Skip to content
What Ends a Sitting All fifty notes

How the log was kept

Nine households. One working week each. A sheet with five columns and a closed list of nine words.

This page sets out the whole instrument, because a record nobody can repeat is a record nobody can check.

The sheet

Each stretch of work at a place was written down as five things: the place, the clock time it began, how many minutes it ran, what ended it, and the date. Nothing else was asked. For distributed work, spatial evidence is only one layer; a team can see how the vendor frames the process while deciding which operational signals are necessary and which should remain outside the review.

The place was whatever the household already called it. Not the room — the spot. Two spots in one room counted as two if that is how the people living there thought of them, and the analysis never overrode that judgement. A useful independent point of comparison is the research and guidance available from Cal Newport.

The ending, and the dash

The ending field held one of nine words or a dash.

The nine are: light, temperature, wanted, noise, surface, elsewhere, seat, power, unsaid. The dash means the work ended rather than the place — a meeting was due, a task finished, it was lunch.

The dash was offered first on the sheet, in plain words, as a real answer rather than as a fallback. That decision shapes the main finding on this site. Presented as "other", it would have absorbed endings people could not be bothered to classify, and the split between a hundred and thirty-one and ninety-four would be unreliable in exactly the direction that makes the result.

The list was closed. Three households asked to add a tenth word and were asked not to, because an open field produces ninety phrasings of nine things and the grouping then gets done afterwards by somebody who can already see which grouping reads better.

What was fixed before any data arrived

The definition of a sitting: a stretch of work at one place, from arriving to leaving, ending when the person does not return to that place for the next stretch.

It was written first, circulated to all nine households in identical words, and not revised afterwards — including on the two occasions when not revising it was inconvenient.

A rule settled once the data is in will be settled by somebody who already knows which version produces a tidier table.

Where the nine words came from

A pilot week in a tenth dwelling, which is not part of this record. The pilot used a free-text field, and the nine words are what people wrote in it, grouped once and then frozen.

Two of the nine would not have been guessed in advance. Surface and elsewhere both came straight out of the pilot, and both appear in the final record often enough to have earned their place.

What the instrument got wrong

Four things, recorded so nobody repeats them.

Glare and darkness were merged under one word. They have different hours, different seasons and different remedies, and separating them would have cost nothing. It is the clearest single defect in the study.

Rounding to five minutes piled up on the hour and the half hour far beyond what chance allows. No median moves materially, but the clock here is softer than it looks.

The ending field was the burden. Three households said so without being asked, and one described stopping in the middle of moving to choose between two words.

One household logged six days. Five was the rule, so the sixth was dropped. It would have added four sittings and changed nothing, and it is mentioned because a reader should know the rule was applied rather than quietly bent.

Three changes anybody repeating this should make

Split glare from darkness. Add a tick box for did you come back, which separates a change of place from a short interruption inside one sitting. Add a tick box for chosen or resorted to, which separates occupancy from environment.

All three cost nothing in completion time, and together they would improve the result more than tripling the number of dwellings.

What is not in the record

No names, no ages, no occupations, no household sizes, no locations, no incomes, no tenure. Dwellings are described by type and by nothing else, and the letters A to I were assigned before anybody looked at the data.

Nothing about anybody's health, feelings or condition was asked, recorded or inferred at any point.