Policy 11: Require states to use automated congressional districting.

Researched: 2026-07-22. Status: draft.

The plank

Gerrymandering is a primary cause of our polarized, gridlocked political system. Requiring automated districting, such as using the shortest split line method, will immediately, irrevocably end gerrymandering. Congress will become accountable to diverse constituents, and will moderate.

README.md and app/templates/index.html match verbatim (checked 2026-07-22) — no drift to fix.

Warning before we go further: two words in this plank do not survive the research below. "Primary" (as a cause of polarization) is contradicted by the best political-science evidence, and "irrevocably" is wrong as a matter of law — any statute can be repealed. The anti-gerrymandering case itself is strong; the plank's causal framing oversells it. See "The opposition's best case" and "Open questions."

The problem

Partisan map-drawing lets politicians choose their voters instead of the reverse, and both parties do it wherever they hold the pen.

• Seats shifted. The Brennan Center estimated that congressional maps used in the 2024 election contained a net of roughly 16 fewer Democratic or Democratic-leaning districts than maps complying with the anti-gerrymandering standards of the (stalled) Freedom to Vote Act — an advantage of about 16 House seats for Republicans versus fair maps (Brennan Center, "How Gerrymandering Tilts the 2024 Race for the House," 2024). For scale: Republicans won House control in 2024 by single digits. The direction of the net advantage reflects which party drew more maps this cycle, not which party gerrymanders when able — Illinois and Maryland are Democratic gerrymanders by the same metrics.
• Competition collapsed. By the final stretch of the 2024 cycle, only 27 of 435 districts were rated toss-ups by the Cook Political Report (2024). The share of districts with a Cook Partisan Voting Index between D+5 and R+5 fell from roughly one-third of all districts (1990s) to under one in ten by 2018 (FiveThirtyEight, "What We Lose When We Lose Competitive Congressional Districts," 2018). Twenty states had zero competitive congressional districts in 2024 (Brennan Center, 2024).
• No judicial backstop. In Rucho v. Common Cause (588 U.S. 684, 2019, 5–4) the Supreme Court held partisan-gerrymandering claims are nonjusticiable political questions — federal courts will never police this. Chief Justice Roberts's majority opinion pointed explicitly at the alternative: the Elections Clause gives Congress the power to act.
• It is getting worse, mid-decade. In 2025 the norm of once-per-decade redistricting broke: Texas redrew its congressional map mid-decade (August 2025) to add roughly five Republican seats, and California voters responded by passing Proposition 50 (November 2025) to suspend its commission map and add roughly five Democratic seats, with Missouri and North Carolina also redrawing (widely reported, 2025; seat estimates are projections — mark as estimates until 2026 results are in). An arms race with no stopping rule.
• How it shows up in an ordinary life: if you live in one of the ~90% of districts that are safe (2024 Cook data), your general-election vote for the House is close to decorative. The only election that matters is a low-turnout primary decided by the most ideological sliver of one party.

What the status quo costs in dollars is harder to pin down than in most planks — the cost here is representational, not budgetary. Direct dollar costs are litigation and consultants (see below), in the low hundreds of millions per decade nationally (estimate, weak evidence); the real cost is a House whose membership is decided in primaries.

Who profits from the status quo

• Incumbents of both parties. The single most consistent output of legislature-drawn maps is incumbent protection. House incumbent reelection rates have run 90–98% in every cycle since 1946 (OpenSecrets, reelection-rate series). Safe seats are the asset; redistricting is how the asset is maintained. This is the rare villain section where the villain is bipartisan and includes the people who would have to vote for the fix.
• Party mapmaking operations. The Republican State Leadership Committee's REDMAP project raised about $30 million in 2010, targeted ~107 state legislative races in 16 states, helped flip roughly 20 chambers (~680 seats), and thereby controlled the 2011 map-drawing in Pennsylvania, Ohio, Michigan, North Carolina, and Wisconsin (RSLC's own REDMAP summary; PBS NewsHour, 2017). Democrats built a mirror image: the National Democratic Redistricting Committee (founded 2017, chaired by Eric Holder) has raised and spent tens of millions per cycle on redistricting politics and litigation (order of magnitude; exact totals unverified).
• Redistricting consultants and litigators. A professional ecosystem — partisan demographers (the late Thomas Hofeller being the documented archetype, per files disclosed in 2019 litigation), map-drawing consultancies, and election-law firms — is paid on both sides of every map fight. Statewide redistricting litigation routinely costs a state's taxpayers millions of dollars per cycle in outside counsel (estimate, weak evidence — worth pinning down per state; North Carolina alone litigated maps essentially continuously from 2011 to 2023).
• Majority-party state legislators, who convert temporary statewide majorities into decade-long congressional delegations.

