Where the plan meets the water — a COA as a route through a banded surface

SPEC-19 · reviewed, endorsed, and ratified (ASSAY-DEC-36) · 2026-07-15 · research note 10-spatial-temporal.md (DEC-11). Every region is its fixture bounding box, every route its stated plans.json legs, every channel value the real compiled world — a projection of fixtures, not a drawing (DEC-4). Dragging a waypoint authors a new plan version and re-scores it through the real in-browser pipeline; what changed glows (G6). No assessed value renders as a single stop anywhere (G2). The same surface now also runs as the Spatial · COA tab of the live demonstrator, over the app's single store — where the map refuses while K12 is contested and appears, glowing, on resolve.

The Meridian scenario is engineered fiction (DEC-8). Every banded value on this page is an assessment, not a fact; bands are shown as bands — there is no midpoint, no most-likely value, no heat stop standing in for an interval (DEC-15).

The review, and its outcome. Research note 10 closed with two questions — is the spatial treatment (split fill, hatch for band width, exact [lo, hi] on hover, regions first) honest and legible, and is the temporal treatment (discrete validity/task extents, the K5→K9 supersession edge, lapse marking, the quoted-not-curved storm peak) honest and legible? The SME checkpoint endorsed the UI model (2026-07-15; findings ledger ASSAY-FIND-7, comms §12), and the conventions + promotion ratified as ASSAY-DEC-36 (register batch 6, closing concept §6 item 21). Still worth trying yourself: find the K6→C4 exposure riser in fac_waters and the K7 route-miss (air_defence untouched) without being told — further reactions remain welcome through the same listening loop.