Skip to content
← All modules

04

Provenance and Audit Trails

How does a system prove what it did?

Building for the stranger who arrives fourteen months later with adversarial intent.

6 lessons · 9 checks · 54 min

Why it sits here

Fourth because determinism is what makes replay possible and replay is what makes provenance meaningful. You cannot prove what a system did until its decisions are reproducible, and you cannot reproduce them until the model is out of the hot path. This module is where module 03 pays off.

Lessons

  1. Nine terms that are not synonymsMost arguments about audit systems are vocabulary collisions. Fixing the words fixes half the disagreement.8m
  2. Event sourcing, and the two things that breakNeither of them is the fold. Determinism is lost somewhere else entirely.10m
  3. Accounting got here firstThe framing is claimed, canonically, since 2015. What survives is the part nobody implements.10m
  4. BitemporalityThe question audits actually ask, and the reason one timestamp column cannot answer it.9m
  5. Lineage has a ceiling, and hash chains have limitsTwo things commonly sold as answers that answer narrower questions than advertised.9m
  6. The anchor caseEvery failure in this module, in one system, with people imprisoned on the output.8m

Pre-reading

13 sources · ~8h if you read all of it

Ranked, not exhaustive. The lessons stand on their own, so treat the first tier as the genuinely load-bearing sources and the rest as depth when a lesson makes you want it.