Experiments4Good · Counter-Map v2 · Published August 1, 2026 · Updated November 3, 2026

The Whole Map

On August 1, the 2026 Election Risk Heat Map scored 733 counties across 23 states, and the other 27 states were not scored at all. Louisiana, Tennessee, and Alabama had zero rows. Since then, Voting Rights Lab has added seven states. As of its October 21 data, it scores 883 counties across 30 states, including all three. Twenty states are still not scored at all. Alabama now scores 76, Very High. Louisiana scores 38 and Tennessee 44, both below the Very High band that starts at 61. On our composite, both sit in the top two tiers. Flip the toggle.

Their index, verbatim

Every number on this view comes straight from the data the tool publishes with its map. States show the Statewide Election Risk Index (1 to 100). Tap a covered state for its full county list. Tap a grey state and see what the model cannot see. Switch between August 1 and October 2 to see what changed.

883
Counties scored
30
States covered
20
States not scored
2,260
Counties without a row
Not scored
Moderate 1–20
Elevated 21–40
High 41–60
Very High 61–80
Extreme 81–100
Tap any state. On this view, covered states show their published risk index and every scored county. Grey states show what the model has no data for.

Four receipts

All of their numbers below are from their own data, August 1, October 2 and October 21, 2026. All of the redistricting facts are current as of October 28, 2026.

Louisiana
August: not scored · 0 rows
October: 15 parishes · statewide 38
October 21: 15 parishes · statewide 38
Signed SB121 on May 29, eliminating its second majority-Black congressional district. House primaries reset to Nov 3, general Dec 12. In August the state had no row in the risk data. In October it has 15, and the field that tracks election deniers in local election offices reads "Research to Come" for every one. On October 27 the three-judge panel in Shreveport heard the state's motion to end the Callais case and the plaintiffs' argument that it should stay open and aimed at the 2nd District, the state's remaining majority-Black seat. The panel took it under advisement without ruling or setting a date. Filings are due December 15 if the case survives.
Tennessee
August: not scored · 0 rows
October: 15 counties · statewide 44
October 21: 15 counties · statewide 44
Enacted a new map May 7 cracking Memphis's majority-Black 9th District. The map survived a federal injunction July 23 and runs in November. In August, Shelby County had no row. In October it scores 38, and in VRL's October 21 data it still does, with 620 counties scoring higher. Justin J. Pearson won the Democratic primary in the new 9th on August 6. In VRL's October 2 data, every district touching Shelby County carries a Solid R rating. On October 5, the DCCC added his race to its Red to Blue program. The only public poll, conducted for a Democratic-aligned PAC, has Republican Brent Taylor ahead 48 to 44. Cook rates the district Solid Republican. Early voting runs October 14 to 29.
Alabama
August: not scored · 0 rows
October: 31 counties · statewide 76
October 21: 31 counties · statewide 76
Ran its court-blocked 2023 map in the August 11 special primaries under a Supreme Court stay issued June 2, and runs it again in November. In August, not one county was scored. In October, and again on October 21, its statewide index is 76, seventh highest in the country. On September 18 the Milligan plaintiffs asked the Supreme Court to dismiss the state's appeal as moot; the Court has not acted.
Virginia
Scored · 55 localities (August and October)
Fully covered in both versions, at a statewide risk index of 1 out of 100. In August, Virginia had 55 localities of coverage while Louisiana had none. Louisiana now has 15.

Sources: the 2026 Election Risk Heat Map, published by Voting Rights Lab; Experiments4Good's Post-Callais + Election Defense Movement Brief v5; and electionriskmap.org.

"The Map They Made": All state and county Election Risk Index values are reproduced verbatim from the data the Heat Map publishes with its map: the August 1 version (generated July 22, 2026; pulled August 1, 2026) and the October 2 version (generated October 2, 2026; pulled October 6, 2026 from the dataset embedded in the map page, since Voting Rights Lab no longer posts a separate file). The October 21, 2026 version (generated October 21, 2026; pulled October 23, 2026, from the same embedded data) is the default view, and the toggle shows all three. The tool's methodology gates coverage on electoral competitiveness; by its own explainer, so much of the map "appears grey" because disputes there are "unlikely to gain traction." The same explainer says coverage is updated regularly. The grey states on this view (27 in August, 20 in October) are absent from the data entirely. They are unscored, not low-scoring. In the October 2 data, Florida's rows carry two statewide values (57 on 17 rows, 53 on 8); this page shows 57, and either value falls below the Very High band. The tool's index bands: Moderate 1–20, Elevated 21–40, High 41–60, Very High 61–80, Extreme 81–100.

"The Whole Map": Experiments4Good's Post-Callais + Election Defense Movement Brief v5, refreshed July 30, 2026, and checked against electionriskmap.org on October 28, 2026. Composite of Callais redistricting status, election defense classifications synced with electionriskmap.org (verified against the Brennan Center and UW State Democracy trackers) except Ohio and Texas as explained below, and 2026 battleground overlap. Scores run 0 to 10, with the weights published here: redistricting status scores live 4, enacted harm 3, pending 3, enacted Democratic counter-redraw 1, none 0; election defense scores critical 4, high 3, complied 3, elevated 2, moderate 1, none 0; battleground overlap adds 2. Battleground is our call, made in the v5 brief on July 30: a state where control of a seat, or whether a community gets represented at all, is being fought over in 2026, whether or not ratings outlets call the race close. Outside measures have since caught up in places. Tennessee is the clearest: on October 5, the DCCC added TN-09 to its Red to Blue program. Complied scores level with high because surrendered voter data is exposure realized, not risk avoided. The gap between enacted harm and enacted counter-redraw keys to majority-minority district harm, not party. And battleground status here is an additive two-point input on a map that scores every state; it is not a gate that decides which states exist in the data. States that handed their full voter file to the Department of Justice are scored here as "complied," worth the same as high, even where ElectionRiskMap.org's October 16 eleven-vector method now rates them lower. The absence of a lawsuit is not the absence of election risk. Texas (December 2025) and Ohio (February 2026) handed DOJ their complete voter files, driver's license numbers and partial Social Security numbers included, without a fight. DOJ then shared state voter data with Homeland Security Investigations for citizenship matching in the SAVE database, and on October 9, 2026 Judge Sparkle Sooknanan vacated the whole program as unlawful, in an order that does not say what happens to the copies DOJ already holds. As of October 28 this applies to Ohio and Texas.

"The Synthesis": compares verdicts, not numbers. The two instruments measure different things on different scales across different coverage, so no blended score is computed. A state is flagged by their instrument at Very High or Extreme (index 61 or above) and by ours at Tier 1 or 2 (composite 6 or above); blind spots are states absent from their data that clear the same Tier 1 or 2 bar on ours.

County denominator: 3,143 counties and county equivalents across the 50 states, per the U.S. Census Bureau's current count of 3,144 including the District of Columbia (a count that includes Connecticut's nine planning regions, recognized as county equivalents in place of its eight historical counties). D.C. sits outside both instruments' scope and is excluded here. 883 counties have a row in the October data (733 in August). 2,260 do not (2,410 in August).

Built by Experiments4Good. This page critiques a framework, not an organization's intentions. Both maps agree the threat is real. They disagree about who counts.