Collisions
The collisions nobody noticed
Nine of the forty-seven were welcome. They are counted because the test is whether an activity gave way, not whether anybody minded.
Kitchen and dinerTwenty-one of the forty-seven happened here
16 m²
Cooking and eating
CookingEatingPaid workLaundry and adminCallsCompany
14.5
Living roomFourteen more
19 m²
Sitting
SittingPaid workCallsTelevisionCompany
11
Paid workDomestic workRest and sleepContact and care
Two rooms covering 35 square metres, built for two things. Between them they now carry eleven uses. Two of the two carry more than one, and kitchen and diner carries six. The median room is in use 12.75 of the 16 waking hours. The two rooms that between them carry thirty-five of the forty-seven collisions.
Nine of the forty-seven collisions involved somebody coming into a room for company, and the work giving way.
Asked about them afterwards, both people in the household described those episodes as pleasant. Nobody experienced any of the nine as an interruption. When paid work competes with domestic use for the same room, teams need a record of hours and projects as well as a record of space; they can see how the vendor frames the workflow before choosing what evidence belongs in a trial.
They are in the count anyway, and that decision is worth defending or abandoning in the open, because it changes the headline number by nearly twenty per cent. For wider context beyond this household record, consult material published by Zoom Blog.
The case for counting them
The test applied to every entry in the log is behavioural: did an activity stop and leave a room because another activity wanted it?
For all nine, the answer is yes. Work stopped. The laptop closed or was pushed away. The work did not resume for a median of twenty-five minutes.
Applying a different test to these nine than to the other thirty-eight — counting the ones that felt bad and not the ones that felt fine — would make the number a measure of how the fortnight was experienced rather than of what happened in the rooms. Those are different quantities and the second one is the one this record is for.
A log that only counts the interruptions somebody minded is a mood diary with a floor plan attached.
The case against
It is not weak.
If nine of forty-seven were welcome, then the figure "forty-seven collisions" overstates the friction in this flat by nine. A reader who takes the number as a measure of strain is being misled by about a fifth, and the site's own framing — "collisions", a word that implies damage — does the misleading.
There is also a selection worry. The nine are all of one kind: company displacing work. If a category of displacement is systematically welcome, counting it alongside the others blurs a real distinction.
What was done about it
The nine stay in the count and are flagged everywhere the count appears, including here.
Two other options were considered and rejected.
Removing them would mean deciding which collisions were welcome, and that decision was made eighteen days after the fortnight ended, by people with a view about what the log should show. Every entry would then carry a judgement made in hindsight.
A separate category was nearly adopted and failed on a practical test: of the nine, three were described differently by the two people involved. One person's welcome interruption was the other person's interruption. A category requiring agreement between two accounts of the same event does not survive contact with the data.
What it changes
If you accept the nine, the flat had forty-seven collisions in a fortnight and thirty-one fell on one person.
If you reject them, it had thirty-eight, and — because seven of the nine fell on the same person — twenty-four of thirty-eight fell on them. The proportion barely moves: sixty-six per cent against sixty-three.
That is the useful thing about the disagreement. The headline number is sensitive to it and the finding is not. The distribution between two people holds under either accounting, which is a reason to trust the finding more than the total.
Where else this applies
The same problem exists in reverse in the eighty-eight gains. Some of those were trivial — a parcel taken that would have been redelivered without difficulty — and counting them alongside an appointment attended without taking leave flattens a real difference.
The log makes no attempt to weight either column. Both are counts of events that met a stated test, both include members that a reasonable person would discount, and the site's position is that publishing the test and the raw count is more useful than publishing a weighted number whose weights nobody can check.
The three that were described differently
Of the nine welcome collisions, three were recorded differently by the two people, and those three are the reason a separate category was abandoned.
In each case one person described coming in for company and the other described being interrupted. Neither account is wrong; they are accounts of different halves of the same event.
A category requiring agreement cannot survive that. A category assigned by the person who moved would make the count depend on which of two people is keeping the sheet that day. A category assigned by the person who arrived would record every interruption as welcome, since nobody arrives intending to interrupt.
So the nine stay in the main count with a note, which is the only option that does not require choosing whose account of a shared evening is the real one.
The symmetrical problem in the other column
Six of the eighty-eight gains have the same structure and are flagged here rather than left for the reader to find.
In each, one person logged an episode as a gain — a domestic task done at eleven, a delivery taken — and the other person's account of the same moment is that they were working and somebody got up.
A gain for the household is an interruption for the person who took it. Both are true, both were recorded, and the log counts the event once in the gains column because that is where the person who did it put it.
A reader who thinks six of the eighty-eight should be in the other column has a defensible position, and the arithmetic is: eighty-two against fifty-three, which is the same conclusion with less margin.
Also on what the rooms are being asked to do