> For the complete documentation index, see [llms.txt](https://osintelligence-llc.gitbook.io/osintelligence/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://osintelligence-llc.gitbook.io/osintelligence/part-iii-the-evidence-what-broke.md).

# Part III · The evidence: what broke

The failure record, placed at the centre of the book deliberately. Nine failure classes, a natural experiment caught at 23:26Z, 1,142 hard blocks of internal telemetry, and the same failures documented in other organisations' systems. A reader who has seen this record reported honestly has reason to believe the results that follow.

***

[**9 · The Drift Taxonomy**](/osintelligence/part-iii-the-evidence-what-broke/9-the-drift-taxonomy.md) · Nine failure classes from sustained production, the architectural-softening cascade as the worked example, and the admission-rate arc. The raw source artifact is reproduced in [Appendix A](/osintelligence/appendices/appendix-a-the-cascade-catalogue-raw.md).

[**10 · The Guard Changes at 23:26Z**](/osintelligence/part-iii-the-evidence-what-broke/10-the-guard-changes-at-23-26z.md) · A natural experiment caught mid-transition: what a model swap does to a governed system, tracked observable by observable against the same enforcement surfaces.

[**11 · The Hook Telemetry Record**](/osintelligence/part-iii-the-evidence-what-broke/11-the-hook-telemetry-record.md) · Sixty-four days of per-hook telemetry from the deterministic layer: 390,515 events, 1,142 hard blocks, every figure a direct count.

[**12 · The Public Case Record**](/osintelligence/part-iii-the-evidence-what-broke/12-the-public-case-record.md) · The same failure classes documented in other organisations' systems, compiled from primary sources under an explicit inclusion rule, with the exclusions named.