heavymap

Adapted from PR #58 (hm/fresh-foundation @ 15adec2) for the #57 layout; CLI/registry-specific parts deferred to a later code PR.

Open decisions

These questions remain for Morgen. Existing policy in AGENTS.md, AGENTS-MCS.md, CONTRIBUTING.md, and heavymap-planning/DATASET-ROUTINE.md still applies while they are open.

  1. defer vs needs_more. Stage 7 of heavymap-planning/DATASET-ROUTINE.md says defer; the proposed vocabulary says needs_more. Choose one term before changing the routine or sheets.
  2. MCS dataset_id vs layer layer_id. Is one MCS dataset a layer or a family? Define how Spreadsheets/17-layer-profile.csv relates to the MCS and when its dataset reference is filled. Do not infer the mapping.
  3. D-13 and owner fields. first-attempts/niagara-atlas/TECHNOLOGY-DECISIONS.md proposes retaining owner data at ingest and later presentation tiers; the current heavymap-planning/DATASET-ROUTINE.md records field names only. Confirm how any future private ingest would work. Until then, put no owner values or real personal data in docs, sheets, PRs, or comments, and publish none.
  4. NIA-86 join grade scale. heavymap-planning/DATASET-ROUTINE.md calls for a grade, but the thresholds and denominators need a decision. Define them before treating grades as comparable.
  5. MCS location conflict. Root AGENTS-MCS.md says the workbook is outside the repo; root CONTRIBUTING.md says the matching workbook and CSV are tracked under Spreadsheets/. Decide the authority and edit process for the existing sheets. Do not relocate them or replace the MCS with another authority here.
  6. G6. Decided 2026-10-18: G5 and G6 are merged into one staged per-jurisdiction beta gate (Stage A reconciled, Stage B publishable). Public publish or deploy still needs Morgen’s sign-off on NA-269. Publish/deploy tooling may be built and tested (HARD-RULES 3). Details: heavymap-planning/BETA-GATE-HAMILTON.md. PR #80 records the migration plan; execution and naming changes need a separate assignment from Morgen. Preparing or merging frontend work does not execute the split.
  7. Score formula. PR #58 proposed v1.0 weights (joinability 25, coverage 15, openness 15, …) in its unmerged data/schema/controlled-values.json. They need calibration on the Welland/Rochester pilot in Spreadsheets/15-pilot-join-matrix-welland-rochester.csv. Decide whether to use weighted totals or keep component assessments only; blank remains unmeasured.