What Is Gerrymandering? Drawing Districts to Decide Elections

The mechanics are simple enough to explain with a small example. Imagine a region with sixty voters, forty supporting party A and twenty supporting party B, divided into five districts of twelve. Drawn one way, party B wins nothing. Drawn another, with B's voters split into groups of seven across three districts while A's are concentrated into two, B wins three seats out of five with a third of the vote. The two operations have names:

• •Packing, which crowds opposition voters into a small number of districts they win by enormous margins, so that most of their votes elect nobody
• •Cracking, which divides an opposition community across several districts so that it forms a minority in each and wins none
• •Both work by manipulating wasted votes, meaning votes cast for losing candidates plus surplus votes beyond what a winner needed
• •The efficiency gap, a measure proposed in 2015, quantifies this as the difference in wasted votes between parties divided by total votes, and has been used in litigation
• •Related tactics include hijacking, which redraws two incumbents of the opposing party into one district so they must run against each other, and kidnapping, which moves an incumbent's home out of their district

The practice depends on a specific combination of institutional features, which is why it is severe in some countries and unknown in others. First, single-member districts with winner-take-all results, since a proportional system distributes seats by vote share and leaves boundary-drawing with little to do. Second, boundaries drawn by politicians with a partisan interest, which in most American states means the state legislature draws both its own districts and the congressional ones. Third, requirement for regular redistricting, in the United States after each decennial census. And fourth, extremely detailed data about how individuals vote, which is the modern accelerant: software combining census blocks, voter registration, consumer data and past results allows a mapmaker to test millions of possible maps and select one that reliably produces a target outcome, a precision that was not available before the 2000s and that has made the practice far more effective.

American courts have drawn a sharp distinction between two kinds. Racial gerrymandering, drawing lines to dilute the voting power of a racial group, is unlawful under the Voting Rights Act of 1965 and has been struck down repeatedly, though the reach of that Act was substantially narrowed by the Supreme Court in 2013 when it invalidated the formula requiring certain states to obtain federal approval before changing electoral rules. Partisan gerrymandering, drawing lines to advantage a political party, was held in 2019 in Rucho against Common Cause to present a political question that federal courts cannot adjudicate, because no judicially manageable standard exists for deciding how much partisan advantage is too much, a five to four decision whose dissent argued that such standards had been developed and were workable. The practical consequence is that partisan maps must now be challenged in state courts under state constitutions, which several have done successfully, most notably in Pennsylvania and North Carolina, with results that have varied as those courts' compositions changed.

Several approaches exist and each has trade-offs. Independent commissions remove the drawing from the legislature and hand it to a body of citizens, retired judges or officials with rules about partisan balance, as in California, Arizona, and at national level in Canada, Australia and the United Kingdom, where the boundary commissions are non-partisan and their recommendations are effectively automatic. Criteria-based rules require districts to be contiguous, compact, to respect existing local boundaries and to keep communities of interest together, which constrains the worst distortions and does not eliminate them, since compactness can be achieved by maps that are still highly skewed. Algorithmic and statistical methods generate enormous ensembles of neutrally drawn maps and compare the enacted plan against that distribution, which gives a rigorous measure of how unusual a map is and has been accepted as evidence in several cases. The structural remedy is to change the electoral system, since multi-member districts with proportional or ranked allocation make the drawing of boundaries far less consequential, which is why the practice is essentially unknown in most of continental Europe.

The second-order effects are arguably worse than the direct ones. A district drawn to be safe means the general election is not competitive, so the contest that decides the seat is the party primary, in which turnout is low and the electorate is more ideologically committed, which rewards candidates who appeal to the party base rather than to the median voter and contributes to polarisation. It reduces accountability, since a representative in a safe seat faces little risk from the electorate as a whole. It depresses turnout, because voters who know the result in advance have less reason to participate. And it corrodes the perceived legitimacy of results, which is a slow cost that is hard to reverse. The counterargument, made seriously, is that some degree of political geography produces uneven outcomes anyway, since voters of one party often cluster in cities and would be over-concentrated by any neutral map, which means a perfectly fair map is itself a contested idea rather than a technical target.

Gerrymandering works by packing opposition voters into a few districts they win overwhelmingly and cracking the rest across districts where they narrowly lose, which maximises their wasted votes. It requires single-member winner-take-all districts, partisan control of boundary drawing and detailed voter data, and modern software has made it far more precise. American federal courts will hear racial but not partisan claims after a 2019 ruling, and the remedies are independent commissions, statistical comparison against neutrally drawn maps, and proportional systems that make boundaries matter less.