A plan only lasts if people who dislike parts of it can still accept how it was made. This page starts with every idea of a fair map we could find, from polls, both parties, civil-rights groups, courts, researchers and other democracies, and checks the plan against each one. Then it looks at Neutral Lines from each group with a stake in congressional districts. For each one it sets out their legitimate concerns, what the formula offers them, and what it costs them, using numbers computed from the same data as the map. Below that: a test of whether the rule can be tuned to favor a party, the federal legal status of every state's plan, and an assessment of the 2025–26 redraws and the reform proposals.
We collected the ideas of fairness that Americans, both parties, civil-rights and reform groups, courts, researchers and other democracies actually hold, and checked the plan against each one. Most can be met at once, and the plan meets them by construction. A few cannot all be met by any plan: some contradict each other outright (a party-blind map cannot also be guaranteed proportional; a race-blind map cannot also guarantee minority opportunity districts), and mathematicians have proved that some combinations are impossible. For those, the table says which way the plan goes and why, and publishes the measurement so nobody has to take our word for it.
Where an idea could be met without adding a setting someone could tune, we changed the rule to meet it: who counts in the population, a public check of the input data, a fixed national definition of every boundary the rule follows, a ban on mid-decade redraws, and a formula for Voting Rights Act remedies. Ideas that need judgment or election data are measured after the fact, never used to draw.
Any rule with a choice in it can be steered. We tested our own earlier version, which chose among compact plans by a partisan-symmetry score, and found it could. Hand-picking among plans that all passed its checks moved the national result from 215 to 227 Democratic-leaning seats. Swapping in other fairness measures that sound just as reasonable (efficiency gap, mean–median, proportionality, competitiveness, single election years), or nudging its tolerances, moved it from 220 to 226. Even which starting angles were searched moved it from 220 to 223.5. Each version was "fair" by some measure, and each gave a different map.
So the rule now has no choice in it. It uses one objective that reads no election data: fewest split counties, then smallest moment of inertia. It searches until doubling the search finds nothing better. The only way to change a state's map is to find a valid plan with a better score, and the score has no party in it. The table compares what the rule produces with what a free choice among the very same valid plans could have produced.
Federal law requires equal population, single-member districts and compliance with the Voting Rights Act as narrowed by Louisiana v. Callais. Today each state adds its own criteria, which is why fifty maps are drawn fifty ways. Neutral Lines is proposed as one national standard under Art. I §4, replacing those state criteria for congressional districts only. The table applies the same automated checks to both plans and lists each state rule the national standard would replace.
Status key: ✓ Pass · ≈ Justified (allowed with a stated, neutral reason) · ! Review (worse than the enacted map or near a common flag) · ✕ Fail. Click a state for its full page.
Each approach is judged on three questions. Legal: would it survive in federal and state court? Consensus: could voters of both parties accept the process, even when they dislike the result? Practical: can election offices run it, and will it last?