Loading issue research
Every proposed congressional map should be tested against a published, million-map algorithmic baseline before whichever body, commission or legislature, adopts the final lines.
Verification Status
AI-researched, unverifiedLast Reviewed
Jul 5, 2026
Cited Sources
17
What is failing, what we would change, and the conclusion we are willing to defend.
Algorithmic tools belong in redistricting as a mandatory disclosure and detection layer feeding the body that draws the final map. Two Supreme Court rulings define the room this proposal has to work in. Rucho v. Common Cause (2019) closed federal courts to partisan-gerrymandering claims. Louisiana v. Callais (2026) narrowed the other tool available, now requiring a race-neutral illustrative map that also meets a state's own districting goals in addition to evidence that a map has a discriminatory effect, a bar scholars expect to be far harder to clear. With one door closed and the other narrower, what's left is what gets published before a map is adopted, and what a state court, a citizen commission, or the surviving federal claims can do with that evidence.
That's what computational redistricting already does, today, in production. RepresentUs and the Princeton Gerrymandering Project already run every state's adopted congressional map through an ensemble baseline of roughly a million alternative plans meeting the same population, contiguity, and compactness rules, then check whether the adopted map is a statistical outlier against it, publishing the results through their Redistricting Report Card. It has failed maps from both parties: Texas's Republican-drawn 2021 map and Illinois's, Maryland's, New York's, and Oregon's Democratic-drawn ones all received F grades in the same cycle. The tool doesn't pick a side. It picks outliers. MGGG's own GerryChain library, the open-source engine behind much of this ensemble work, has been used to evaluate maps in active litigation and in retrospective reviews of real commission cycles, including a 2024 joint assessment of Michigan's first full redistricting cycle with the advocacy group that helped create its commission.
Two things this position refuses to do. It won't hand the final decision to software: every serious implementation, including the Institute for Computational Redistricting's work with Arizona's independent commission and, in Missouri, with the League of Women Voters, produces evidence for a human body to weigh, never a map with legal force. And it won't treat "race-blind" as a synonym for neutral: an ensemble built without racial data has been shown to significantly undercount minority-opportunity districts.
Require, as a condition of a congressional map taking effect, that it be tested against a published, race-neutral ensemble of at least one million alternative plans meeting the state's traditional redistricting criteria, with results public before final adoption.
Require a second, published audit using racial and language-minority population data, to catch the undercount a race-blind baseline produces (see Extended for why this is a second, separate analysis with its own legal purpose).
Fund the technical assistance, through federal grants, that lets any state run these analyses on existing open-source tools, opening access beyond litigants who can afford expert witnesses.
Make this kind of ensemble analysis admissible as evidence in redistricting litigation, the same practice courts in Pennsylvania and North Carolina already accept.
When a commission deadlocks, as Virginia's did in 2021 before its Supreme Court stepped in with special masters, whoever draws the map next remains bound by the same requirement as every other map-drawer: test the result against the neutral ensemble before adoption. A breakdown in the ordinary process creates no disclosure exemption.
Editorial note: this issue is marked stance_type: normative. Requiring
the audit baseline to use racial data takes an explicit side in a
live debate over "race-blind" redistricting; a procedural framing would understate that
this argues against a specific legal theory (see Extended for the full reasoning).
Turn frustration into useful pressure.
If this position misses evidence or a lived consequence, challenge it. If it holds up, help test it locally and connect it to the issues around it.