of exclusion is 1/10, or 10%. Andrew Douglas, Cambridge, Massachusetts Elections a Model for America, FairVote (Nov. 1, 2013), https://www.fairvote.org/cambridge-massachusetts-elections-a-model-for-america [https://perma.cc/Q46G-D96M]. Because the threshold of exclusion (25%) is lower than the size of Mobile’s Black community (35%), Black voters could elect a candidate of choice if they voted as a bloc—the same outcome as in the case of single-member districts.
The distinctive feature of RCV, the ranked-choice ballot, would enter the picture in the scenario where some of the top candidates only earned tiny fractions of the vote. If fewer than three candidates earned more than 25% of the vote after a first pass, RCV would use the other preferences collected on the ballot to find the most preferred remaining candidate. Voters whose first-preference candidate had already won or who were assuredly going to lose would have their ballots reallocated to a second-preference candidate. This process would repeat until a third winner, with more than 25% of the vote, emerged.93See How RCV Works, supra note 32. A vote that needs reallocating—a “wasted vote”—has a complicated meaning in the multimember RCV setting. A wasted vote is any vote that goes to a candidate in last place, but also any vote for a candidate who has earned in excess of what they need to win. Say that Candidate A has 35 votes but needed only 26 votes to win. In the aggregate, 9 votes were wasted. But how much of every individual vote was wasted? Nine divided by 35, or 25.7%, of every vote for Candidate A was wasted. That fraction of a vote will be reallocated to the second preference of every voter whose first choice was Candidate A. See Multi-Winner Ranked Choice Voting, FairVote, https://www.fairvote.org/multi_winner_rcv_example [https://perma.cc/T3CN-6Y7Q].
Multimember RCV, like single-member districting, can solve vote dilution because it permits a minority coalition to earn seats. But the example of Mobile has suggested that multimember RCV takes a comparatively less intrusive approach. Single-member districts divide a territory into small chunks, so that a minority at a jurisdiction level becomes a majority at the district level. Multimember RCV’s method is to simply lower the number of votes required for victory in the jurisdiction writ large.
The discussion thus far has suggested that multimember RCV offers a simple value add. The status quo of single-member districts requires that courts divide up communities into districts. Multimember RCV might absolve the court of the need to draw any districts at all. This description is accurate so far as small jurisdictions like Mobile go. In general, however, a multimember RCV election requires some districts rather than none. As explored below, an RCV ballot that seeks to elect more than a few candidates from a single ballot becomes unmanageable.94See Expert Report of Nolan McCarty, Ph.D. at 7, Hagopian v. Dunlap, 480 F. Supp. 3d 288 (D. Me. 2020) (No. 20-cv-00257) [hereinafter Expert Report], ECF No. 1-1 (demonstrating a relationship between the problem of ballot exhaustion and the number of candidates on a ballot). No matter the jurisdiction, however, multimember RCV delivers a simplified remedial task for the judge because it reduces the number of districts that need drawing.
RCV elections require districts when the number of representatives to be selected is too great for one ballot. Mobile, for instance, sought to elect three city commission members. An RCV ballot in Mobile would accordingly require that voters rank roughly four or five candidates, a task that scholars assume voters can handle.95See Drutman, supra note 23, at 231 (suggesting a five-winner election as the properly sized multimember RCV election). See infra note 101 for analysis of how many candidates a voter needs to rank given the number of candidates in a race. But suppose a court imposed at-large RCV on Texas for elections to the Texas State House. That chamber has 150 members.96House Members, Tex. House Representatives, https://house.texas.gov/members/ [https://perma.cc/TYU9-APX8]. An RCV ballot in these circumstances would ask voters to articulate, for instance, their seventy-fifth preference among candidates for state legislature. If about half of Americans cannot name a Supreme Court justice,97Ed Kilgore, Most Americans Can’t Name a Supreme Court Justice, N.Y. Mag.: Intelligencer (Sept 5. 2018), https://nymag.com/intelligencer/2018/09/most-americans-cant-name-a-supreme-court-justice.html [https://perma.cc/VTA9-PYQH]. voters almost certainly cannot articulate a seventy-fifth preference among candidates in a state legislature race.
It would be one thing if an RCV ballot requiring hundreds of rankings were futile but inconsequential. In fact, RCV cannot work as advertised if voters cannot articulate a sufficient number of preferences on their RCV ballot.98The following is my own presentation based on How RCV Works, supra note 32. But it echoes the common refrain that ballot exhaustion means RCV does not select candidates preferred by a majority of voters. E.g., Craig M. Burnett & Vladimir Kogan, Ballot (and Voter) “Exhaustion” Under Instant Runoff Voting: An Examination of Four Ranked-Choice Elections, 37 Electoral Stud. 41 (2015). For one study that attempts to systematically measure the effects of ballot exhaustion, see D. Marc Kilgour, Jean-Charles Grégoire & Angèle M. Foley, The Prevalence and Consequences of Ballot Truncation in Ranked-Choice Elections, 184 Pub. Choice 197 (2020) (finding that “small amounts of truncation can alter election outcomes”). When voters mark only a few preferences but not an entire slate—say, preferences 1 through 5 but not 5 through 150—an RCV election may succumb to a problem called “ballot exhaustion.”99E.g., Simon Waxman, Ranked-Choice Voting Is Not the Solution, Democracy (Nov. 3, 2016, 3:03 PM), https://democracyjournal.org/arguments/ranked-choice-voting-is-not-the-solution [https://perma.cc/JZ2H-TEWB] (describing this problem as ballot exhaustion). That problem works as follows. When ballots are left uncompleted, RCV is unable to reallocate “wasted” votes (e.g., those spent on Ralph Nader) to candidates who remain viable. This means RCV cannot protect against the undesirable outcomes of vote splitting. Most voters, for instance, might prefer Candidate A to Candidate B. But suppose these voters have even stronger preferences for Candidates C, D, E, or F. If these voters split their top-preference votes among C, D, E, and F, and fail to indicate their more weakly held preference for Candidate A, Candidate B can win. In other words, when voters do not articulate a full slate of preferences, RCV may elevate unpopular candidates, like B, who win only because opponents split votes.