No trade association lobbies for gerrymandering by name; the defense of the status quo is done by the parties themselves, which is why it never gets fixed from inside.

The proposal, concretely

Vehicle. A federal statute under the Elections Clause (U.S. Const. Art. I, § 4, cl. 1: states prescribe the times, places, and manner of congressional elections, "but the Congress may at any time by Law make or alter such Regulations"). This is the same hook used by the Freedom to Vote Act (S. 2747, introduced September 2021; failed Senate cloture January 2022), which already contains a statutory ban on partisan gerrymandering of congressional maps, uniform criteria (VRA compliance, communities of interest, no incumbent-address consideration), quantitative partisan-fairness tests, and a private right of action in federal court. The realistic legislative move is to amend that framework's remedy: when a state's map fails the fairness test, the replacement map is generated algorithmically rather than redrawn by the offending legislature.

The named mechanism. The shortest-splitline algorithm (Warren D. Smith, Center for Range Voting, rangevoting.org/Splitlining.html, proposed mid-2000s): to divide a state into N districts, find the shortest straight line that splits the population in the ratio ⌈N/2⌉:⌊N/2⌋; recurse on each half until every piece is one district. Fully deterministic — given the census data, everyone's computer produces the same map. Ties broken by fixed rules. The Center published demonstration splitline maps for all 50 states from 2000 census data. Its virtue is exactly its indifference: no input about party, race, or incumbency exists anywhere in the procedure, so there is nothing to manipulate except the census itself.

Honest engineering note. Pure splitline is the maximal version of the idea and the easiest to defend against manipulation, but it ignores everything except population: it slices cities, counties, and communities of interest arbitrarily, and it cannot deliberately create the majority-minority districts that Voting Rights Act § 2 has historically required (see Evidence). A defensible statute would more likely mandate an open-source, seeded, constraint-based algorithm (equal population, contiguity, compactness, county/city integrity, VRA compliance as constraints — the academic ensemble/optimization literature, e.g. DeFord–Duchin–Solomon's recombination Markov chains, arXiv:1911.05725, 2019, shows this is tractable) with the code and inputs published before the census, so the output is reproducible by anyone. The plank should name splitline as the illustration, not the specification.

Where it's in use. No U.S. jurisdiction uses fully automated districting today. Mexico's INE has used optimization-based automated first-draft maps with human adjustment since 2015 (secondary sources; details unverified). The closest U.S. analogues are the independent commissions (AZ 2000, CA 2008/2010, MI 2018, CO 2018), which remove the legislature but keep human drawers — their track record is the best available evidence for "what happens when you take the pen away from incumbents."

Evidence

1. Removing the legislature from map-drawing reduces partisan skew — moderate evidence. Commission-drawn maps in the 2021 cycle scored materially fairer than legislature-drawn maps: analyses of California and Michigan commission maps found efficiency gaps averaging under 5%, versus over 10% in contemporaneous legislature-drawn maps elsewhere (Duke Journal of Constitutional Law & Public Policy, "The Good, the Bad, and the Ugly: Redistricting Commissions in the 2021 Cycle," 2022; PlanScore state scores, 2022). Arizona's and Colorado's 2021 commission maps produced near-proportional partisan outcomes (same source). Turnout in uncompetitive commission-drawn districts in Michigan and Colorado ran ~10 points above comparable districts (Brennan Center, "The Turnout Effects of Redistricting Institutions," 2023).

