an audit is finished only when its findings have an order

An interrogation of a system's own documentation returned eighty-seven questions, every one answered, and that is where most exercises of the kind end: a document of findings, correct, complete, unread. Findings are not work. A list of them is an archive with a short half-life, because nothing about it says what to do first.

The interesting part is not the answering but the disposition. The findings were consolidated into a single plan organised by commitment horizon: being done now, next, later, far, and a fifth category for what nobody has taken at all. That last one is the honest one. Work without an owner is real, and the choice is between a category for it and a quiet deletion.

Ordering by horizon rather than by subject looks trivial and is the actual decision. A subject-ordered list tells a reader what the findings are about, which they already know, because they did the audit. A horizon-ordered list tells them what to do next, which is the only question that keeps a plan alive. Each phase carried a scope and a status, and finished work was kept as ticked lines rather than removed, so the document accumulates a record of what was completed instead of only a list of what is left.

The gaps found in components with no tests were also marked where those components are, so the absence is visible at the point of use rather than only in the plan.

The cost is a document that goes stale the moment priorities move, which they do. What it proves is that the team can end an investigation by converting it into a queue rather than publishing a report.