Here is a guest post from Wendy K Tam Cho, Yan Y. Liu, Emily R. Zhang, & Bruce E Cain:
In addressing the sometimes impossible problem of having to choose between sesame, poppy seed, onion, garlic or salt, the everything bagel presents a nice, well-rounded solution. Beyond breakfast, it also offers some inspirations for partisan gerrymandering doctrine. Plaintiffs and amici in Whitford v. Gill, Benisek v. Lamone and Common Cause v. Rucho have suggested a number of possible measures of partisan effect. For instance, plaintiffs in Whitford have put forth the efficiency gap as advocated for by Nick Stephanopoulous and Eric McGhee in their University of Chicago Law Review article, and in Rucho, plaintiffs have also advanced the mean-median difference and partisan bias. Elsewhere, in the political science literature, there are other proposed measures of partisan fairness. Scholars have proposed competitiveness measures as well as ways to quantify the mismatch between the number of votes a party receives and the number of seats they win.
Setting aside the mechanics of each of these measures, is there, in fact, a single silver-bullet partisan effects measure? Partisan fairness is a complex democratic ideal that encompasses many different values. Can we expect a single, simple measure to take all those values into account? Might an everything-bagel approach to partisan fairness be more appropriate? An analogy to health is instructive here. Medical professionals do not expect to discover a single measure of health. A patient whose doctor’s visit consists of a sole demand to open up her mouth and say “aah!” would be greatly puzzled (and overcharged). While there are disputes over the relative weight to accord each health indicator—longevity, mobility, functionality, enjoyment—it is generally assumed that health is a meta-ideal that can be manifested (and hence measured) in many different ways.
Political fairness is a similarly multidimensional—and if anything—a more heavily contested concept. It encompasses many values, and hence many different ways to measure injury to those values. The currently proposed partisan effects measures capture some of those values, but not all. Both the efficiency gap and partisan bias measure are concerned about fairness vis-à-vis the two political parties: do the district lines allow the two parties to compete fairly against each other. Competitiveness is concerned with a different kind of fairness: do the district lines give rise to genuine competition between the parties for votes? Notice how competitiveness is a voter-centric value while the symmetry measures are party-centric (party defined broadly as activists, party officials, candidates, elected officials, etc.). While parties may care a great deal about whether a redistricting plan makes it easier for them to win relative to their opponents, they may care less about whether it diminishes overall competition between the parties. Indeed, they may prefer an uncompetitive plan because it saves them time and campaign expenses. Bipartisan, incumbent-protective gerrymanders of the early 2000s, for instance, were politically symmetrical but highly uncompetitive.
Voters may have other fairness aspirations for their districts: responsiveness (to be heard), representativeness (to have one’s representatives speak, act, and stand for one’s interests), and accountability (to throw the bums out if they fail to perform the aforementioned duties satisfactorily). Bias measures, for instance, are concerned with whether the function that translates votes to seats is the same for both parties. Even assuming that perfect proportionality is not required (and indeed rarely possible when redistricting), gross mismatches between voters’ preferences and the kind of representatives who are elected undermine political legitimacy. As with competitiveness, the values of responsiveness, representativeness and accountability may not be ones that political parties care much for—as long as a redistricting plan produces an equal number of seats that are unresponsive/unrepresentative/unaccountable between the parties. But for citizens seeking a better government, they are paramount.
Certainly, courts—the Supreme Court through Whitford and Benisek and state supreme courts in developing partisan gerrymandering doctrines—may decide to elevate one particular brand of partisan fairness above others. Indeed, as Pam Karlan has written, the Supreme Court views the right to vote less as a snowflake (unique and fragile), and more as a calculator (for aggregating up to political outcomes). And in many other areas of the law, the Court has elevated certain strains of an overarching concept over others: intentional and animus-based discrimination over subconscious bias, for instance, financial injury over stigmatic injury in the context of standing, or in the antitrust context, injuries to consumers over those to competitors.
In some of these instances, there are philosophical or theoretical rationales for picking one strain of a concept over others. But if administrability is the central concern, it should not dog partisan gerrymandering doctrine. With the help of ever more potent computing power, extreme outcomes along each dimension of partisanship—voter or party-centric—can be simultaneously taken into account. In our article in the William & Mary Law Review, entitled A Reasonable Bias Approach to Gerrymandering: Using Automated Plan Generation to Evaluating Redistricting Proposals (now available online), we show how automated redistricting algorithms and a reasonable bias approach make it possible to generate an “everything bagel” set of feasible maps that encompass all the important values under the umbrella ideal of partisan fairness. Mapdrawers, be they legislatures, independent redistricting commissions or judges/special masters in remedial proceedings, can tailor the automated redistricting algorithms to conform to negotiated terms, state laws or other guidelines. They can weigh each measure as desired (more poppy seeds, less onion), and set minimum threshold values for each measure (at least a convincing smattering of sesame seeds). Insofar as each component is deemed a worthwhile addition (and a consensus can develop over time about what constitutes extreme outcomes), automated redistricting algorithms can produce maps that encapsulate the multi-faceted notion of partisan fairness. Party-centric values need not tread on the toes of voter-centric values: indeed, they can exist in reasonable tradeoffs with each other. The same can be done with compactness measures and other multidimensional redistricting values. Seldom in life (or in doctrine) can one have a combination of all. Bagels and voting districts are exceptions to this rule.