2. Effects on competitiveness are real but modest — moderate-to-thin evidence. Studies of California's Citizens Redistricting Commission find enhanced competitiveness and slightly reduced polarization, with effects described by the authors as "modest to date" (Scientific Research Publishing / Open Journal of Political Science study of Prop 20, 2016). Other peer work finds "little evidence" commissions enhance competition (Best et al., APSA preprint, "Do Redistricting Commissions Avoid Partisan Gerrymanders?"). Takeaway: fair process reliably fixes skew; it only somewhat fixes competitiveness, because…

3. …geography does a lot of the gerrymandering by itself — strong evidence. Chen & Rodden, "Unintentional Gerrymandering: Political Geography and Electoral Bias in Legislatures," Quarterly Journal of Political Science 8:239–269 (2013): using thousands of simulated neutral maps built from precinct-level 2000 presidential returns, they show Democrats' concentration in cities means even intent-free maps give Republicans a seat bonus — Democrats can win 50% of votes and well under 50% of seats with nobody cheating. Direct implication for this plank: an automated map is not a proportional map. Splitline ends deliberate gerrymandering; it does not end pro-Republican geographic bias, and the plank should never promise proportionality.

4. The causal claim about polarization — strong evidence against the plank as written. McCarty, Poole & Rosenthal, "Does Gerrymandering Cause Polarization?", American Journal of Political Science 53(3):666–680 (2009): comparing actual polarization against simulated neutral districting, they find districting explains little of the rise in polarization. Polarization is driven mainly by Democrats and Republicans representing the same kinds of districts increasingly differently. The killer facts: the Senate — which cannot be gerrymandered — polarized on nearly the same trajectory as the House, and members from identical districts diverge by party. Subsequent work (McCarty & Shor 2015 review) confirms. This plank's first sentence is, per the best available evidence, wrong.

5. Automation per se — thin evidence, cautionary literature. Altman & McDonald, "The Promise and Perils of Computers in Redistricting," Duke Journal of Constitutional Law & Public Policy 5:69 (2010): "both the promise and peril of computers have been greatly exaggerated" — computers made map-drawing cheaper and faster (including for gerrymanderers) but every automated system embeds value choices in its objective function and constraints. Choosing the algorithm is a districting decision; the gain from automation is that the choice is made once, in public, in advance, rather than in a back room after the election returns come in.

6. Legal viability — strong. Arizona State Legislature v. Arizona Independent Redistricting Commission, 576 U.S. 787 (2015), upheld taking congressional redistricting away from state legislatures entirely. Rucho (2019) both closed the courthouse door and expressly reaffirmed that the Elections Clause lets Congress regulate congressional districting. Congress has used this power before: it required single-member districts by statute (2 U.S.C. § 2c, enacted 1967) and imposed compactness/contiguity requirements in apportionment acts from 1842 to 1929. A federal automated-districting mandate for congressional maps is on unusually firm constitutional ground compared with most planks' preemption questions. (It could not constitutionally reach state-legislative maps — the plank correctly limits itself to congressional districting.)

Complication (2023–2026): Allen v. Milligan, 599 U.S. 1 (2023) reaffirmed that VRA § 2 can require drawing majority-minority districts — historically the strongest legal objection to race-blind algorithms like splitline. But Louisiana v. Callais (No. 24-109, decided April 29, 2026, 6–3) sharply cut back § 2, holding Louisiana's second majority-minority district was an unconstitutional racial gerrymander and raising the bar for § 2 claims toward intentional discrimination (CRS Legal Sidebar LSB11431, 2026). Post-Callais, the doctrinal conflict between race-blind automation and the VRA is much smaller than it was in 2022 — an uncomfortable assist, since it comes from a decision that weakens minority-voter protections. If Congress later restores § 2's results test by statute, the algorithm must take VRA constraints as input; pure splitline cannot.

Overall strength of evidence: moderate for "automation/commissions end deliberate partisan skew"; thin for "and Congress will therefore moderate."

Who wins, who loses

Transition: one cycle. Maps change once, at the next census (or effective date); there is no phase-in pain in the usual economic sense. No compensation mechanism is proposed or needed except for the VRA problem, which is a design constraint, not a payment.

Fiscal impact

Effectively zero on the federal budget — this is a governance plank, not a fiscal one, and we should say so plainly rather than stretch.