RCV cannot, therefore, facilitate an at-large election (that is, a districtless election) when dozens of seats need filling. But this does not mean RCV cannot apply in any form in a jurisdiction like Texas. Where an at-large RCV election is impractical, jurisdictions can adopt RCV with multimember districts.100See, e.g., Fair Representation in Texas, supra note 22. As an example, Texas might not hold a single RCV election for 150 representatives. Instead, it could divide its territory into fifty districts of three members each. No individual voter would need to articulate more than perhaps five preferences, since each ballot would also seek to elect three representatives in a particular district.101The number of preferences a voter must articulate to avoid ballot exhaustion is a function of the number of candidates in the race. E.g., Expert Report, supra note 94, at 10 (defining a fully participating voter as a voter who ranks one less than the total number of candidates in the race). Supposing that the two major political parties, and only those parties, nominated candidates for each of the three seats, there would be six candidates total, and a fully participating voter would need to rank five candidates. But the election as a whole would produce 150 winners, as desired.
The jury is still out on exactly how many preferences an RCV election can rely on voters to articulate. In other words, debate continues about how many multimember districts a jurisdiction like Texas might require, and therefore about the extent to which RCV would absolve map drawers of the need to draw districts.102The debate continues not only because of the conflicting evidence, see infra notes 110–111, but because it is unclear how voter behavior with RCV will change over time. Some RCV advocates suggest, for instance, that studies of behavior right now are misleading as voters will grow more competent at ranking long lists of candidates as RCV becomes more common. Expert Report, supra note 94, at 7. Some scholars have suggested that voters can competently participate in RCV elections that select five representatives.103See Drutman, supra note 23, at 213 (describing the potential campaign benefits in a hypothetical five-party race); John M. Carey & Simon Hix, The Electoral Sweet Spot: Low-Magnitude Proportional Electoral Systems, 55 Am. J. Pol. Sci. 383, 390 (2011) (finding, in a cross-national study, that benefits of multimember districts in terms of better representation diminish after a district includes more than 5 representatives). Empirical studies of ballot exhaustion in recent RCV elections suggest even five may be too many.104See Expert Report, supra note 94, at 11 (finding that only 36 percent of voters in a recent RCV election in Maine ranked three candidates). While research on voters’ competencies develops, it makes sense for judges ordering RCV to err on the side of caution by selecting multimember districts of sizes of two, three, or four.
Regardless, the essential benefits of RCV for remedial redistricting remain, because RCV allows larger districts, and larger districts simplify the judicial task. An RCV election that used multimember districts could increase the population size of every district by a factor of at least two or three and reduce the number of districts needing drawing by the same amount. Crucially, when districts are bigger, a given district is more likely to subsume an area and less likely to pose difficult questions about how to draw lines through that area. When judges can draw districts that preserve existing boundaries—county, city, municipal, or political—judges need not make sui generis decisions about which voters should belong together.105As Section II.B.2 explains, the law requires courts to redistrict according to state redistricting principles, and many states require that districts follow county or city boundaries and preserve communities of interest. Even if this weren’t true, judges could naturally rely on these subdivision boundaries as guidance about which voters should go together. The existing boundaries provide the relevant guidance. But when districts must break these provided-for lines, the rules have run out and judges must make decisions about defining political community themselves. By increasing the sizes of districts, multimember RCV absolves judges of making some of these choices.
Multimember RCV may generate appreciable gains even if it only permits elections of two or three representatives at a time. To make these benefits more concrete, consider a stylized example of the relationship between the size of a state house district, and the number of counties that a map drawer of state house districts must break. Whenever a county’s population is bigger than the size of a state’s house districts, a map drawer must necessarily break up that county, since the county cannot fit inside the district. Counties that can fit inside a district may require splitting, too, since a district that encircles a smaller county will not achieve the required population without adding pieces of other counties. But not every such big county needs breaking up. Thus, the percent of a state’s counties that exceed the size of a state house district provides a rough cut of how many counties a map drawer must break up when designing state house elections.
By this measure, RCV could significantly simplify redistricting remedies if it merely allowed multimember districts of two or three members. Across states with single-member districts in their lower houses,106Thirty-nine states have lower state houses with single-member districts. See Am. Acad. of Arts & Scis., supra note 27, at 26, 71 n.28 (listing the ten states that already use multimember districts in state house races). Nebraska has no lower house as it is a unicameral legislature. Nebraska Unicameral, Nebraska.gov, https://www.nebraska.gov/government/legislative [https://perma.cc/87MS-XRWK]. 35 percent of the counties in the typical (median) state are bigger than the corresponding house districts. If districts doubled in size, however, only 18 percent of the counties in the median state would exceed the size of a district; if the district size increased by a factor of three, only 12 percent of the median state’s counties would exceed the size of a district.107These results reflect a comparison of state house district sizes in the 2010s and county sizes. See 2010 Constituents Per State Legislative District Table, Nat’l Conf. St. Legislatures, https://www.ncsl.org/research/about-state-legislatures/2010-constituents-per-state-legislative-district.aspx [https://perma.cc/799Y-NQ7B]; County Population Totals: 2010–2019, U.S. Census Bureau, https://www.census.gov/data/datasets/time-series/demo/popest/2010s-counties-total.html [https://perma.cc/MP8C-7Q9E]. These numbers suggest that judges would break up significantly fewer counties under multimember RCV because districts could be drawn to include more voters.
The upshot of all this is that whether in its at-large or multimember form, RCV requires much less ingenuity from the court even as it provides a remedy. Rather than ask a judge to decide which voters belong to the same community as defined by certain district lines, multimember RCV permits larger districts that allow voters to join, or not join, in coalition as they choose.108See Richard Briffault, The Local Government Boundary Problem in Metropolitan Areas, 48 Stan. L. Rev. 1115, 1156–57 (1996) (describing proportional voting schemes like RCV as a way to let voters rather than map drawers decide on jurisdictional boundaries). Since multimember RCV does not dilute minority votes but still keeps courts away from precarious fact-specific inquiries, courts should order the practice as a remedy in redistricting cases.
