The Martha’s Vineyard climate reading

An island’s climate conversation enters — 31 documents, four voices: the town and its harbors, policy, science, and the Wampanoag — spanning the island and the climate-migration domain beyond it. Minted and read the same day. A first read, declared exploratory before it ran. A reading, not a finding.

the readings, printed — by the instrument

A first read, declared exploratory before it ran: no predictions were registered, so nothing below is a tested claim — these are values, laddered, published as such. Coupling per named axis is warp; displacement is observed warp minus the median of a matched-n permutation null (300 draws, seed 7), each construction against its own null. A quoted warp states raw or ridged. The correlation matrices behind every value are published beside this page — reading-data.json, both constructions, its fingerprint in the data island above. local_gov = town & harbor · policy = policy · science = science · wampanoag = Wampanoag.

observed warp, by voice

Observed warp, raw construction, all seventeen axes, canonical order.

displacement against each voice’s own null

The highlighted rows carry this first read’s two starkest value-contrasts. On Agency and Boundary the Wampanoag column shows the largest displacements in the table (+1.90, +1.76). On Telicity and Viewpoint the science and Wampanoag columns carry opposite signs (+0.65/−1.23, +0.94/−1.30) — the completion axes tension one voice and release the other. First-read values against matched-n nulls; none of these values is a tested claim, and what the contrast means is the reader’s.

Displacement against the matched-n null, raw and ridged per voice. Where profiles differ by voice the coupling is conditioned — and what that means is yours to resolve.

bridge anatomy — per-axis separation, by pair

The bridge scalar on the door is a cut; this table is the shape it cuts away. The strongest bridge in the read, science ↔ wampanoag (0.484), spans the widest axis separation in its column — Viewpoint, 2.21 apart — while on Modality Type the same two voices sit 0.01 apart, nearly coincident. Bridge viability is mutual navigability, not sameness; where it rides high across a wide gap, that gap is part of the reading.

Per-axis warp separation, raw construction: the absolute difference between the two voices’ observed warp on each named axis — arithmetic on the warp table above, nothing newly computed. Each bridge on the door compresses one of these columns to a single number.

floors, as values

Floors are the diagonal — per-axis variance, the ground the warp above stands on. The thinnest floor in this read is science on Transitivity (0.0013 raw). On Telicity — the axis where the science and Wampanoag displacements oppose — the floor thins from 0.0502 under science to 0.0181 under the Wampanoag: the axis they contest is one voice’s thick ground and the other’s thin ice. Where a floor is thin, the declared +0.01 ridge does proportionally more of the ridged column’s work.

Floor values as values, both constructions; the ridged floor carries the declared constant inside it.

what was read — a first snippet

Four things surface from the tables above, offered as values with the resolution left to you. The Wampanoag voice carries the largest displacements in the read — Agency +1.90, Boundary +1.76, Influence +1.63 — the axes whose framing questions are who can act, who is inside this decision, whose voice carries. The science and Wampanoag columns pull opposite ways on the completion axes — Telicity +0.65 against −1.23, Viewpoint +0.94 against −1.30: one voice tensions endings and open-or-closed, the other releases them. The strongest bridge on the island runs between those same two voices (0.484), and its anatomy is printed above: widest apart on Viewpoint (2.21), nearly coincident on Modality Type (0.01) — a bridge is mutual navigability, not sameness. And the curation note that travels with it, document by document: the science register is four monitoring and coastal-observation documents, three risk-research documents, and one scientist’s bio; the Wampanoag register holds two climate-adaptation plans and a restoration project beside cultural, biographical, and land-context documents. Where the writing is genre-adjacent — plans, monitoring, practice — is one candidate reading of the bridge; it is not the only one. And the policy voice decouples on Causality — −1.60 where every other voice sits at +0.05 or above — the planning register, itself almost entirely the island’s own, pulling away from the three voices it plans among. First-read values, single mint — none of these four is a tested claim.

a first read, and what that means

The declaration was written before the run: no predictions were registered, so nothing in the tables above is a tested claim — values, laddered, published as such. Of the two registered tests this corpus exists for, the first has since run under its own registration (2026-08-14): does coupling move with domain — island against beyond-island — the way it moves with voice? All three of its registered predictions were falsified. Against document-permutation nulls, free and register-stratified, the beyond-island partition’s displacement profile sat mid-null — rank 194 and 233 of 301. At this size — seven beyond-island documents, single mint — the domain partition reads as a relabeling. That is a bound on this corpus, not a law about islands — and it bounds domain only: the test neither touches nor upgrades the voice displacements above, which remain first-read values, untested. The second registered test — what near-equal presence changes — has not run and requires its own registration first. This page is not a replication of the Cape Ann reading and will not be described as one; comparisons between the two are values beside values.

the fourth voice

Seven of the thirty-one documents are the Wampanoag voice — public documents from public sources, read after conversation with Wampanoag representatives, under a hard-stop this architecture enforces rather than promises: nothing trains on anything, here or anywhere in this instrument. Presence in a corpus is a condition, not a sample — and this voice’s standing with this place is measured in millennia, not documents. Its rows are printed in the same tables as every other voice, at its own presence, against its own null. What its geometry means is not ours to say.

how this reading is kept

Every document was extracted through the live path on 2026-08-10, one fresh session per document, and frozen as per-document artifacts with provenance sidecars — one machine, one mint. A single-mint corpus supports no independence claim across builds, and none is made: that limit is part of the record. A quoted warp states raw or ridged, here as everywhere.

what this reading does not say

Seventeen thousand filings, one hundred sixty-four letters, a cape’s climate record — and now an island’s, four voices, read the day it was minted. One instrument, unchanged.

← habitat · the Cape Ann reading · method