Open the interactive simulator →
An electoral map looks like a technical document. It shows settlements, administrative boundaries, roads, rivers, and neatly drawn district lines. This technical appearance is exactly what creates the impression that the map merely records objective geography.
But an electoral district does not exist in nature. Someone creates it.
Someone decides which territories to combine, which communities to split, how much weight to give compactness, whether crossing municipal boundaries is acceptable, and whether the composition of the population should be taken into account at all. Even the decision not to consider population composition is itself a choice — a choice in favor of whatever consequences a formally neutral map will produce.
So the central question isn’t whether districting is political. It inevitably is. The real question is: who draws the lines, for what purpose, under what rules — and can the cost of that decision be checked?
Formal equality does not guarantee representation
In its simplest form, equal elections work like this: each person has one vote, each vote is counted the same way, and districts are comparable in population.
But individual vote equality does not by itself mean equal ability to shape power.
A social or ethnocultural group may make up a significant share of the population, share common interests, and vote consistently — and still end up with no representative at all. This doesn’t require banning its candidates from running or its members from voting. It’s enough to draw the lines so that the group remains a stable minority in every district.
Particularly vulnerable are communities that are:
relatively small in number; excluded from major parties and political funding; geographically concentrated, yet split apart by boundary lines; distributed across territory in a way that leaves them nowhere in the electoral majority.
The result: every individual keeps the right to vote, but the group as a political community is systematically “dissolved” into the majority.
This is exactly the gap — between the right to vote and the real chance to elect a representative — that our districting simulator is built to show.
One territory, four outcomes
In the model, the territory consists of 64 equally-populated cells. It is divided into eight single-member districts of eight cells each. Part of the population belongs to Group B, shown in orange; the rest of the residents make up Majority A.
Users can adjust the share of Group B, its degree of territorial concentration, and the configuration of the districts. They can choose one of four ready-made schemes or draw their own map by hand. The population stays the same — only the boundaries change.
Neutral districting. In the baseline scenario, the territory is divided into equal administrative blocks without regard to where Group B lives.
It’s tempting to call such a map neutral: the algorithm doesn’t try to help or hurt the group. But a neutral procedure doesn’t guarantee a neutral outcome.
If a compact community happens to be split across several rectangular districts, it may end up with no seats at all. Under a different settlement pattern, the very same geometrically regular boundaries could give it one or two districts.
A neutral rule isn’t necessarily unjust. But it doesn’t exempt anyone from examining its consequences.
Cracking. Under cracking, members of the group are spread across districts so that they form a majority almost nowhere.
Say Group B makes up a quarter of the population. If its members are distributed roughly evenly across all eight districts, they’ll be a noticeable share of each one — and lose in all eight. Twenty-five percent of the population can end up with zero percent of the seats.
Everyone voted. Every ballot was counted. All the districts may be equal in population and even reasonably compact. And yet the design of the map turns a stable social group into a permanent electoral minority.
Packing. Under packing, as many members of the group as possible are concentrated into one or a few districts.
The group wins that district comfortably — but with far more votes than needed. If five votes out of eight are enough to win, the sixth, seventh, and eighth don’t add any more seats. In the remaining districts, too few of the group’s members are left to compete.
So packing doesn’t always eliminate representation entirely. It can grant the group one guaranteed seat while minimizing its influence everywhere else.
Opportunity districts. The fourth scenario is a protective map, or “opportunity plan.” The algorithm estimates roughly how many seats would correspond to Group B’s share of the population and tries to create several districts where it has a real chance of electing its preferred candidate.
This isn’t an “objectively fair map.” It’s a map deliberately optimized for one particular fairness criterion — effective representation.
It may bring the share of seats closer to the share of population, but it can create new problems: less compact districts, crossed administrative boundaries, or excessive weight given to a single population characteristic. So the simulator doesn’t declare the protective map correct in advance. It checks this map the same way it checks all the others.
A map has to be able to criticize itself
The most important part of the simulator isn’t the colored cells — it’s the audit panel.
Users can deliberately draw a single-cell district, an empty district, a territory made of disconnected islands, or a map with wildly unequal populations. The simulator doesn’t block these choices. It keeps counting winners — while also flagging violations.