One final clarification is in order. Multimember RCV is not the only election method that simultaneously avoids single-member districts and remedies vote dilution. Multimember RCV avoids vote dilution by collecting second and third preferences. But other multimember election schemes solve vote dilution by collecting different kinds of information: for instance, in Europe, many proportional systems collect voters’ preference for party.109See Issacharoff et al., supra note 85, at 1213–22. This Note focuses on multimember RCV for several reasons. In the broader universe of multimember election methods, only three election types allow voters to cast votes for candidates (as opposed to parties);110See Richard Briffault, Lani Guinier and the Dilemmas of American Democracy, 95 Colum. L. Rev. 418, 432–41 (1995) (reviewing Lani Guinier, The Tyranny of the Majority: Fundamental Fairness in Representative Democracy (1994)); Mulroy, supra note 92, at 339–43. American law and tradition approves only of voting for candidates.111See Issacharoff et al., supra note 85, at 1213 (explaining that the United States practices majoritarian democracy, where candidates earning the most votes win). Many state constitutions prescribe elections that select a majority winner or plurality winner. See, e.g., Tex. Const. art. IV, § 3. This would immediately foreclose elections where a party was on the ballot, not a candidate. Of these three—RCV, cumulative voting, and limited voting112Cumulative voting has two key features: First, a voter has as many votes as there are seats to fill. Second, voters can assign more than one vote to a candidate. In Mobile, then, each voter would receive three votes and could allocate up to three votes to a single candidate. Since Mobile’s Black community is bigger than the threshold of exclusion in a three-candidate race, the Black community could elect a candidate of choice if it were to pool its votes behind a single candidate. See Richard H. Pildes & Kristen A. Donoghue, Cumulative Voting in the United States, 1995 U. Chi. Legal F. 241, 253–54. Limited voting works like cumulative voting, except that voters receive less votes than there are seats to fill. Id.—multimember RCV has received the most recent sustained attention from the press as well as from jurisdictions contemplating a shift in election method.113See supra text accompanying notes 25–31. This makes it most compatible with the doctrines discussed in Section II.B, which sensibly take account of whether a jurisdiction is familiar with a remedial election method. Setting aside popularity, however, this Note takes no strong stand on the merits of multimember RCV relative to these other methods—cumulative voting or limited voting—as a remedy.
B. Redistricting Doctrine and the Errors of a Junior-Partner Approach
American law resists multimember RCV as a remedy not only because of concerns about vote dilution but because of its particular view of the relationship between courts and legislatures. The law calls for courts to craft a remedy that the legislature might plausibly have created itself.114White v. Weiser, 412 U.S. 783, 795 (1973) (explaining that the remedy should honor state policy by embodying the redistricting principles that the state legislature itself follows when redistricting). Usually, that remedy is single-member districts with SCV. This Section applies the doctrine to multimember RCV and situates it in broader context. It then critiques the quasi-legislative choices that this approach forces upon courts.
The Supreme Court has explained that a court’s redistricting remedy must not “intrude upon state policy any more than necessary.”115Whitcomb v. Chavis, 403 U.S. 124, 160 (1971). The Court elaborated, in White v. Weiser, that courts must not “pre-empt the legislative task” and must “honor state policies.”116412 U.S. at 795. As for how to do so, the Court instructed lower courts to consult the priorities and policies as expressed in “statutory and constitutional provisions or . . . the reapportionment plans proposed by the state legislature.”117Id. We can understand this as a “hypothetical intent” test: the court inquires into what the legislature might have done had it created an election system without the invalid element.118E.g., Cane v. Worcester Cnty., 35 F.3d 921, 928 (4th Cir. 1994) (“[T]he court, in exercising its discretion to fashion a remedy that complies with § 2, must to the greatest extent possible give effect to the legislative policy judgments underlying the current electoral scheme or the legally unacceptable remedy offered by the legislative body”); cf. Emily Sherwin, Rules and Judicial Review, 6 Legal Theory 299, 305 (2000) (describing a similar inquiry, about severability, as a hypothetical intent test).
This understanding of deference is not unique to redistricting. In many areas of law, courts defer to a legislature.119The most prominent example is statutory interpretation. See Richard H. Fallon, Jr., On Viewing the Courts as Junior Partners of Congress in Statutory Interpretation Cases: An Essay Celebrating the Scholarship of Daniel J. Meltzer, 91 Notre Dame L. Rev. 1743 (2016); infra Section III.A. And, in many areas of law, courts have deferred precisely as the Court in Weiser advised by seeking to carry out the legislative purpose.120Id. at 1746 (describing the junior-partner approach). This approach is familiar from purposivist methods of statutory interpretation,121See id. at 1745. and it calls to mind the “junior partner” model of the courts.122See id. at 1746. That is, under purposivism as well as remedial redistricting doctrine, courts act as partners of the legislature to fill in gaps in a statutory scheme. In the process, judges may exercise some of their own ingenuity, practical judgment, and discretion. That is not only tolerable, but commended: as junior partners, courts are trusted collaborators with the legislature.123Id. at 1745.
By adopting a junior-partner approach to redistricting, the courts have rejected what scholars have called the “faithful agent” approach.124Id. We may recognize this theory from one of its instantiations, textualism. Under a faithful-agent approach, courts should not inquire into legislative intent and purpose. These indicia are too vague to offer guidance. Courts better respect the limits of the judicial role by adopting decision rules, often formalistic ones, that prevent courts from acting as legislatures themselves.125Id. at 1745–76. Multimember RCV, this Note argues, is a useful remedy because it provides such a decision rule.126Admittedly, the argument for RCV does not echo another central justification for textualism: that textualism respects what the law is, where “law” is understood to be only the text that survives bicameralism and presentment. Id. at 1773.
