Answers are grounded in this paper and cite their sources.
Consolidated legal codes are versioned information objects whose quality depends on traceable provenance and consistent terminology. This exploratory study presents an open, reproducible protocol for auditing the documentary consistency of the amendment record of Kazakhstan’s Criminal Code, 2015–2026, and its terminological substitutions using official sources. The consolidated text, revision history and 79 laws cited in its footnotes were hashed, and the laws’ 1754 directives were checked against footnotes and the consolidated text. Footnotes and directives identify the same law–article pairs with precision of 0.986 and recall of 0.983, and with 0.973 and 0.996 on held-out Criminal Procedure Code points of the same laws (post hoc). Under the protocol rule, 84% of checkable lexical changes are visible in the current text. Only 70 of 262 lexical replacements substitute term-level expressions; repeated replacements were largely uniform within laws, while consistency across laws could rarely be tested. Concentration indices over surface forms registered inflection, not terminological choice. Six editorial notices, found post hoc by reading final provisions that the pipeline does not parse, misstate the entry-into-force date of the changes they announce. Post hoc, two legal experts jointly judged that none of the 31 possible inconsistencies signal a real defect and that all six notices are contradictory. Machine signals mark places to read, not legal errors.
No takes yet. Share an insight, caveat, or question.
Malikova et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: