Packing and cracking

A gerrymander works through geometry. To weaken a group of voters, a mapmaker has two moves — and they are usually used together, not separately.

Packing concentrates the opposing bloc into as few districts as possible. Packed voters win those seats by enormous margins — seventy, eighty percent — while wasting nearly all votes cast above the threshold needed to win. Those excess votes return nothing. The packed party holds a seat it would have held anyway, but by a landslide that changes no other outcome.

Cracking does the opposite. The same bloc is divided into fragments, each too small to constitute a majority in any district where it lands. Cracked voters become perpetual minorities: enough of them exist to matter in aggregate, but in each individual contest they are outnumbered.

A full gerrymander combines both moves. The most concentrated pockets of the target group are packed into a small number of seats, making the waste there spectacular. The remaining voters — the fragments left over — are cracked across surrounding districts, where they lose narrowly. The effect is that a party winning, say, forty-five percent of the statewide vote might hold only a quarter of the seats. The map is working as designed.

The diagnostic most often applied is the efficiency gap, formalised by Nicholas Stephanopoulos and Eric McGhee in the early 2010s. It measures wasted votes — those cast for a losing candidate, plus those cast for the winner beyond what was strictly needed to win — and computes the difference in wasted votes between the two parties as a share of total votes cast. In a neutral map the gap sits near zero; a map engineered through packing and cracking pushes one party's waste systematically higher than the other's.

The efficiency gap has critics. It assumes a baseline that itself encodes political judgements, and a gap generated by a heavily packed district looks arithmetically identical to one generated by voluntary residential clustering. That second problem is not trivial. Voters with similar political preferences often live near each other; any district map will reflect that geography to some degree. The question — easy to pose, genuinely hard to answer — is how much of the observed skew comes from the mapmaker's pen and how much from where people chose to live.

Courts have tried to answer it. In Rucho v. Common Cause (2019), the United States Supreme Court held that federal courts could not adjudicate partisan gerrymandering claims, leaving the matter to state courts and state constitutions. State courts in several jurisdictions have subsequently applied their own standards, some of them explicitly invoking efficiency-gap evidence or related measures, to evaluate whether observed packing-and-cracking patterns crossed a threshold the legislature was not permitted to cross.

District shape is not proof of packing or cracking, but it is often a clue. Compactness measures — perimeter-to-area ratios, convexity scores, the radius-deviation approach — capture the odd elongations and tentacles that sometimes appear when a mapmaker is threading a district through disparate communities to achieve a numerical result. A district connecting two distant urban precincts by a narrow corridor, bypassing intermediate suburbs, warrants asking what the connection achieves politically.

What the shape cannot tell you on its own is the intent. A non-compact district may reflect county-line constraints, language communities, municipal boundaries or simply awkward geography. The numbers — margin distributions, efficiency gaps, partisan bias statistics — do more explanatory work than the outline. Used together, shape evidence and vote-share evidence are more powerful than either alone.

The arithmetic of packing and cracking is neutral about which party deploys it; the tool is available to whoever controls the redistricting pen. Understanding it is a matter of reading what the district lines, taken as a set, actually do to votes before they are cast.