1. The Current Doctrine Applied to RCV
The Supreme Court has never weighed in on multimember RCV as a remedy.127The closest the Court has come is a concurrence by Justice Thomas. Holder v. Hall, 512 U.S. 874, 910 (1994) (Thomas, J., concurring). Justice Thomas spoke favorably about the legality of multimember RCV, but only in an effort to illustrate his criticisms of section 2 of the Voting Rights Act. He meant that current law would be amenable to defining vote dilution not only in reference to single-member districts but also proportional remedies like RCV. Id. But we can readily apply the Court’s junior-partner approach to remedial multimember RCV. In the typical case, where a legislature passed an invalid single-member district map, the court would seek to recreate state policy and therefore recreate a single-member map. Multimember RCV would fail not only because it swapped in a multimember map where the legislature wanted a single-member map. It would fail because—except in a very unusual case—the legislature has a preference for SCV, not RCV.128See Where Ranked Choice Voting Is Used, FairVote, https://www.fairvote.org/where_is_ranked_choice_voting_used [https://perma.cc/NQM7-SE5L] (providing a map of jurisdictions that use a version of ranked-choice voting). Evidently, most jurisdictions do not use RCV. Imposing multimember RCV on a legislature that never requested it would, very clearly, intrude too much on state policy. A court doing so would not be acting as the hypothetical legislature would.
The junior-partner approach would not disqualify multimember RCV in all cases. Where the legislature passed an invalid multimember map, multimember RCV might be tempting as a way to respect a legislature’s status quo preference for multimember elections. Several district courts have ordered multimember cumulative voting for just this reason.129E.g., Mo. State Conf. of the NAACP v. Ferguson-Florissant Sch. Dist., 219 F. Supp. 3d 949 (E.D. Mo. 2016); Cottier v. City of Martin, 475 F. Supp. 2d 932 (D.S.D. 2007), aff’d, 551 F.3d 733 (8th Cir. 2008), and vacated, 604 F.3d 553 (8th Cir. 2010); Cane v. Worcester Cnty., 847 F. Supp. 369 (D. Md. 1994), aff’d in part, rev’d in part, 35 F.3d 921 (4th Cir. 1994); see supra note 112 for a description of cumulative voting. In 2016, for instance, after finding that Ferguson, Missouri’s school districts violated the Voting Rights Act, a district judge ordered multimember cumulative voting to preserve the city council’s existing multimember elections.130Ferguson-Florissant Sch. Dist., 219 F. Supp. 3d at 962.
But multimember remedies for a multimember status quo are unusual. Courts usually put forward single-member districts to replace multimember districts without even explaining the choice.131See Guinier, supra note 83, at 1098 (describing the often rote association of single-member districts with section 2 violations). When courts have addressed the issue head-on, they have suggested that forcing a legislature to adopt an alternative scheme for tabulating votes, like RCV, would be too grave an intrusion—even if the intrusion also preserves a multimember status quo. In Harper v. City of Chicago Heights, for example, the Seventh Circuit reversed a court order for multimember cumulative voting, noting that “[p]rior to the district court’s order, the parties had never thought of cumulative voting.”132223 F.3d 593, 602 (7th Cir. 2000). In a similar case, the Eleventh Circuit put matters more strongly: “The federal courts have no authority to conjure up such an election scheme and impose it on a state government, regardless of the theoretical prospect of increasing minority voting strength.”133Dillard v. Baldwin Cnty. Comm’rs, 376 F.3d 1260, 1268 (11th Cir. 2004). Under a junior-partner vision, in short, the imposition of multimember RCV or another alternative voting scheme unacceptably contravenes the preferences of the legislature whose map needs remedying.
2. Criticism of the Current Doctrine
Part and parcel of the junior-partner approach to redistricting is the fact that courts will exercise their discretion and practical judgment to fill out a statutory scheme.134Fallon, supra note 119, at 1745–47. The junior-partner approach indeed envisions courts as trusted partners. But the pitfall of a junior-partner approach is that courts may play more than a “junior” role; they might become major decisionmakers. Under current doctrine, courts are asked to draw new maps and make enormous numbers of unguided, quasi-legislative, and potentially arbitrary decisions. There are significant drawbacks to the current approach.135Part III emphasizes two of these drawbacks in greater detail. First, a junior-partner approach is out of step with contemporary doctrine, which generally rejects legislative intent. Second, this approach involves the court in difficult decisions about the scope of racial identity.
One way to appreciate how major the role of the court is under current doctrine is to revisit the case of the multimember status quo. Return to the case of Mobile in 1980, in which the votes of Black voters were diluted using single-choice voting in multimember districts.
The current doctrine would ask the courts to order the remedy that Mobile might have created itself. But this ultimately requires that courts speculate about questions that Mobile’s City Commission never answered. Mobile’s past map expressed a preference for multimember elections and SCV.136City of Mobile v. Bolden, 446 U.S. 55, 59 (1980), superseded by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97-205, sec. 3, § 2, 96 Stat. 131, 134. But the vote dilution claim requires that Mobile choose between its multimember elections and SCV. Would Mobile prefer to retain its familiar ballot but subdivide its territory into single-member districts? Or would Mobile rather retain its multimember elections but adopt an unfamiliar RCV ballot? The junior-partner approach trusts courts to guess without much guidance.
The same absence of structure reappears in more alarming form when courts seek to draw new single-member districts. When courts determine that “state policy” advises in favor of single-member districts, the junior-partner model advises courts to design the map that the legislature would have drawn itself. To do so, courts attempt to make changes where necessary to fix the illegal map.137Persily, supra note 1, at 1136–38. Courts also look for guidance in state policy on redistricting—in the constitutional provisions, statutes, and legislative committee reports that explain how a state believes districts should look.138See Perry v. Perez, 565 U.S. 388, 395 (2012) (per curiam).
In the example of Texas in the 2010s, for instance, the Supreme Court instructed the district court to hew closely to state policy as embodied in the previous map, moving lines and districts only where necessary to cure that plan’s legal defects.139Id. at 395–96. The Court encouraged the district court to rely on redistricting criteria from state law as well.140Id. These criteria often include compactness, contiguity of districts, and respect for the existing boundaries of political subdivisions.141Districting Principles for 2010 and Beyond, Nat’l Conf. St. Legislatures, https://www.ncsl.org/research/redistricting/districting-principles-for-2010-and-beyond.aspx [https://perma.cc/67QU-S74X]. See infra text accompanying notes 145–148 for a discussion of compactness. Contiguity of districts refers to the idea that “one can walk to each part of the district without having to go through another district.” Persily, supra note 1, at 1159. The principle of respect for political subdivisions means that a district should avoid splits of cities and counties. Id.
