The lag
What a six-month pilot can see
Run this transition as a standard six-month trial and the report is positive, accurate, and silent about three of the five problems.
The standing slot for work with no ownerThe fastest of the five problems
week 5 · surfaced after 14 weeks
Reading supplier contracts before renewal
week 2 · surfaced after 22 weeks
The quarterly reconciliation of the project ledger
week 3 · surfaced after 31 weeks
Shadowing new starters through their first week
week 6 · surfaced after 47 weeks
Logging why a job was turned downThe slowest on the record
week 9 · surfaced after 64 weeks
Surfaced 5
Five removals were written down with the week each of them stopped. Every one of them produced an event inside the 78 weeks watched. Of these, five surfaced as a problem. All five removals whose absence surfaced as a problem, in the order they surfaced.
Almost every published account of a four-day week reports success. The unanimity is striking and the usual explanations for it are unsatisfying: that the organisations self-selected, that the measurement was soft, that everybody wanted it to work.
All three may be true. None of them is needed. This record offers a simpler account, which is that the trials are measuring an interval shorter than the thing they are measuring, and that this alone is sufficient to produce a positive result from a transition with four unresolved problems in it.
The exercise
Take this record. Run it as a six-month pilot — twenty-six weeks from the first day of the transition — and report at the end.
Counting in absolute weeks from the start rather than in lag, the five problems surfaced in weeks nineteen, twenty-four, thirty-four, fifty-three and seventy-three.
Two inside the window. Three outside it.
What the report says at week twenty-six
| what had happened by week 26 | n |
|---|---|
| asked back | 8 |
| crept back | 9 |
| surfaced as a problem | 2 |
| nothing at all | 19 |
Nineteen of thirty-eight entries had produced no event whatever. Seventeen things had come back, most of them smaller. Two problems had arisen, both had been dealt with, and both had obvious fixes.
The report writes itself, and every sentence in it is true:
Thirty-eight procedures were removed. Seventeen were restored in some form, most at reduced cost. Two gaps were identified and closed. Half the removals have produced no consequence of any kind. Output and service levels were maintained throughout.
Nobody has shaded anything. The pilot was run honestly, measured properly and reported accurately.
What the report cannot say
Three of the five problems on this record had not yet happened. More than that: at week twenty-six, the three future problems were sitting in the report as part of the nineteen entries that had produced nothing.
They were, in other words, counted on the positive side of the ledger. The sentence "half the removals have produced no consequence of any kind" included the shadowing of new starters, the log of refused jobs and the reconciliation — which between them went on to cost the organisation three resignations, four duplicated pieces of work and a scramble through an audit.
This is the mechanism, and it is worse than silence. A removal whose consequence has not yet arrived does not appear as unknown. It appears as evidence of success.
Why this would happen to anybody
There is nothing peculiar about this organisation's distribution of lags. The things with long lags were the things whose absence is noticed by somebody standing in a place they only stand in occasionally — an auditor, a new starter's manager a year on, a contract renewal.
Every organisation has such places. A pilot of any length will catch the removals whose observers are frequent and miss the ones whose observers are rare, and the second group is systematically the one with the expensive failures, because rare observation and slow discovery are the same property seen from two sides.
So the published accounts are not wrong and their authors are not careless. They are reporting the head of a distribution and calling it the distribution, which is what any honest six-month report of an eighteen-month process will do.
What an honest report looks like
It is no longer than the usual one and no more hedged. It carries one extra paragraph:
- State the window: this covers twenty-six weeks.
- State what is known to sit outside it: on this record, three of five problems arrived later than that.
- Name who will look again and when.
- List the entries that have produced nothing, by name, and say that these are the ones about which nothing is yet known.
The fourth point is the one that does the work, and it costs a page. It converts nineteen silent entries from evidence of success into a list of open questions, which is what they are.
None of this argues against shorter weeks. Two problems in twenty-six weeks against seventeen cheap restorations and nineteen clean removals is a good result, and so is the eighteen-month picture. The argument is only that the six-month report should not be read as the eighteen-month one, and that nothing in the six-month report itself tells you which you are holding.
The same exercise at twelve months
Running the record to week fifty-two rather than twenty-six changes the picture in one direction only.
| week 26 | week 52 | |
|---|---|---|
| asked back | 8 | 8 |
| crept back | 9 | 10 |
| surfaced | 2 | 3 |
| nothing yet | 19 | 17 |
Twelve months buys one further problem and one further quiet return. It does not buy the two slowest failures, which arrive at weeks fifty-three and seventy-three.
That flatness after week twenty-six is the practically important feature and it is easy to misread as reassurance. The curve going flat does not mean the process is finished; it means the fast channel has stopped and the slow one reports too rarely to produce a visible rate.
A reviewer at twelve months sees almost the same picture as one at six, plus one incident, and is correspondingly more confident. The confidence is not better founded.