Gerrymander Mapgerrymandermap.us

How scores work

This site measures how compact each congressional district's shape is. A compact district is roughly round or square. An irregular one snakes, sprawls or has pieces cut out of it. Each state page also scores the state's map for partisan fairness, explained below.

The three measures

Each runs from 0 to 1, where 1 is as compact as a shape can be.

Polsby-Popper
Compares a district's area with its perimeter. A circle scores 1. Jagged edges and long thin arms add perimeter without adding area, so they pull the score down.
Reock
Compares a district's area with the smallest circle that fits around it. A stretched-out district fills little of its circle and scores low.
Convex hull
Compares a district's area with a rubber band stretched around it (its "convex hull"). A district with bites taken out of it, or tentacles reaching out, scores low.

Measuring against land

Reock and convex hull count only the part of the circle or rubber band that falls on the district's own state. Water between islands or past a coastline is space no district could use, so it doesn't count against a district. Land a district leaves out still does. Without this, island districts such as Hawaii's 2nd or the Florida Keys would rank among the least compact for their geography alone.

Polsby-Popper has no such adjustment, so a ragged coastline still lowers it. Each district is measured in an equal-area map projection centered on the district, so a district in Alaska is measured the same way as one in Maine.

Each district page also says how much of the district's border is the state's own edge: a coastline, a national border or a state line. Mapmakers don't choose those lines, and a ragged one, like a coastline or a river, lowers Polsby-Popper. A stretch of border counts as the state's edge if it lies within 100 meters of the state's outline, which allows for the small shifts from simplifying the newer state maps (see "Which map" below).

From three measures to one score

For each measure, a district gets a percentile: where it falls among the 429districts that have lines to draw. The six states with a single district covering the whole state are measured but not ranked. The irregularity score is 100 minus the average of the three percentiles, so 0 is the most compact district and scores near 100 are the least compact. Percentiles put the three measures on one scale, since their raw values sit in very different ranges.

"Less compact than X% of districts" on each district's page comes straight from its rank by irregularity score.

What a score can't tell you

Who lives in each district

District pages note when most of a district's voting-age residents (18 and over) belong to minority groups: either one group is over half, or minority groups together are. The Voting Rights Act can require districts like these, so a minority community can elect candidates it prefers, and they can be irregular on purpose. The note doesn't say a district was drawn for that reason, because the site can't know that.

The counts come from the 2020 Census redistricting data, the same data the maps were drawn with. Each census block is placed in the district that contains the point the Census marks inside it, and the blocks are added up. Hispanic or Latino residents of any race are one group. The other groups (Black, Asian, American Indian or Alaska Native, and Native Hawaiian or Pacific Islander) count people who aren't Hispanic and who gave that race alone or in combination with another, as Voting Rights Act analyses do, so someone who is Black and white counts as Black. Minority residents are everyone except non-Hispanic white residents.

As a check, each district's 2020 population should equal its state's average, since every map in use was drawn to equal populations. Most are within 0.1%, typically a few dozen people. Districts in nine states are off by more, up to 2.3%, because those states adjust the Census counts before drawing maps: eight of them count people in prison at their home addresses rather than at the prison, and Hawaii leaves out military members and students who live elsewhere.

Cities and towns in each district

District pages list the largest places with people in the district, to show where it is and what it covers. Places are the Census Bureau's: incorporated cities, towns, villages and boroughs, plus census-designated places, which are communities the Census names for statistics that have no local government of their own. Some states have many of these: every community in Hawaii is one, and so are large suburbs such as Metairie, Louisiana. The site calls them communities.

The counts come from the same 2020 Census blocks as above. Each block records the place it lies in, so adding up a place's blocks by district gives how many of its residents live in each district. Districts follow block lines, so the split is exact for 2020 residents. As a check, a place's residents across all districts must add up to all of its 2020 residents. Places are listed by their residents in the district, up to 10, and a place split between districts says how much of it is in this one. Residents outside every place live in rural areas or small settlements the Census doesn't name.

State maps mark the largest places in the state with a dot, as landmarks. Each dot is at the census block nearest where the place's residents live on average, so it always falls where people live, though not always at the downtown. (The Census's own point for a place can be in open water: most of San Francisco's area is ocean and bay.) Each map marks up to 10, biggest first. A place is skipped when it sits right next to one already marked, such as a suburb beside its city, so the marks spread across the state, and when its label would cover another one. So a map can skip a big suburb for a smaller city elsewhere, and a smaller map, such as on a phone, can mark fewer. A city that shares one government with its county is labeled by the city's name alone, such as Nashville for Nashville-Davidson.