Yet how much guidance can courts truly elicit from state policy about how to draw districts? The previous map discloses to a court where the legislature wanted to draw lines. But it cannot tell a court how to choose among the different ways to draw new lines. Curing a redistricting violation means adding and subtracting a group of voters from an invalid district.142The principle of one-person, one-vote fixes the size of each district in a jurisdiction at the same level. Reynolds v. Sims, 377 U.S. 533 (1964). If a district receives new voters without any corresponding subtraction, and assuming the district’s population was previously at the required level, the district will have become unlawfully big. A court will need to enlist its own judgment to decide among different ways to add and subtract. Furthermore, because every district must have equally sized populations, a court cannot take voters out of the invalid district and put them in a neighboring district without, in turn, subtracting a new round of voters from that neighboring district, and so on.143Id. at 622. These cascading effects may radiate outwards from the invalid district, requiring courts to make not only minor tweaks to the district in question but also changes throughout the redistricting map.144Cf. Fletcher, supra note 59, at 645 (identifying redistricting as a “polycentric” problem, where each subproblem is related to all the other subproblems, “such that the solution to each depends on the solution to all the others”).
As for the guidance that courts can expect to find in state redistricting policies, the principles themselves are undefined and conflict with each other. Compactness, for instance, is the widely accepted idea that districts should look square shaped and “regular.”145See, e.g., Aaron Kaufman, Gary King & Mayya Komisarchik, How to Measure Legislative District Compactness if You Only Know It When You See It, Am. J. Pol. Sci. (forthcoming) (manuscript at 1), https://gking.harvard.edu/files/gking/files/compact.pdf (explaining that compactness is the idea that districts should look “regular”). Neither experts nor courts can agree on precisely how to measure compactness, and the various definitions in use often contradict each other.146Persily, supra note 1, at 1159–60. Compactness can be defined, for instance, as the ratio of the area of the district to the area of the smallest circle, square or rectangle that circumscribes the district. It has also been defined as the ratio of the perimeter of the district to the area of the district. A measure might depend, as well, on the number of peninsulas or “protrusions” that emerge from a district. All of these measures are plausible but also mutually exclusive, as a district may score as highly compact along one measure but not compact on another. Id. at 1159. What’s more, compactness can conflict with other legally mandated redistricting goals. For example, the Voting Rights Act prohibits racial vote dilution, and racial vote dilution often entails “packing” voters of color into one district in order to confine their influence to only that one district.147See supra text accompanying notes 51–54. But because many voters of color live in denser urban areas, even a compact redistricting scheme may lead to city districts “packed” with voters of color. By contrast, a scheme with oddly shaped district maps can make the votes of marginalized communities appropriately strong.148Persily, supra note 1, at 1159–60. Suppose that a city with a large Black population is surrounded by suburbs with a significant Black minority. A compact districting scheme might draw one district around the city, and other districts around the suburbs. This would have the effect of making Black voters a majority in the city district, and a minority in the suburban districts. But it would also waste many Black votes: Black voters, if they voted together, would win by landslides in the city, and lose narrowly in each suburb. A more efficient but noncompact districting scheme could draw each district so that it looked like a wedge. Each wedge would include parts of the city and parts of the surrounding suburbs. This arrangement could make Black voters a majority in many more districts and might even be required under the Voting Rights Act. Thornburg v. Gingles, 478 U.S. 30, 49–51 (1986) (requiring that courts draw majority–minority districts, where a minority is geographically compact, where there is racially polarized voting, and where other contextual factors are met).
A final sign of the major role that courts have come to play under a junior-partner model is related to personnel. To redistrict, courts usually hire not only a special master, but also experts on local political geography, technicians, computer specialists, and “a staff of people producing and handling the enormous amount of documentation that often accompanies a plan.”149Persily, supra note 1. For a description of typical special master duties in redistricting cases, see, for example, Taft Wireback, State Must ‘Pay Up’ for Special Master in Gerrymandering Suit, Judges Say, News & Rec. (Mar. 27, 2018), https://greensboro.com/news/state/state-must-pay-up-for-special-master-in-gerrymandering-suit-judges-say/article_f7776451-1736-5610-b8cb-1bd329fa8630.html [https://perma.cc/X47V-VG25] (describing Stanford Law professor who was hired as a special master in advising about redistricting in North Carolina). A court transfigures into a small administrative agency or legislative office in order to achieve a new redistricting map.150Cf. Fed. Radio Comm’n v. Gen. Elec. Co., 281 U.S. 464, 467, 469 (1930) (“We think it plain . . . that the powers confided to the commission . . . are purely administrative and that the provision for appeals to the Court of Appeals does no more than make that court a superior and revising agency in the same field. . . . But this Court cannot be invested with jurisdiction of that character, whether for purposes of review or otherwise. It . . . is invested with judicial power only . . . . It cannot . . . exercise or participate in the exercise of functions which are essentially legislative or administrative.”).
III. In Support of Multimember RCV as a Remedy
A junior-partner approach accepts that remedial redistricting relies heavily on courts’ ingenuity. A faithful-agent approach would prefer a strict framework that prevents courts from making unguided policy decisions. This Part argues in favor of a faithful-agent approach to redistricting and multimember RCV as a remedy, for two main reasons. First, multimember RCV as a remedy would align redistricting doctrine with broader currents in the law that have come to emphasize a faithful-agent understanding of the courts. Second, multimember RCV helps courts avoid weighing in on the boundaries of racial identity.
A. Multimember RCV as Best Aligned with the Surrounding Law
The first reason that courts should adopt multimember RCV as a remedy, as well as a faithful-agent approach to redistricting, is that it would align redistricting doctrine with the dominant view of the proper relationship between courts and legislatures. The debate between junior-partner and faithful-agent approaches has played out most thoroughly in the field of statutory interpretation.151See Fallon, supra note 119, at 1745–48. And the trends there point strongly in a faithful-agent direction.152See, e.g., Harvard Law School, The 2015 Scalia Lecture: A Dialogue with Justice Elena Kagan on the Reading of Statutes, YouTube, at 08:29 (Nov. 25, 2015), https://www.youtube.com/watch?v=dpEtszFT0Tg (“We’re all textualists now.”); Abbe R. Gluck & Richard A. Posner, Statutory Interpretation on the Bench: A Survey of Forty-Two Judges on the Federal Courts of Appeals, 131 Harv. L. Rev. 1298, 1304–05 (2018) (finding that many younger judges “unquestionably have been influenced by changes in legal education and statutory litigation over the past three decades”). As this Section shows, moreover, these trends have already left their mark on one quasi-remedial doctrine: severability. Remedial redistricting law should take the cue and change as well.