• Direct outlays: de minimis. Census Bureau already produces the input data; certifying an open-source algorithm and publishing maps is single-digit millions one-time (back-of-envelope).
• Direct savings: states and DOJ save recurring redistricting-litigation and consultant costs — plausibly tens of millions per year nationally averaged over a decade (back-of-envelope, weak evidence).
• No CBO score exists for automated districting; the Freedom to Vote Act's scored costs were driven by its election-administration provisions, not its redistricting title.
• The honest fiscal argument is indirect and unprovable in advance: the platform's other eleven planks all die in a gridlocked House; this plank is infrastructure for passing them. Claim it as strategy, not as savings.

The opposition's best case

1. "Gerrymandering doesn't actually cause polarization — you've misdiagnosed the disease." — McCarty, Poole & Rosenthal (AJPS, 2009), the strongest names in the polarization literature. The Senate polarized without districts to gerrymander; polarization shows up within identical districts depending on the party of the representative; neutral simulated maps yield polarization close to actual levels. Concession, in writing: they are right, and the plank's first sentence should be rewritten. Polarization is driven more by partisan sorting, primaries, media, and money than by district lines. What survives: gerrymandering demonstrably shifts seats (Brennan: ~16 net in 2024), entrenches minority rule risk, kills accountability in ~90% of districts, and Rucho left no other remedy. End gerrymandering because rigged maps are illegitimate and seat-distorting — not because it will de-polarize Congress. It probably won't, much.

2. "Race-blind algorithms dismantle minority representation." — the NAACP LDF position across Milligan (2023) and Callais (2026); the classic objection to splitline. Splitline cannot see race, so it cannot create the majority-minority districts that § 2 required from Thornburg v. Gingles (1986) through Milligan (2023). Partial concession: true of pure splitline, and it is why the statute should mandate a constraint-based algorithm with VRA compliance as an input rather than splitline literally. Post-Callais (April 2026) the legal force of the objection is reduced, but the representational concern stands on its own and we should treat it as binding design requirement, not an obstacle.

3. "Communities of interest matter; your algorithm slices cities and counties like a pizza." — Altman & McDonald (2010); California's constitutional criteria (Voters FIRST Act, 2008) put communities of interest above compactness on purpose. A district that merges half of Tulsa with 200 miles of ranchland is "fair" only in a geometric sense. Partial concession: real cost of the pure version. Answer: constraint-based algorithms can respect county/municipal boundaries; and the status quo's alternative use of "communities of interest" has too often been a euphemism for incumbent protection. We accept somewhat worse community fit as the price of removing the pen from interested hands.

4. "Neutral maps still favor Republicans — you're locking in geographic bias and calling it fairness." — the left-of-center reading of Chen & Rodden (2013); proportionality advocates would rather have multi-member districts with proportional representation (e.g., the Fair Representation Act's ranked-choice multi-member proposal) than neutral single-member maps. Concession: correct on the facts — automation ends cheating, not bias. Answer: a known, stable, modest geographic tilt that both parties can see in advance is categorically better than an unbounded, escalating, mid-decade arms race (Texas/California, 2025). PR is a bigger reform with far weaker political feasibility; this plank does not preclude it.

5. "Districting is inherently political; hiding the politics inside an objective function is technocratic evasion, and 'irrevocable' is fantasy." — Altman & McDonald (2010) on the first half; any lawyer on the second. Every algorithm choice (compactness metric, tie-breaking, constraint weights) allocates power, and whoever picks the algorithm picks the winners once. And no Congress can bind a future Congress — a statute is repealable, and Rucho means courts won't preserve it. Concession on "irrevocably": the plank overclaims and the word should go. Answer on the substance: making the value choices once, publicly, in statute, before the census, with open-source code anyone can run, is the opposite of hiding politics — it is the only version of districting where the public can check the work. Durability comes the same way it did for the 1967 single-member-district statute (2 U.S.C. § 2c): once neither party knows who'd win the repeal fight, nobody repeals.