Each district page's outline marks the district's own largest places the same way, ranked by how many of their residents live in the district. A place split between districts gets a dot in each part, at the census block nearest where that part's residents live on average, so the dot always falls inside the district.

Close-ups of crowded areas

Some districts cover only a few city neighborhoods, so on a state map they're too small to see or click. Below those maps, close-ups redraw the crowded areas larger, each outlined with dashes on the map and named after the largest city in it. A district counts as too small when, on a map drawn 512 pixels tall, it has less than 8 pixels from its middle to its edge. Nearby small districts share a close-up, which enlarges its area 2 to 8 times and is redrawn from the detailed district lines, not just magnified. Where districts are still too small even then, as in central Los Angeles, Manhattan and Brooklyn, and central Houston, a close-up gets a close-up of its own. Close-ups mark the largest places in them the same way as state maps, so they can mark cities too close together to mark on the map. Close-ups are for display only. Scores always use each district's full shape.

Which map

Scores are for the districts used in the November 3, 2026 election. Most come from the U.S. Census Bureau's map of the 2024 districts. 9 states used new lines in 2026 (Alabama, California, Florida, Louisiana, North Carolina, Ohio, Tennessee, Texas, Utah), and for those the site uses each state's official 2026 map. Missouri and Virginia passed new maps that were not used in 2026, so they keep their earlier lines.

The state files are much more detailed than the Census map, and more detail means a longer measured perimeter. So the new maps are simplified to the Census map's level of detail first. The settings were calibrated against the Census Bureau's own detailed and simplified versions of the same districts for 12 states, using 12 extra Census files.

Partisan fairness

Shape is one way to look at a map. Another is how it turns votes into seats: whether one party's voters are packed into a few districts it wins by huge margins, or spread thin across districts it narrowly loses. Each state page scores this with three common measures. All three use the same sign: positive favors Republicans, negative favors Democrats, and 0 favors neither. State pages show the size of the number and the party it favors in words.

Efficiency gap
A vote is "wasted" if it went to a losing candidate, or to a winner beyond the half of the district's votes needed to win. The efficiency gap is Democratic wasted votes minus Republican wasted votes, as a share of all votes. A gap of 10% means one party wastes more votes than the other by 10% of all the votes cast.
Mean-median difference
The Democratic share of the vote averaged across the state's districts, minus the Democratic share in its middle (median) district, in percentage points. If the middle district is less Democratic than the average, Democratic voters are concentrated in a few districts, which favors Republicans. The reverse favors Democrats.
Declination
Lines up the districts from most Republican to most Democratic and compares how lopsided each party's wins are, as an angle scaled from -1 to 1 (Warrington, 2018). It needs districts won by both parties, so it can't be measured where one party carried every district.

Where the votes come from

The measures use the 2024 presidential vote (Kamala Harris and Donald Trump only, so third parties don't count) as a consistent measure of how each district leans. House races are less even: some candidates run unopposed, and incumbents draw extra votes. For states that used the same districts in 2024 and 2026, precinct results are added up within each district. Texas and North Carolina drew new districts for 2026 and published official 2024 results for them, so those are used directly. The other redrawn states published none, so their scores will use the 2026 House results once they are final.

District pages show each district's 2024 presidential vote the same way, as Harris's and Trump's shares of the votes for the two of them, and a district covering a whole state gets the state's vote. Where a precinct is split between districts, its votes are divided the way its House votes were, so those districts' figures are close estimates. Districts in the redrawn states without official results show none.

What fairness scores can't tell you

Senators

District and state pages name each state's two U.S. senators, with their parties, from the Senate's own list of current senators. Senators are elected by the whole state, so every district in a state shares the same two. The list is dated on each page. The winners of the November 2026 Senate races take office January 3, 2027, and the list will be updated after January 3, 2027.

November 2026 House and Senate races

District pages name the winner of each district's November 3, 2026 House race, and say who won each Senate seat that was on the ballot. Winners take office January 3, 2027. Names, parties and counts come from each state's election office, as published before the states certified them, so they're unofficial.

A race shows a winner only once news organizations' decision desks have called it. The site never calls a race itself: until a race is called, its page says so and names nobody. The list of uncalled races was checked 2026-11-06 09:32 ET. Only that list comes from news organizations.

Some states hadn't published statewide results when they were gathered. Some post results only by county until their official count. Those races say results aren't available yet, and are filled in as the states publish.

Data sources

District maps

Population

Votes

Senators

2026 results

Every step, from download to score, is open source and can be rerun: github.com/rpivo/gerrymandering-map.

The code is under the MIT License. The scores, ranks and shapes this site produces are under Creative Commons Attribution 4.0: you can reuse them for any purpose if you credit Gerrymander Map.