> "Our documentation is out of date and nobody knows which parts are still true." ^pain A collective decay pain. The artefacts exist and can be found, but readers cannot tell whether they still reflect reality. Each failed-trust event pushes the reader back to asking a person, which makes the artefact even less referenced, which means even fewer people maintain it — the negative-feedback loop that turns a wiki into a graveyard while everyone still has read-access to it. ## Discovery questions - "When was the last time someone on your team updated documentation without being prompted by an incident?" - "If a new hire reads your team's wiki page on X, what's the chance they're getting today's truth versus last year's?" - "How do people on your team decide whether a doc is current enough to act on?" ^discovery-questions ## Examples - Google's postmortem of the 2013 Gmail outage notes that some internal documentation and runbooks were outdated, contributing to delayed mitigation.[^1] - The 2017 AWS S3 outage analysis describes how operators followed an outdated internal procedure that removed more servers than intended.[^2] - Atlassian's incident handbook describes an internal outage where stale runbooks failed to reflect the current architecture, slowing recovery.[^3] - NHS England serious-incident report on a patient death highlighted outdated clinical protocols and guidance as a contributory factor.[^4] [^1]: https://sre.google/sre-book/postmortem-culture [^2]: https://aws.amazon.com/message/41926 [^3]: https://www.atlassian.com/incident-management/postmortem [^4]: https://www.england.nhs.uk/patient-safety/serious-incident-framework