Other shapes
Where each shape hides its lag
Every arrangement produces slow consequences. What differs is whether they are attached to anything that can be found afterwards.
The Friday clean-down of the workshopThe removal a compressed week cannot make
week 9 · crept back after 22 weeks
A named person on the phone until sixThe removal staggered cover is designed to avoid
week 7 · crept back after 5 weeks
Visits to the two furthest sitesThe removal a nine-day fortnight makes instead
week 7 · asked back after 12 weeks
Crept back 2Asked back 1
Three 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, one was asked back and two crept back. Three removals that a different shape of week would have handled differently.
The five costly removals on this record were slow because nobody was positioned to notice. That property belongs to the removal rather than to the arrangement, which means every shorter-week arrangement has some version of it.
What differs between the arrangements is not whether slow consequences occur. It is whether they leave a trace that can be followed back.
Four arrangements, four kinds of trace
Four-day week. The best trace of the four, for an unflattering reason: it is the most disruptive. A whole day disappears, everybody notices, a list gets written because somebody has to coordinate the thing. The removals are attached to a specific event in a specific quarter.
Even the seven unproposed removals on this record were recoverable, five of them within a month, because there was a question that worked: what did you used to do on the day that went?
Nine-day fortnight. A weaker trace. Half the hours, so about half the removals, and no single day that went. Things that lived in slack still stop; nothing attaches them to an identifiable gap. The retrospective question does not work, so the removals have to be caught as they are made or not at all.
Compressed hours. No removals of this kind, so no trace needed. The costs are of a different sort and show up in who can take the arrangement and how the longer days are tolerated — both of which are visible immediately rather than in a year.
Annualised hours. The weakest trace of the four. Procedures stop by deferral rather than by decision, nothing is announced, nothing is listed, and the expectation that they will resume suppresses the attention they would otherwise attract.
The ranking is backwards
Put those in order of how much is removed and you get: four-day week, nine-day fortnight, annualised hours, compressed hours.
Put them in order of how well you can find out afterwards and you get the same order.
The arrangement that removes most is the one with the best record of what it removed, and the arrangement that removes least by decision is the one with the worst record of what it dropped anyway.
Disruption is what produces documentation. The gentler arrangements are gentler partly because nobody has to write anything down, and that is also why they are harder to audit.
What this implies
Not that a four-day week is safer. It removes four to eight times as much and it had five costly failures on this record; a nine-day fortnight attempting half the reduction would be expected to have fewer.
What it implies is about where to spend the effort, and the answer inverts.
- Under a four-day week, the list writes itself and the useful effort goes into the read-backs — the twelve and eighteen-month passes, which nobody does.
- Under a nine-day fortnight, the list does not write itself, and the useful effort goes into capturing removals as they are made, because no retrospective question will recover them.
- Under annualised hours, the list is of deferrals rather than removals, nobody has ever written one, and it is the only artefact that would make the arrangement auditable at all.
- Under compressed hours, none of this applies, and the effort belongs somewhere else entirely: on who is excluded by the longer day and on whether the exclusion is being noticed.
The common element
All four need the same thing, which is a written record of what stopped, made at the time, by somebody who will not be reassigned the moment the change is finished.
The arrangements differ only in how hard that record is to produce and how easy it is to pretend it is unnecessary. The four-day week makes it nearly unavoidable. The others make it optional, and on the evidence of this record an optional list does not get written and an unwritten list cannot be read back.
The trace that none of them produces
There is one category that no arrangement captures and that every account of shorter weeks is silent about: the work that moved rather than stopping.
On this record four things happened quietly on the day off, by people who saw a message and could not leave it. None was logged. Several procedures were performed slightly faster rather than being removed. Somewhere between one and three hours a week per person is unaccounted for by any entry on the list.
That unaccounted hour is a removal with no name. It is not on the list, so no read-back reaches it, no question recovers it and no arrangement makes it visible — and it is probably the largest single component of how the hours were actually found.
The honest position is that every arrangement described here leaks in the same place, that the leak is invisible to the method this site recommends, and that nothing in this record measures it. A list of thirty-eight removals accounts for most of a reduction and not all of it, and the part it does not account for is the part where the remaining work quietly got denser.