Four properties are checked:
equal population across districts; territorial contiguity; average compactness; fragmentation of Group B’s compact communities.
This is a deliberate design choice. An educational model shouldn’t quietly steer the user toward whatever answer its author considers correct. It should let people build a bad map and see exactly what’s wrong with it.
But the panel also reveals a harder problem: improving one metric can worsen another.
A geometrically compact district can cut a long-established community in two. A district that keeps a community whole may end up an odd, elongated shape. Strict population equality can conflict with municipal boundaries or transit connectivity. A map that gives a minority two competitive districts may look messier than a map that gives it none.
There is no single geometric shape for fairness.
Not every accommodation of a minority is gerrymandering
The word “gerrymandering” usually implies abuse: boundaries drawn deliberately to entrench partisan power or weaken particular voters.
That’s why the phrase “positive gerrymandering” sounds striking but actually conflates two different things.
One is manipulating boundaries for a predetermined political winner. The other is accounting for a map’s consequences in order to prevent the systematic dilution of a minority’s voice.
The technique can look similar — in both cases, population composition affects where the lines fall. But the legal purpose, the procedure, and the system of constraints are different.
Better terms are opportunity districts, protective districting, or community-conscious districting. None of these concepts assume that a good intention automatically justifies any configuration. On the contrary, they require proof that intervention is necessary and an account of every trade-off made.
The United States: between vote dilution and racial classification
American law illustrates the internal tension of protective districting with particular clarity.
Section 2 of the Voting Rights Act prohibits election rules and practices that give citizens protected on the basis of race, color, or membership in certain language minorities less opportunity to participate in the political process and to elect representatives of their choice.
In Thornburg v. Gingles (1986), the Supreme Court set out three threshold conditions for vote-dilution claims: the minority must be sufficiently large and geographically compact to form a majority in a reasonably configured single-member district; its voters must be politically cohesive; and the majority must vote as a sufficiently cohesive bloc to usually defeat the minority’s preferred candidate.
This test made it possible to challenge maps that, through cracking or packing, effectively denied racial minorities the ability to elect candidates of their choice.
But American constitutional doctrine simultaneously limits the use of race in drawing districts. In Shaw v. Reno, the Supreme Court held that deliberately sorting voters into districts predominantly by race can violate the Equal Protection Clause. A district’s bizarre shape is not by itself a separate legal test, but it can be evidence that traditional criteria — compactness, contiguity, administrative boundaries, and genuine shared interests — were subordinated to racial classification.
The result is an almost paradoxical construction: a legislature sometimes must account for a map’s racial consequences to avoid diluting a minority’s vote, yet may not let race dominate uncontrollably over every other criterion.
A new turn in 2026. In April 2026, the Supreme Court substantially tightened this framework in Louisiana v. Callais.
The majority updated the Gingles test: an alternative map proposed by plaintiffs must not use race as a basis for construction and must perform no worse than the state’s map on all its permissible goals — including traditional criteria, incumbent protection, and stated political goals. When analyzing cohesive voting, race must be separated from party affiliation. The Court also required the totality-of-circumstances analysis to focus on current data and present-day evidence of racial discrimination.
The majority did not formally state that every plaintiff must directly prove discriminatory intent. However, the dissenting justices noted that the new requirements effectively bring the proceeding closer to a search for specifically racial motivation and weaken the original analysis of discriminatory effects enshrined in the 1982 amendments to the Voting Rights Act.
For our purposes, this dispute matters more for what it reveals than for its outcome in one particular American case. It shows how hard it is to draw a legal line between correcting structural inequality and impermissibly classifying people by identity.
The US therefore offers not a ready-made model to copy, but a warning: even a mature body of case law can spend decades without settling, once and for all, when accounting for group membership restores equality — and when it creates a new inequality.
Māori electorates: neither a quota nor an ordinary district
New Zealand offers a different model.
Four special Māori electorates were established by the Māori Representation Act 1867. After the Electoral Act 1993, the fixed number was replaced by a formula tied to the size of the Māori electoral population. Seven Māori electorates remain in place for the 2026 elections.
