Module 06
Re-measurement
Run the baseline again under your new manual, compare honestly, and write the verdict.
A framework that exempts itself from evidence is not one.
Movement III — Check whether it held. Choose measurements that resist gaming, then run the baseline again under the new manual and write an honest verdict — including that nothing improved.
The comparison, and what will confound it
Repeat module 00 exactly: one ordinary week, the same four things recorded, the same three time totals. Then compare against the baseline using the measurements from module 05 and their pre-written failure conditions.
Two things will confound the result, and naming them is part of the artifact. You are now practised at observing yourself, which improves the record independently of the practice. And the weeks differ — different work, different pressure, different interruptions. Neither invalidates the comparison; both mean the honest verdict is directional rather than precise. Say so rather than implying a precision you did not obtain.
Verdicts that are allowed
No change is a legitimate result and a common one — a redesign of a practice takes longer than a week to show. Worse is legitimate and informative: a boundary drawn too tight, or a review protocol too heavy to sustain. Better is legitimate, and requires the most scepticism, because it is the answer you were hoping for.
The artifact must include what you are reverting. A manual that survives contact entirely was not specific enough to be wrong anywhere, which means it did not commit to much.
Method
- 01Run one ordinary week under your manual, recording exactly as in module 00.
- 02Produce the same three time totals: done, supervised, repaired.
- 03Compare against your module 05 measurements and their failure conditions.
- 04Write the confounds plainly, including the observer effect and how the two weeks differed.
- 05State the verdict, and list what you are reverting and what you are keeping.
How this goes wrong
Reading improvement into noise
Two weeks is a small sample and you are invested in the outcome. If the difference is within the range the two weeks would have differed anyway, the honest verdict is no change.
Moving the baseline
Adjusting what you count between the two weeks makes the comparison meaningless. If a measurement has to change, report both the old and new figure.
Reverting nothing
A manual with no reverted parts usually means the changes were too cautious to test anything, or the verdict was written to protect the work.
The exercise
- Re-run the baseline for one full week under your manual.
- Compare against module 05's measurements and failure conditions.
- Write the verdict, the confounds, and what you are reverting.
Verdict
The sections your artifact contains:
- The week, and how it differed from the baseline week
- The three totals, side by side
- Each measurement against its failure condition
- Confounds, including the observer effect
- Verdict — what changed, what did not
- Reverting / keeping
Done means. A verdict that could have gone the other way. Concluding that AOS did not help you is a completed artifact, not a failed one.