Origin Map

where Bay Area artists come from → where they played

BLOCKED.md — venue-anchored framing gate failure (Phase 0)

Project record, rendered verbatim from BLOCKED.md in the source repository.

Recorded: 2026-09-08T01:55Z. See PROBE.md for the full probe report.

What we found

What we expected

≥ 20 events at Gilman and ≥ 70% origin-coordinate resolution, per the Phase 0 gate thresholds.

The three options we see

  1. Proceed with the fallback build (what the gate prescribes for this outcome): the same origin→venue map, seeded by the 1,577 Bay-Area-origin artists (capped at 1,500), with the venue/edge layer populated from whatever real MusicBrainz events those artists have — sparse but honest, degrading gracefully. The venue layer stays additive in the schema.
  2. Re-anchor on the venue with the most events (Shoreline, 72). Rejected: still under any reasonable threshold for a venue-*anchored* framing, and it abandons the Gilman/DIY concept without fixing the underlying sparsity.
  3. Stop and wait for a richer event source. setlist.fm is excluded by instruction; no other free keyless source covers per-venue gig history. Rejected: the run brief says thin venue data must degrade gracefully, not block the run.

Decision

Proceeding with option 1, as the gate itself instructs ("proceed with the fallback… The site still ships"). The venue-anchored build is NOT being built; what ships is the Bay-Area-origin-seeded map with a sparse, honestly-labeled venue layer. /gaps will carry the counts, including per-venue event sparsity, so the hole is visible rather than papered over.

Trim rule for the 1,577 → 1,500 cap: keep all artists with a P740 formation location first (bands — the heart of the concept), then fill with P19 birthplace artists in ascending numeric QID order. Deterministic and recorded in the DB (seed table keeps the full pool with an included flag).