This system differs both from an ordinary opportunity district and from a pure quota.
Voters of Māori descent can choose which roll to join — the general roll or the Māori roll. Those on the Māori roll vote for a candidate in their Māori electorate; those on the general roll vote in the corresponding general electorate. Since 2023, the roll can be changed outside the narrow window that used to be the only opportunity to switch, though temporary restrictions apply ahead of elections.
The MP is not appointed by the state and does not receive the seat automatically — they win it in an election. But the very existence of a separate electorate, and the right to vote in it, are tied to a specific electoral roll.
So it’s more accurate to describe Māori electorates as a hybrid of territorial representation, a distinct electoral mechanism, and guaranteed institutional seats. This model shows that the choice between ordinary districts and quotas isn’t binary — it’s a whole spectrum of possible designs.
Can districts replace quotas?
Sometimes — yes. But far from every underrepresented group.
Protective districting makes sense when a group:
genuinely faces persistent underrepresentation; has a sufficiently clear territorial connection; lives compactly or forms recognizable communities of interest; can gain a real, not merely symbolic, chance of election; itself regards territorial representation as an appropriate form of protection.
For a compactly settled Indigenous, ethnic, or linguistic community, an opportunity district can preserve the link between representative and territory while still allowing election through the ordinary competitive process.
But the territorial model fits poorly for groups spread evenly across a country. Ensuring women’s representation through districting alone, for example, is almost impossible: women don’t occupy a distinct geographic territory. The same holds for many people with disabilities, professional groups, or dispersed diasporas.
In such cases, party quotas, reserved seats, proportional systems, special rolls, or other mechanisms may work better.
Districts, in other words, are not a universal alternative to quotas. They are one tool of institutional design, applicable only where a political community has a territorial dimension.
Why this matters for Indigenous peoples
Everything above can sound like an abstract debate about geometry and procedure. For Indigenous peoples, it rarely is.
Indigenous nations and communities are, almost by definition, the case districting was never designed around. Colonial and post-colonial states drew their internal administrative maps — provinces, districts, electoral boundaries — over territories that already had their own political geography: clans, bands, villages, seasonal migration routes, language groups. Those older boundaries were treated as irrelevant, or actively erased, when the new ones were drawn.
The consequence is structural, not incidental. A single Indigenous nation can be split across several electoral districts, several administrative regions, sometimes several states, simply because the map was drawn by someone else, for someone else’s purposes, without asking. Once that happens, the nation stops being one political community in the eyes of the electoral system. It becomes several small minorities, each too small to matter in its own district — the same mechanism the simulator calls cracking, except it was never framed as districting at all. It was framed as “how the map has always looked.”
This is why the Māori electorates example is not just an interesting comparative case — it is close to the only real historical instance where a settler state built a dedicated territorial-representation mechanism for an Indigenous people, tied to genuine political agency: people choose their own roll, vote for their own candidate, hold an actual seat in Parliament. It shows that the “opportunity district” logic can be extended from racial minorities in general to Indigenous peoples specifically, and that doing so is not the same as a quota or a symbolic gesture.
But the same case also shows the model’s real precondition, one that matters more for Indigenous peoples than for almost any other group: self-determination has to sit at the center of the design, not at its edge. A map “for” an Indigenous community, drawn without its consent, by an external commission that decided who counts as the community and what representation it should have, repeats the very pattern of external classification that Indigenous rights frameworks — FPIC among them — exist to prevent. The legitimacy criteria set out earlier in this piece — proven underrepresentation, a genuine territorial basis, participation by the people affected, public criteria, independent oversight, periodic review — read almost like a paraphrase of what free, prior and informed consent already requires of any decision affecting Indigenous territory. Districting is not exempt from that standard just because it looks like a technical, internal matter of electoral law rather than a resource project.
There is also a narrower, very concrete stake: Indigenous communities are frequently the ones directly affected by extractive projects, land-use decisions, and environmental policy that get decided at the level of the very representative bodies whose districts are in question. A community cracked across several districts loses not an abstract seat, but real leverage over decisions about the land it lives on. In that sense, the question this piece keeps returning to — who draws the lines, and can that decision be checked — is not separate from the question of who gets to decide what happens to Indigenous land. It is very often the same question, one step removed.