Through the 1970s, judges approached statutory interpretation on a theory that the courts have since rejected.153See Jeremy K. Kessler & David E. Pozen, Working Themselves Impure: A Life Cycle Theory of Legal Theories, 83 U. Chi. L. Rev. 1819, 1848 (2016). But that theory lives on in today’s remedial redistricting doctrine. In the previous era of statutory interpretation, judges read statutes so as to effect state policy.154See Fallon, supra note 119, at 1746 (explaining that courts sought to help legislatures develop a “just and workable body of law”). They inquired into the goals and logic of statutes.155Id. at 1746–47. They operated as partners of the legislature, conceiving of their judicial task in quasi-legislative terms.156Id.; Gluck & Posner, supra note 152, at 1303. The seminal cases on redistricting remedies emerged at the same time, during this pre-textualist era of statutory interpretation.157The Court decided White v. Weiser in the early 1970s. 412 U.S. 783 (1973). According to William Eskridge, the tide began to change in the 1980s. William N. Eskridge, Jr., The New Textualism, 37 UCLA L. Rev. 621, 646 (1990).
Judges and scholars today have moved in a textualist direction—“we are all textualists now”158Harvard Law School, supra note 152, at 08:29.—in part because of a preference for constraints on judicial discretion over adherence to state policy. Reading a statute so as to carry out its intent was said to empower judges to make policy; it was akin to “looking over a crowd and picking out your friends.”159Patricia M. Wald, Some Observations on the Use of Legislative History in the 1981 Supreme Court Term, 68 Iowa L. Rev. 195, 214 (1983). Statutes and their legislative histories may contain multiple and contradictory purposes.160E.g., Joseph A. Grundfest & A.C. Pritchard, Statutes with Multiple Personality Disorders: The Value of Ambiguity in Statutory Design and Interpretation, 54 Stan. L. Rev. 627, 628 (2002) (suggesting that statutes will often have conflicting or ambiguous purposes because such ambiguity greases the wheels of legislative deliberation). A method of statutory interpretation that pursues legislative purpose counts on a judge to resolve these tensions in a sensible way.161See supra text accompanying notes 128–130. But the process of resolution requires something akin to policy judgment. It might allow, or even require, that judges “write their own preferences into the statute” by looking out over available readings and picking their preferred one.162Eskridge, supra note 157, at 648; see also Kessler & Pozen, supra note 153, at 1848–49.
This critique of statutory intent bears an obvious resemblance to the criticisms Section II.B levied at the contemporary approach to redistricting. Under contemporary doctrine, judges seek to draw the map that the legislature might have drawn. But that legislative intent proves hard to divine, just as statutory purpose is hard to divine. Courts are empowered to make quasi-legislative decisions according to their own best judgment about where to place district lines.163See supra Section II.B.2.
Textualists’ alternative to intent mirrors this Note’s solution to the problem of court-drawn districts. Textualists advise that a judge respect the separation of powers by focusing only on the concrete question of what the statute’s words actually mean.164Eskridge, supra note 157, at 648. By emphasizing text at the expense of legislative intent, textualism puts a thumb on the scale in favor of interpretations that may conflict with a statute’s goals.165Cf. Sherwin, supra note 118, at 310 (“[E]ven a good rule can, and almost certainly will, produce errors in some of its applications. Rules are blunt, because they do not restate the moral principle on which they depend, but instead reduce these principles to a set of instructions about what to do in specified circumstances.”).Textualists accept friction with state policy as the cost of an interpretive method that avoids the policymaking pitfalls of an intent-focused approach.166See Fallon, supra note 119, at 1746 (suggesting that textualism requires a certain indifference to whether the outcomes of an interpretation produce a “just and workable body of law”). As an example of how pure textualist commitments may undermine state policy, consider King v. Burwell, 576 U.S. 473 (2015). Chief Justice Roberts and the Court’s majority held that the Affordable Care Act (ACA) granted tax credits not only to those who enrolled in healthcare plans via state-run marketplaces, as the plain language of the Act seemed to suggest, but to those who enrolled via federal marketplaces. The majority did so in light of the ACA’s purpose to expand health coverage. Id.. Justice Scalia’s opinion, striking a more traditionally textualist note, would have pursued plain meaning even at the risk of undermining the ACA’s federal marketplaces. Id. at 499 (Scalia, J., dissenting); see also Fallon, supra note 119, at 1770 (offering the majority opinion, to which Justice Scalia objected, as an example of a junior-partner approach to statutory interpretation). A multimember RCV remedy likewise sacrifices fit with state policy for the sake of rule-like constraints on the judge.
The trend towards faithful-agent approaches has not meant that courts must discard legislative intent entirely. Where the alternative to legislative intent is plainly implausible, courts may continue to rely on it. For instance, in another area of remedies, severability, courts still foreground an intent-based question: how a legislature would choose to rewrite a statute after a court invalidates one of its provisions.167David H. Gans, Severability as Judicial Lawmaking, 76 Geo. Wash. L. Rev. 639, 640, 643 (2008) (describing the precarious act of reconstruction that the Court undertook in United States v. Booker, 543 U.S. 220 (2005)). Justice Thomas has disparaged the severability analysis in familiar, faithful-agent terms. Severability involves the Court taking a “blue pencil” to statutes as part of a “nebulous inquiry into hypothetical congressional intent.”168Murphy v. Nat’l Collegiate Athletic Ass’n, 138 S. Ct. 1461, 1486 (2018) (Thomas, J., concurring) (quoting United States v. Booker, 543 U.S. 220, 320 n.7 (2005) (Thomas, J., dissenting in part)). And he is not wrong that the inquiry into legislative intent is difficult and indeterminate, since Congress does not always leave clear instructions about how a court should reconstruct a partially invalidated statute.169Reflective of the scholarly mood is Emily Sherwin’s article, which argues (or, simply, takes for granted) that the “search for legislative intent—that is, intent about specific severability problems that arise in constitutional adjudication—will normally be fruitless.” Sherwin, supra note 118, at 307. But severability doctrine persists,170Abbe R. Gluck, “A Scalpel Rather than a Bulldozer”: Severability Is in the Spotlight as the Newest ACA Challenge Looms, SCOTUSblog (July 28, 2020, 10:33 AM), https://www.scotusblog.com/2020/07/a-scalpel-rather-than-a-bulldozer-severability-in-the-spotlight-as-the-newest-aca-challenge-looms [https://perma.cc/2KAE-UEF4]. in part because the alternative is hard to imagine. As Michael Dorf has argued: “[W]ould the invalidation of some snippet of text imply the invalidation of everything else that was enacted as part of the same omnibus bill as that snippet?”171Michael C. Dorf, Fallback Law, 107 Colum. L. Rev. 303, 370 (2007).Courts have no choice but to reconstruct intent during a severability inquiry, in other words, because the more mechanical alternative—a blunt invalidation of the entire statute—seems patently unacceptable.