Talking points

• One line: Politicians shouldn't pick their voters — make congressional maps by open-source formula anyone can check, so nobody's thumb is on the scale.
• Thirty seconds: In 2024, only 27 of 435 House races were competitive, and rigged maps shifted roughly 16 seats — more than the House majority's margin. The Supreme Court said in 2019 that courts will never fix this, but pointed at who can: Congress, under Article I, Section 4. Require every state's congressional map to come from a published, open-source algorithm — same code, same census data, same map on everyone's computer. No back rooms, no mid-decade redraws, no exceptions for either party.
• Two minutes: adds — The villains are bipartisan: incumbents with 90-plus percent reelection rates, and party operations like REDMAP, which spent $30 million in 2010 flipping statehouses precisely to control the 2011 maps; Democrats built their own version, and in 2025 Texas and California started redrawing maps mid-decade in an open arms race. The honest caveats: political scientists (McCarty, Poole, Rosenthal, 2009) find gerrymandering is not the main driver of polarization — the Senate polarized too — so we claim fair maps and accountable seats, not a cure for tribalism. And because Democrats cluster in cities, even neutral maps tilt slightly Republican (Chen & Rodden, 2013) — automation ends cheating, not geography. The algorithm must also carry Voting Rights Act constraints so minority representation is protected by design. Commissions in Michigan, California, and Arizona already show what happens when you take the pen from incumbents: partisan skew drops to near zero in one cycle. The formula does the same thing, cheaper, everywhere, permanently — and it's the plank that makes the other eleven passable.

Open questions

1. Which algorithm, exactly? Splitline is rhetorically clean but fails VRA/communities tests. Before this becomes model legislation we need a named, open-source, constraint-based specification (recombination-ensemble selection? optimization with statutory constraint ordering?) and a governance answer for who certifies the code and the tie-breaking seed. An expert should be asked: can a fully deterministic constraint-based method resist adversarial manipulation via the constraint weights?
2. Post-Callais VRA landscape (decided 2026-04-29). How much of § 2 districting law survives in practice, and should the statute hard-code minority-opportunity constraints regardless? Track lower-court application through 2026–2027; this materially changes the design.
3. Does ending gerrymandering moderate anyone? The plank promises Congress "will moderate"; the evidence says probably only slightly. Would we keep the plank if the honest claim is only "fair, stable, legitimate maps"? (Recommendation: yes — and rewrite the plank text. Under the working agreements, that means updating README.md and index.html together.)
4. Pin down dollar figures currently flagged weak: per-state redistricting litigation costs 2011–2025; NDRC and RSLC cycle-by-cycle spending; consultant-market size.
5. 2026 election as natural experiment: compare seat-vote skew in the mid-decade-redrawn states (TX, CA, MO, NC) against commission states after November 2026 — first post-arms-race data point.
6. Does this plank need a companion ban on mid-decade redistricting even before automation passes, as a severable fallback?

