Packing and Cracking, the Two Moves Behind Nearly Every Gerrymander

In February 2016, a North Carolina legislative committee met to approve the rules for a congressional map it had to redraw in a hurry, because a federal court had just struck down two districts as racial gerrymanders. One of the criteria the committee adopted was called "Partisan Advantage." State Representative David Lewis, a Republican and the committee's co-chair, explained what he meant with unusual candour. "I propose that we draw the maps to give a partisan advantage to 10 Republicans and 3 Democrats because I do not believe it's possible to draw a map with 11 Republicans and 2 Democrats," he said. The sentence was later quoted by the Supreme Court in Rucho v. Common Cause, the 2019 case that ended federal court review of partisan maps. What made it remarkable was not the goal, which mapmakers in both parties have pursued for two centuries. It was that he said the number out loud, before the lines were drawn, in a state where the two parties split the vote nearly evenly. To get from a state divided almost in half to a delegation of ten to three, the mapmakers needed only two techniques. They have names, and once you know them you will see them everywhere.

In short: Almost every partisan gerrymander combines packing (concentrating the other party's voters into a few districts they win overwhelmingly) and cracking (spreading the rest thinly across many districts they narrowly lose). The result is a party winning more seats than its share of votes would suggest. Measures like the efficiency gap count the "wasted" votes these moves create, and knowing how they work is the first step to spotting and challenging them.

The arithmetic of a seat

Start with a fact so obvious it is easy to miss. In a single-member district, a candidate needs just over half of the votes to win. Every vote beyond that is surplus; it does not add anything to the winner's seat count. Every vote for the loser, from the party's point of view, also adds nothing. Both kinds of votes are what political scientists call "wasted," not because the voters did anything wrong but because their ballots did not change who represents the district.

A mapmaker's job, if the goal is maximum partisan advantage, is to arrange voters so that the other party wastes as many votes as possible and one's own party wastes as few as possible. There are only two ways to waste someone's vote: make them win by too much, or make them lose. Those are packing and cracking.

Packing: let them win, by a mile

Packing concentrates the opposing party's voters into a small number of districts where they win by enormous margins, say 75 or 80 percent. The opponent wins those seats, but every vote over 50 percent is spent where it cannot help anyone else.

In practice, packing often follows existing geography. Democratic voters in many states cluster in cities, so a mapmaker can draw a district around the urban core and fill it with as many Democrats as possible. Republican voters cluster too, in rural and exurban areas, and Democratic mapmakers pack them the same way when they hold the pen.

Packing is the subtler of the two moves because it can look generous. The packed party wins its seats easily and its incumbents are safe. Some of those incumbents even like it, which is one reason packing sometimes survives with quiet support from the very party it harms.

Cracking: let them lose, narrowly

Cracking does the opposite. It takes the opposing party's remaining voters and spreads them across many districts so they are a minority in each, perhaps 40 or 45 percent. They lose every one of those seats, and all of their votes count as wasted.

Cracking is where most of the seats are gained. A city that could anchor one competitive district is sliced into three or four pieces, each attached to a large rural or suburban area that outvotes it. A college town is split down its main street. A Black or Latino neighbourhood is divided among districts where it never forms a majority. The voters in those places still vote; they just never decide anything.

A worked example

Here is a simplified, made-up state to show how it adds up. Imagine 500 voters to be divided into five districts of 100 each. Party A has 250 voters and Party B has 250. A truly even split.

A neutral map. If each district happened to mirror the state, every race would be 50-50 and a coin toss. In real life, geography means some districts lean one way and some the other. A map that followed natural communities might produce, say, two safe seats for each party and one toss-up. Seats would roughly track votes.

A map drawn for Party A. The mapmaker packs Party B into one district: 90 B voters and 10 A voters. That leaves 160 B voters and 240 A voters for the other four districts. Spread evenly, each of those four districts gets 60 A voters and 40 B voters. Party A wins four seats with 60 percent each. Party B wins one seat with 90 percent.

Result: with exactly half the votes, Party A wins four of five seats. Party B wasted 40 surplus votes in its packed district and 160 losing votes in the cracked ones. Party A wasted only 40 surplus votes across four districts and 10 losing votes in the packed one.

Real maps are messier, and margins of 60-40 carry risk, since a bad year for Party A could flip several seats at once. A cautious mapmaker trades some of the maximum gain for safety, building districts at 55 or 58 percent instead of the bare edge. That trade-off helps explain, as a matter of arithmetic, why a mapmaker might settle for ten seats rather than reach for eleven: an eleventh seat would have meant spreading Republican voters thinner and risking several districts in a good Democratic year.

The supporting moves

Packing and cracking do the heavy lifting, but mapmakers have a few lesser tools that show up alongside them.

Pairing. Two incumbents of the same party are drawn into one district, forcing them to run against each other or forcing one to retire. The opposing party loses experience and seniority, and sometimes a seat.

Kidnapping. An incumbent's home is moved into a new district full of unfamiliar voters, often far from the area the member has represented. The incumbent can still run, but without the name recognition and relationships that made them hard to beat.