The central virtue of multimember RCV is that it provides courts an entirely plausible alternative to the search for intent. A court that applies multimember RCV enacts a rule-like remedy, consistent with the current vision of the relationship between courts and legislatures. But this remedy is also a manageable and workable substitute for single-member districts.
B. Multimember RCV as a Restraint on the Courts in Issues of Race
A second reason to prioritize a faithful-agent approach to redistricting, and multimember RCV in particular, is that multimember RCV allows courts to sidestep constitutionally and politically fraught questions about the scope of a racial community. A multimember RCV remedy offers a remedial bargain. By imposing an unfamiliar balloting system onto a legislature, a court can avoid the task of defining racial communities. The first imposition is not like the second. The Constitution provides no special protection for single-member districts or the single-choice vote.172The Constitution provides that state legislators determine the time, place, and manner of elections to Congress. U.S. Const. art. I, § 4. But this clause does not bear on elections for state office. Moreover, it cannot answer the question of how a court—seeking to intrude as little as possible on the state legislature’s control over elections—should design a redistricting remedy. There is also a long tradition of multimember elections in American politics. E.g., City of Mobile v. Bolden, 446 U.S. 55, 60 (1980) (opinion of Stewart, J.) (explaining that “literally thousands of municipalities and other local governmental units throughout the Nation” use multimember elections), superseded by statute, Voting Rights Act Amendments of 1982, Pub. L. No. 97-205, sec. 3, § 2, 96 Stat. 131, 134. By contrast, drawing districts requires courts to juggle a range of race-related harms that are cognizable under the Equal Protection Clause and the Voting Rights Act. Given the difficulty of the balancing that districting requires, and the significance of the racial harms at stake, it is all the more important to reserve these decisions for legislatures, not courts.
Drawing districts threatens at least two serious race-based harms: one of constitutional status, and one protected by the Voting Rights Act. The first limitation, of constitutional status, is that map drawers cannot too obviously take race into account when drawing districts, lest a map send the message that voters are defined only by their race.173Shaw v. Reno, 509 U.S. 630, 644 (1993) (holding that strict scrutiny applies when “redistricting legislation  is so bizarre on its face” that it cannot be explained except by reference to race) (citing Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 266 (1977)); see also Richard H. Pildes & Richard G. Niemi, Expressive Harms, “Bizarre Districts,” and Voting Rights: Evaluating Election-District Appearances After Shaw v. Reno, 92 Mich. L. Rev. 483, 501 (1993) (explaining that Shaw addresses the problem of districts that send a “signal” to “government officials [that] race has become paramount and dwarfed all other, traditionally relevant criteria”). In a line of cases beginning with Shaw v. Reno, the Supreme Court has held that the shape of a district cannot suggest that the district was drawn solely or predominantly for the purpose of encircling voters of color.174Richard H. Pildes, Principled Limitations on Racial and Partisan Redistricting, 106 Yale L.J. 2505, 2510–11 (1997). If district lines take too many twists and turns, look too gerrymandered or “highly irregular” and “bizarre,”175Shaw, 509 U.S. at 646, 655–56. the district broadcasts a message of racial essentialism. It suggests that “members of the same racial group—regardless of their age, education, economic status, or the community in which they live—think alike, share the same political interests, and will prefer the same candidate at the polls.”176Id. at 647. This harm is serious enough to trigger strict scrutiny.177Id. at 644.
Shaw thus proscribes courts—or any map drawer—from taking race too much into account when drawing districts. But courts engaged in redistricting cannot ignore race either. The Voting Rights Act requires that map drawers empower minorities to elect candidates of choice when possible.178Thornburg v. Gingles, 478 U.S. 30, 49–51 (1986). Usually, this involves drawing single-member districts in which a community of color can constitute a majority (a so-called majority-minority district).179See supra Section II.A. Hence, a court engaged in redistricting always walks a fine line between relying on race too much and too little—between a VRA violation that consists in racial vote dilution and a constitutional violation of excessive racial gerrymandering. One judge has analogized complying with Shaw and the VRA to making the passage between Scylla and Charybdis.180Harper v. City of Chicago Heights, 223 F.3d 593, 599 (7th Cir. 2000) (describing a multimember cumulative voting remedy as an attempt to “steer between the Scylla of racially based district lines . . . and the Charybdis of ineffectual Section 2 remedies” (citations omitted)). On all sides in the redistricting process looms the risk of cognizable racial harms.
