Physically Complete, Semantically Indeterminate: Field-Level Evidence Coordinates for the Analytical Usability of Ecuador's Official Illicit-Firearms Records, January–July 2026

Resumen

Official administrative records can be physically complete and still fail to carry usable information. We audited the open dataset on illicit firearms published by Ecuador's Ministry of the Interior for January–July 2026 (7,252 records, 27 fields) against its published data dictionary and proposed, for each record, separately grounded evidence coordinates for the proposition "the record is analytically usable": T from positive validity evidence (dictionary-consistent values in  attribute fields), I from explicit indeterminacy (sentinel values, values undefined in the dictionary, schema non-applicability) and F from contrary evidence in  cross-field consistency checks, with duplication risk D recorded separately and no unit-sum constraint. The file has no null cells, yet clase_arma holds the sentinel SE DESCONOCE in 1,797 records (24.78 %); the sentinel is itself a dictionary category, 1,702 of 1,704 REVOLVER records carry it, and the conditional entropy of clase_arma given the object type is 0.026 bits: the indeterminacy is schema-induced, not a recording failure. Contrary evidence was almost absent (  in 2 records). The national mean indeterminacy was  (  in 28.0 % of records) and fell to 0.009 when clase_arma was treated as inapplicable to revolvers. Across the 63 cantons with at least 30 records, mean I ranged from 0.017 to 0.096 and correlated at  with the value expected from each canton's object mix; quality-aware volume (Σ T) reproduced the raw canton ranking (Spearman ). Exact repeated rows (1,972, in 714 configurations) clustered in same-day, same-subcircuit batches consistent with multi-weapon seizures and were not counted as falsity. A first version of this audit had labelled 34.7 % of records T, 48.0 % I and 17.3 % F with a mutually exclusive rule; the revised coordinates show that most of that indeterminacy and all of that falsity were artefacts of the partition. Results support quality-aware territorial review; they do not estimate crime rates, population risk, causal effects or police effectiveness.