The Setup
Boston holds 992 civic records — 29% of every civic record in the corpus, and the second-largest metro by volume behind San Francisco. None of them has ever been scored for hostility, upzoning probability, or litigation risk. civic_records stores government transcripts, notices, agendas, and permits, each of which can carry three composite scores: hostility_index, upzoning_probability, and litigation_risk_score. Across the full corpus of 3,400 records, only 364 — about 11% — carry any of the three. Those 364 are concentrated in exactly three metros: San Francisco (336 of its 1,169 records, 29%), Chicago (25 of 202, 12%), and Houston (3 of 14, 21%). Every other metro tracked — Boston, Nashville, Phoenix, Miami, Atlanta, LA, NYC — sits at zero.
Boston is the outlier worth naming because of scale. Nashville (573 records), Phoenix (248), and Miami (150) are all smaller unscored pools. Boston's 992 records is the second-largest civic-records collection in the platform, trailing only SF's 1,169 — and unlike SF, none of it has ever been touched by the scoring layer. The feed itself is not stale: records run from April 5 through September 1, ingestion current within days. The gap is scoring, not collection.
The Chain
All 992 Boston records are record_type = 'transcript' — the same record type that accounts for 339 of the 364 scored records nationally, 336 of them SF's. Transcript is not an unscoreable type; SF proves the scoring pipeline runs against it at meaningful volume. What differs is which metro's transcripts the scorer has actually processed. Three metros account for effectively all coverage; seven, Boston included, account for none. That points to a scoring job scoped to a metro allowlist — or one that silently stopped expanding past its original three targets — rather than a type-based exclusion.
The Implication
Litigation risk, upzoning probability, and hostility index are the fields meant to tell a cell-scoring model whether the local civic environment is adversarial or accommodating to development. For Boston, that whole channel of civic signal is dark, even though the underlying raw material — 992 transcripts, current within three days — is sitting in the table. A composite cell score for Boston is implicitly treating civic hostility as zero everywhere in the metro, not because Boston's civic environment is quiet, but because nothing has ever read the transcripts to find out.
The asymmetry compounds at the portfolio level. A model comparing development risk across metros is, in practice, comparing SF's actual measured hostility against six other metros' silent zero — Boston included — which is indistinguishable in the data from "no civic friction here" even though it really means "never scored." Any ranking or filter built on these three fields will systematically favor unscored metros over scored ones, independent of what's actually happening on the ground in each.
What to Watch
Whether the metro allowlist for hostility/upzoning/litigation scoring expands past SF, Chicago, and Houston, and whether Boston's 992-record backlog gets a first pass or keeps growing unscored month over month.
Limitations
This reflects the state of hostility_index, upzoning_probability, and litigation_risk_score as populated in civic_records at query time. It does not evaluate the scoring methodology itself, only its metro coverage, and does not rule out an intentional, not-yet-expanded rollout rather than a stalled one.
--- Data as of 2026-09-04. Source: Locus `civic_records`, metro_slug='boston' (992 records, all record_type='transcript', 0 with hostility_index/upzoning_probability/litigation_risk_score set, earliest 2026-04-05, latest 2026-09-01) vs. full-corpus scoring coverage (3,400 records total, 364 scored, concentrated in metro_slug IN ('sf','chicago','houston')).