Even when districts neither violate Shaw nor the VRA, moreover, districts may disturb norms that are adjacent to constitutional bars. In particular, the process of districting may require that judges decide on the boundaries of racial communities. Whenever a community of color is too big to fit in one single-member district, the map drawer must decide who belongs together and who does not. Traditional redistricting principles will sometimes provide other bases on which to draw the lines. But because these principles are so often undetermined, the map drawer will often be left to make decisions about who constitutes a racial community.181See supra Section II.B.2 (emphasizing the indeterminacy of redistricting principles). The Supreme Court has not squarely labelled this sort of undertaking constitutionally suspect.182But see League of United Latin Am. Citizens v. Perry, 548 U.S. 399, 424 (2006) (rejecting, under the VRA, the state’s attempt to draw a district around different Latino populations that did not share in common class, education, health, and other markers of community). And the act of identifying racial communities of interest is, of course, not categorically bad; on the contrary, remedying histories of discrimination requires it.183Cf. Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 245 (1995) (Stevens, J., dissenting) (suggesting that to ignore that histories of discrimination are experienced by particular groups is akin to “disregard[ing] the difference between a ‘No Trespassing’ sign and a welcome mat”). But doing so requires sensitivity to nuance, careful use of data, and skepticism of rigid generalizations.184Lani Guinier, The Tyranny of the Majority: Fundamental Fairness in Representative Democracy 255 (1994) (warning that districting can require the application of rigid or essentializing views). One example of a term that has proved overbroad in perhaps harmful ways is the “Latino vote.” Around the 2020 election, scholars of Latino politics worked to disabuse commentators of the view that there was a single “Latino vote.” Jasmine Aguilera, Why It’s A Mistake to Simplify the ‘Latino Vote,’ Time (Nov. 10, 2020, 7:53 AM), https://time.com/5907525/latino-vote-2020-election [https://perma.cc/PU94-RDEH] (describing social scientists whose decades of study have equipped them to understand and honor the nuances of Latino voting behavior); see also Theodore R. Johnson, Don’t Fall for the Mythic ‘Black Voter’ Analysis After South Carolina, Brennan Ctr. Just. (Feb. 28, 2020) (citing Ryan Pougiales & Jessica Fulton, A Nuanced Picture of What Black Americans Want in 2020, Third Way (Dec. 30, 2019), https://www.thirdway.org/memo/a-nuanced-picture-of-what-black-americans-want-in-2020 [https://perma.cc/K4TQ-UFHA]), https://www.brennancenter.org/our-work/analysis-opinion/dont-fall-mythic-black-voter-analysis-after-south-carolina [https://perma.cc/TP9Y-LKCE] (cautioning against treating Black voters as a monolithic group). There is good reason to believe that the federal judiciary, especially the current one, has neither the competence nor the ideological inclination to handle this delicate subject effectively.185The contemporary federal bench is not a beacon of racial justice. See, e.g., Shelby Cnty. v. Holder, 570 U.S. 529, 531 (2013) (rendering inoperative section 5 of the Voting Rights Act because “things have changed dramatically” in the South); Rebecca R. Ruiz, Robert Gebeloff, Steve Eder & Ben Protess, A Conservative Agenda Unleashed on the Federal Courts, N.Y. Times (Mar. 16, 2020), https://www.nytimes.com/2020/03/14/us/trump-appeals-court-judges.html [https://perma.cc/3JYG-VX48]. What’s more, the law has frowned upon careful use of the very tools that would help the courts develop more sensitive understandings of racial community, such as sociological or social scientific evidence. Oliver Roeder, The Supreme Court Is Allergic to Math, FiveThirtyEight (Oct. 17, 2017, 6:00 AM), https://fivethirtyeight.com/features/the-supreme-court-is-allergic-to-math [https://perma.cc/Y8H5-28R7] (recounting when Chief Justice Roberts referred to the use of data in redistricting litigation as “sociological gobbledygook”).
All this is to say that the act of drawing districts requires that judges balance, avoid, and navigate around various racial harms. Some have constitutional status, such as the prohibition against making race the dominant motive when redistricting. Others flow from landmark civil rights legislation, the Voting Rights Act. And others are adjacent to constitutional norms, namely, the concern about the competence of federal judges to decide how to divide up racial communities. Multimember RCV may impinge on a legislature’s choices as between balloting methods. But a practice of court-ordered multimember RCV has the decided virtue of saving for the legislature, and keeping away from courts, fact-intense, higher-stakes decisions about racial identity.
As for how exactly multimember RCV helps avoid these problems: multimember RCV does not entirely absolve courts of the responsibility for drawing districts, as Part II explained. But multimember RCV permits courts to draw fewer districts of much greater sizes, while respecting other preexisting districting norms.186See supra Section II.A. A multimember RCV district might have two, three, or four times the population of a typical single-member district. This lessens the impact of both problems discussed above.
RCV with multimember districts means that courts will less often face the scenario that triggers the tension between the VRA and Shaw. In this sort of scenario, voters of color live somewhat far apart—say, on the opposite sides of a city.187In Shaw, the congressional district in question connected Black communities separated by as much as 160 miles. Shaw v. Reno, 509 U.S. 630, 635–36 (1993). But districts that encompass smaller populations than congressional districts do might well raise the Shaw issue within one municipality. The map drawer can draw, unartfully, a single district that includes both communities, at the risk of violating Shaw. Alternatively, the map drawer can keep the communities separate, at the risk of violating the VRA, which might require that these communities occupy a single district.188See supra text accompanying notes 51–52. Multimember RCV permits districts that are at least twice the size—and perhaps five times the size—of a single-member district. As districts grow in size, more and more decisions about whether or not to include two distantly separated communities become moot. This is why proponents of multimember RCV and similar systems have said that RCV would allow voters “to coalesce along whatever lines they wish without requiring the state to involve itself in these groupings.”189Heather K. Gerken, Understanding the Right to an Undiluted Vote, 114 Harv. L. Rev. 1663, 1729–30 (2001).Rather than asking judges to decide on district lines, voters decide who they belong with based on their own votes.
Multimember RCV will, likewise, moot many decisions about how exactly to divide up a particular community of color. A map drawer needs to make decisions about who constitutes a racial community whenever the number of voters of color is too big to fit in one district. But the bigger a district, the more likely it is that a community of color will fit squarely inside a single district, and the less likely it is that a judge will need to make the decision of who belongs in and who belongs out.190See supra text accompanying note 108 (making a similar argument about divisions of county lines).
When the Supreme Court took on abuses of redistricting in the 1960s, the Court appeared to face an unappealing choice at the remedial stage: order multimember districts at the risk of causing vote dilution, or order single-member maps and risk the fraught enterprise of line drawing. However, the courts can cure vote dilution without drawing a district for every representative elected through the use of multimember RCV. As doctrinal climates change, and the federal bench retrenches on matters of race, we should look to a remedy during the next redistricting cycle that places only mild remedial demands on the federal judge.