Splitting a base. A line is drawn to separate a rising candidate's home town or core support from the rest of the district, so they cannot build the coalition they would need.

Prison gerrymandering. In many states, the census counts incarcerated people where they are imprisoned rather than at their home address. Districts containing large prisons, often rural, get population credit for people who cannot vote there and frequently come from cities elsewhere in the state. A number of states have changed their laws to count incarcerated people at their home addresses for redistricting.

These moves matter less to the overall seat count than packing and cracking, but they show the same principle: a map can decide a race before a single vote is cast.

Measuring the damage

If packing and cracking produce wasted votes, you can count them. That is the idea behind the efficiency gap, a measure developed by the law professor Nicholas Stephanopoulos and the political scientist Eric McGhee. It adds up each party's wasted votes across a state, takes the difference, and divides by the total votes cast. A large gap means one party's votes were converted into seats much less efficiently than the other's, which is what packing and cracking are designed to do. The measure was central to the Wisconsin case Gill v. Whitford, which reached the Supreme Court in 2018 and was sent back on standing grounds without a ruling on the merits.

Other measures look at the same thing from different angles. Partisan symmetry asks whether each party would win the same share of seats with the same share of votes: if Party A would win 60 percent of seats with 52 percent of the vote, would Party B, too? Mean-median difference compares a party's average vote share across districts with its share in the middle district; a large gap suggests packing. And ensemble analysis uses computers to draw thousands of random, rule-following maps and asks where the real map falls among them. If the enacted map is more lopsided than nearly every simulated one, it is very hard to explain without intent.

None of these measures is perfect. The efficiency gap can swing in a wave election; symmetry relies on hypothetical vote shifts; ensembles depend on which rules the simulation follows. That imperfection is part of why the Supreme Court majority in Rucho concluded there was no single legal standard for federal courts to apply. State courts, applying state constitutions, have been willing to use these tools anyway.

When packing is required, and when it is a cover

There is an important complication. Under Section 2 of the Voting Rights Act, as the Supreme Court interpreted it in Thornburg v. Gingles in 1986, a state may be required to draw a district where a large, compact minority community that votes cohesively can elect a candidate of its choice, if white bloc voting would otherwise defeat that community's candidates. Such "opportunity districts" have elected many Black and Latino members of Congress who would otherwise not be there. In Allen v. Milligan in 2023, the Court upheld that framework and required Alabama to draw a second such district.

That obligation is not packing. An opportunity district is drawn to give a community a fair chance, not to waste its votes. But the line can blur. A mapmaker can pack minority voters into a district at far higher percentages than needed, claim the Voting Rights Act required it, and use the move to drain those voters from neighbouring districts where they might have swung an election. Courts have struck down districts like that as racial gerrymanders, including the North Carolina districts whose invalidation led to the 2016 redraw described at the top of this post. The honest test is whether the percentage is what the community needs to elect its candidate, or far more.

Both parties, same playbook

It would be comfortable for progressives to treat packing and cracking as a Republican habit. The record does not support that. Maryland Democrats cracked Republican voters in the western part of the state in 2011; that map was the Democratic half of the Rucho case. Illinois Democrats have drawn maps that pack Republicans into downstate districts. In 2022 New York's Democratic legislature passed a map that the state's highest court struck down as a partisan gerrymander.

It is also true that the scale has not been symmetrical in every decade. After the 2010 census, Republicans controlled the line-drawing in more states with many seats, and maps in places like Pennsylvania, Wisconsin, Michigan and North Carolina produced some of the largest efficiency gaps on record. In Pennsylvania's 2012 House races, Democratic candidates won about half of the statewide vote and took 5 of 18 seats. The point is not that one side is pure. It is that whoever holds the pen uses these two moves, and the only durable fix is to change who holds it and what rules they must follow.

None of this calls for cynicism about the people who represent those districts. Many members elected from gerrymandered seats work hard and serve their constituents well. The problem is the system that chose their voters for them, not the individuals sitting in the seats.

What changes when you can see it

Once you understand packing and cracking, a district map stops looking like abstract art. You can look at your own city and ask: is it whole, or sliced into pieces attached to far-off places? Is there one district where one party wins with 80 percent while the neighbouring districts all go the other way by small margins? Those patterns do not prove intent on their own, since geography produces some of them naturally. But they tell you where to look, and what questions to ask at the next hearing.

The people in a cracked district are not abstractions either. They are the union local in a mid-sized city that never has its own member of Congress, the rural Black community split three ways, the suburban independents whose vote is diluted on purpose. Neighbours of every party lose something when representation is engineered. Fixing it is not a partisan project, and the reform campaigns that have succeeded, from Michigan to Utah, did so by saying exactly that.

What you can do

• Look at your own district. Find your representatives and compare your district's shape to your city or county. Is your community whole?
• Follow the map fights. Our voting rights and redistricting topic page gathers news on new maps, lawsuits and commissions in every state.
• Watch the court cases. The court watch page lists Supreme Court opinions this term with the Court's summaries and the emergency docket, where election cases often land.
• Join people who draw maps for a living, or for free. Voting rights groups run map-literacy workshops before redistricting hearings. Find them in the organizations directory.