Provenance. §1 = reviewed unit (candidate gold, Watson-grounded; inter-annotator agreement is an LLM-panel proxy for human IAA). §§2–3 = candidate (proxy IAA, not human-calibrated). §§4–73 = machine-assisted, unreviewed heuristic exploration. Interpretive layers are valid-but-heuristic rule-sets — coverage ≠ accuracy, not a digital edition. Editorial statement · Error budget

Honest Error Budget

Coverage measures ANALYSIS, not accuracy. Structural defaults (EVENT_ANCHOR/PARTICIPANT role scaffold, PREDICATION frame, given/topic info, unscored rhetoric) are excluded from coverage.

Per-layer coverage (content tokens = 5818)

layerhonest coveragenote
thematic role0.55% (gross incl. EVENT_ANCHOR scaffold: 28.62%)honest = real thematic roles (agent/patient/...); gross counts EVENT_ANCHOR/PARTICIPANT structural scaffold
event frame11.77%excludes PREDICATION/CONTEXT/EVENT filler
information structure27.04%excludes given + definitional clause-head topic
rhetoric14.92%categorical priors, NOT measured accuracy

Open review queue (the deficit)

1834 open review items of 1873 total ({'already-reviewed': 39, 'review-needed': 1834}). This is the published error budget the project owes the reader.

Section status (first 12; full in edition_status.json)

§tiercontentroleframerhetoricqueue open
§1reviewed-unit5732345718
§2candidate-iaa82071537
§3candidate-iaa76015821
§4exploration91052044
§5exploration850171834
§6exploration82001737
§7exploration5800719
§8exploration790151125
§9exploration4901413
§10exploration7005818
§11exploration990202139
§12exploration710151431

Status computed by compute_edition_status.py — data-driven, cannot be hand-set.