The cost of protective districting
Even with a good purpose, serious risks remain.
Packing disguised as protection. Power holders can create a single district with an overwhelming majority of the group and declare the problem solved, while its influence everywhere else is wiped out.
Legitimizing partisan manipulation. Protecting a minority can be used as public cover for a map actually designed to preserve a particular party’s power.
Essentialism. The very notion of a “group representative” assumes the group’s members share a common political interest. But Indigenous peoples, language minorities, migrants, and residents of a region do not vote as a single organism.
Substituting external classification for self-determination. A state can unilaterally decide who counts as a community and what kind of representative it needs, without asking the people it claims to protect.
Conflict with vote equality. If districts of sharply unequal population are created for the sake of representation, one vote becomes heavier than another.
Freezing a temporary measure in place. A map justified by a particular settlement pattern and political situation at one moment can persist long after those conditions have changed.
When is protective districting legitimate?
The project’s original presentation set out six baseline requirements: proven underrepresentation, territorial community of interest, an independent commission, public criteria, judicial review, and a ban on using minority protection for partisan gain.
For a fuller legal model, this list is worth expanding.
Protective districting can be considered legitimate when:
The problem is proven with data. A general claim that a group is historically vulnerable is not enough. It must be shown how a specific district system affects the group’s ability to participate in elections and elect representatives. A territorial basis exists. The community must have a genuine geographic connection, rather than being artificially assembled on the map to produce a desired result. Affected people take part in the decision. A community cannot be “protected” without giving it a chance to define its own interests, boundaries, and preferred form of representation. Criteria and data are public. The public should know which metrics were optimized, what alternatives were considered, and why one map was chosen over another. An independent body makes the decision. Those who directly benefit from winning seats under a map should not be the sole authors of its boundaries. All maps face the same scrutiny. Population equality, contiguity, compactness, administrative boundaries, and community preservation should be assessed for protective maps and formally neutral ones alike. There is judicial and public oversight. Interested parties must have access to the data, the algorithms, the hearings, and an appeals process. The measure is periodically reviewed. Demographics, political behavior, and community needs change over time. A protective mechanism must not turn into a permanent, unaccountable classification.
What the simulator doesn’t prove
The model deliberately simplifies reality.
It assumes every cell has an equal population, that all members of Group B vote the same way, that elections are single-member and decided by simple majority. In reality, groups are internally diverse, turnout varies, parties form coalitions, populations are unevenly distributed, and representatives don’t necessarily belong to the community whose interests they express.
The cracking, packing, and protective-districting algorithms are also heuristics, not exact solvers. They don’t find a mathematically optimal map and sometimes produce imperfect districts. That’s precisely why the simulator separately displays contiguity, compactness, and other constraints instead of hiding them.
The model doesn’t prove that a group’s share of seats should always match its share of the population. It doesn’t claim that every minority has a right to its own representative. Nor does it settle when collective representation should outweigh individual freedom of political choice.
It does one thing: it makes the mechanism visible.
The same group can end up with zero, one, or two seats — not because its numbers or political preferences changed, but because someone drew the lines differently.
A boundary is a decision
Maps carry a peculiar kind of power: a political choice, once drawn as a line, begins to look like natural geography.
But no rectangular map, no compact map, no opportunity-district map is a neutral reflection of territory. Each embodies a particular view of what matters most: geometry, administrative integrity, partisan competition, population equality, community preservation, or effective representation.
The problem isn’t solved by banning any consideration of identity. Sometimes it’s precisely such a ban that lets formally neutral boundaries continue a pattern of structural exclusion.
But it isn’t solved by permitting districts to be drawn for any proclaimed good purpose either. Without a transparent process, protecting a minority easily becomes a convenient language for the next manipulation.
So the right question isn’t:
Can gerrymandering be used for good?
It’s more precise to ask:
Under what conditions is a society entitled to take an underrepresented community’s position into account when drawing boundaries — and what constraints will keep that protection from becoming just another instrument of power?
The simulator doesn’t offer a single answer.
It doesn’t ban bad maps.
It shows them honestly.