Honeycombing

It lays an H3 hexagonal grid over a state as a neutral reference frame and aggregates census and precinct-level election data onto it. Separately, it locates the enacted congressional map within an ensemble of 5,000 computer-simulated alternative maps drawn under documented redistricting constraints (equal population, contiguity, compactness, county preservation, and a Voting-Rights-Act-aware target). The question is positional: is the enacted map an ordinary draw from the space of constraint-satisfying maps, or an outlier?

North Carolina case study, using 2020 presidential returns as the partisan-lean proxy: the ensemble's median outcome is 6 Democratic seats of 14. The 2023 enacted map yields 4, near the 2.8th percentile of the distribution; the 2022 court-ordered plan it replaced sits near the 86th. The method's precinct-assignment approximation was validated against exact assignment (≤0.12 percentage-point district-level difference).

Scope and caveats: one state, prototype-grade, pre-publication; partisan lean is proxied by presidential vote, not congressional results. Percentile positions are relative to the ensemble's documented constraints and are not evidence of intent or legal injury.