Sources

• Brennan Center for Justice, "How Gerrymandering Tilts the 2024 Race for the House" (2024). https://www.brennancenter.org/our-work/research-reports/how-gerrymandering-tilts-2024-race-house — secondary (advocacy-affiliated research; methodology published).
• Brennan Center for Justice, "How Gerrymandering and Fair Maps Affected the Battle for the House" (2024). https://www.brennancenter.org/our-work/analysis-opinion/how-gerrymandering-and-fair-maps-affected-battle-house — secondary.
• Brennan Center for Justice, "The Turnout Effects of Redistricting Institutions" (2023). https://www.brennancenter.org/our-work/research-reports/turnout-effects-redistricting-institutions — secondary.
• Brennan Center for Justice, "The Freedom to Vote Act Is a Big Deal for Redistricting" (2021). https://www.brennancenter.org/our-work/analysis-opinion/freedom-vote-act-big-deal-redistricting — secondary.
• Rucho v. Common Cause, 588 U.S. 684 (2019). https://en.wikipedia.org/wiki/Rucho_v._Common_Cause (opinion at supremecourt.gov) — primary.
• Arizona State Legislature v. Arizona Independent Redistricting Commission, 576 U.S. 787 (2015). https://supreme.justia.com/cases/federal/us/576/787/ — primary.
• Allen v. Milligan, 599 U.S. 1 (2023); CRS Legal Sidebar LSB11002 (2023). https://www.congress.gov/crs-product/LSB11002 — primary (CRS summary of primary).
• Louisiana v. Callais, No. 24-109 (U.S. Apr. 29, 2026). https://www.supremecourt.gov/opinions/25pdf/24-109_21o3.pdf ; CRS Legal Sidebar LSB11431 (2026), https://www.congress.gov/crs-product/LSB11431 — primary.
• U.S. Const. art. I, § 4, cl. 1 (Elections Clause); 2 U.S.C. § 2c (single-member districts, 1967) — primary.
• Freedom to Vote Act, S. 2747, 117th Cong. (2021); Campaign Legal Center, "What the Freedom to Vote Act Means for Partisan Gerrymandering" (2021), https://campaignlegal.org/update/what-freedom-vote-act-means-partisan-gerrymandering — primary (bill text) / secondary (summary).
• McCarty, N., Poole, K. T., & Rosenthal, H., "Does Gerrymandering Cause Polarization?" American Journal of Political Science 53(3): 666–680 (2009). https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1540-5907.2009.00393.x (open PDF: https://legacy.voteview.com/pdf/ajps_393.pdf) — primary (peer-reviewed).
• Chen, J., & Rodden, J., "Unintentional Gerrymandering: Political Geography and Electoral Bias in Legislatures," Quarterly Journal of Political Science 8: 239–269 (2013). https://www.nowpublishers.com/article/Details/QJPS-12033 — primary (peer-reviewed).
• Altman, M., & McDonald, M., "The Promise and Perils of Computers in Redistricting," Duke Journal of Constitutional Law & Public Policy 5: 69 (2010). https://lawreview.richmond.edu/files/2013/03/Altman-473.pdf (companion; original via Duke) — primary (peer-reviewed).
• Smith, W. D., Center for Range Voting, "The Shortest Splitline Algorithm" and "Splitline districtings of all 50 states." https://rangevoting.org/Splitlining.html ; https://www.rangevoting.org/SplitLR.html — primary for the proposal (self-published by proponent; site returned 403 during this research — content confirmed via secondary descriptions; re-verify on second pass).
• DeFord, D., Duchin, M., & Solomon, J., "Recombination: A Family of Markov Chains for Redistricting," arXiv:1911.05725 (2019; Harvard Data Science Review 2021). https://arxiv.org/pdf/1911.05725 — primary (peer-reviewed).
• Duke Journal of Constitutional Law & Public Policy, "The Good, the Bad, and the Ugly: Redistricting Commissions in the 2021 Cycle" (2022). https://scholarship.law.duke.edu/cgi/viewcontent.cgi?article=1205&context=djclpp — primary (law review).
• Best, R. E., et al., "Do Redistricting Commissions Avoid Partisan Gerrymanders?" (APSA preprint). https://preprints.apsanet.org/engage/api-gateway/apsa/assets/orp/resource/item/5d7c0b8ed0706700120e052d/original/do-redistricting-commissions-avoid-partisan-gerrymanders.pdf — primary (unrefereed preprint).
• Study of California CRC / Prop 20 competitiveness and polarization, Open Journal of Political Science (2016). https://file.scirp.org/Html/7-1670438_71249.htm — secondary (low-prestige venue; treat findings as suggestive).
• Cook Political Report, 2024 House Race Ratings and Cook PVI. https://www.cookpolitical.com/cook-pvi — secondary (industry standard).
• FiveThirtyEight, "What We Lose When We Lose Competitive Congressional Districts" (2018). https://fivethirtyeight.com/features/what-we-lose-when-we-lose-competitive-congressional-districts — secondary.
• REDMAP: Wikipedia, "REDMAP," https://en.wikipedia.org/wiki/REDMAP ; PBS NewsHour, "GOP gerrymandering creates uphill fight for Dems in the House" (2017), https://www.pbs.org/newshour/politics/gop-gerrymandering-creates-uphill-fight-dems-house — secondary ($30M/2010 figure traces to RSLC's own reporting).
• OpenSecrets, House incumbent reelection rates series. https://www.opensecrets.org/elections-overview/reelection-rates — secondary (cited from memory; verify exact figures on second pass) — unverified.
• 2025 mid-decade redistricting (Texas, August 2025; California Proposition 50, November 2025; Missouri; North Carolina) — widely reported contemporaneous events; specific seat-shift figures are projections — unverified pending 2026 results.
• NDRC fundraising totals — unverified (order-of-magnitude from press coverage; pin down before "reviewed").
• Mexico INE automated first-draft districting since 2015 — unverified (secondary mentions only; verify before use in the ebook).