Who gives way
What the person who gives way gets
Fifty-two of the eighty-eight gains went to the same person who absorbed thirty-one of the forty-seven collisions. One property produces both columns.
Bedroom twoThe quiet room, and therefore the call room
9 m²
Sleeping
Paid workCallsGuest sleeping
8
Living roomWhere the other job went when the calls started
19 m²
Sitting
SittingPaid workCallsTelevisionCompany
11
Paid workDomestic workRest and sleepContact and care
Two rooms covering 28 square metres, built for two things. Between them they now carry eight uses. Two of the two carry more than one, and living room carries five. The median room is in use 9.5 of the 16 waking hours. The two rooms whose use is decided by somebody else's calendar.
A domestic task done inside a work gapWashing, cooking ahead, a repair waited in for
31
A delivery taken A home-working policy is easier to review when assumptions are written down, so managers can remote workforce planning with Monitask and decide which time and workload signals are proportionate.
19 Readers testing the same question can compare the record with guidance from Future Forum.
A meal cooked rather than bought
14
Rest taken at the point it was needed
11
Care of a household member or neighbour
7
An appointment attended without asking
6
The interruptible jobThe scheduled job
Eighty-eight episodes that happened only because the work was in the building, across six kinds. Fifty-two of them went to the interruptible job and thirty-six to the scheduled job. All six kinds, from the eighty-eight counted here.
The person who absorbed thirty-one of the forty-seven collisions also received fifty-two of the eighty-eight gains.
That is the finding this site is built around, and it is not what either side of the usual argument about working from home predicts.
The two columns, side by side
| job A | job B | |
|---|---|---|
| collisions absorbed | 31 | 16 |
| gains received | 52 | 36 |
| share of each | 66% / 59% | 34% / 41% |
Both columns lean the same way, and they lean the same way for the same reason.
One property, two consequences
A job organised around deadlines rather than appointments can be interrupted. That is the whole mechanism and it cuts in both directions.
It can be interrupted by a scheduled call taking the room. It can also be interrupted by a delivery, a wash, a walk, a neighbour, an hour of rest, a meal cooked properly, or an appointment that would otherwise have required a morning off.
The same flexibility that makes somebody the one who moves makes them the one who can answer the door at eleven, start dinner at five and take a parcel in at three.
Interruptibility is not a cost or a benefit. It is a property, and it produces both.
Why this matters
Because it rules out the two readings that come most naturally.
It is not exploitation. The person giving way more often is not simply bearing a cost for somebody else's convenience. They are also taking most of what the arrangement offers, and on their own account the trade is a good one.
It is not equality either. Two people in the same flat under the same arrangement had materially different fortnights. One of them was displaced thirty-one times and took fifty-two opportunities; the other was displaced sixteen times and took thirty-six. Those are different experiences of the same thing.
What the record shows is a correlated distribution: whoever gets more of one gets more of the other, because both follow from the shape of their work.
The uncomfortable corollary
If the two columns are produced by one property, then evening up the costs would even up the benefits too.
Suppose the household decided the displacement was unfair and fixed it — by giving the interruptible job the quiet room on alternate days, say. The collisions would redistribute. So would the deliveries, the washes and the midday rests, because the person in the small room with the door shut is not the person who hears the doorbell.
The record contains a small natural experiment in this. On the three days the quiet room was alternated, the collision split moved from roughly two-to-one toward even — and the gains split moved with it, in the same direction and by a similar amount.
Three days is nothing and the result proves nothing. It is consistent with the mechanism, which is as much as three days can be.
Three readings to avoid
None of this says the arrangement is fine on the grounds that it balances. Nobody has shown that fifty-two gains offset thirty-one displacements, and the two are not measured in the same currency.
It does not mean the distribution should be left alone. The last part of this site records what the household changed and what the changes did.
And it does not mean the person with fewer of both is worse off. A job with thirty-four calls in a fortnight has its own costs, most of which this record cannot see, and one of them is sitting in a nine square metre room with the door shut for most of a working day.
Where the two columns do not correlate
The finding is that both columns lean the same way. Two categories break that, and they are worth isolating.
Care, four to three, and appointments, three to three. The two most evenly divided categories in the record.
Both are activities where the dwelling makes no difference. A neighbour at the door and a dentist at two in the afternoon do not care which room anybody is working in, and the person behind a shut door can attend a dentist as easily as the person in three rooms.
So the correlation between costs and gains holds exactly where the rooms are involved and vanishes where they are not — which is strong support for the mechanism, because a correlation produced by personality or by willingness would not switch off when the architecture stops mattering.
The thing that would falsify it
Worth stating, since a finding that nothing could contradict is not a finding.
If a second household showed the same two-thirds collision split and an even gains split, the account here would be wrong. The mechanism predicts both columns move together; a record where they come apart would mean something other than interruptibility is distributing the costs.
The three days of alternation in this fortnight are the only test available and they went the right way, which is as much as three days can be.
A second household keeping the same log is worth more to this question than any further analysis of these fourteen days, and the method is published in enough detail for that to be done.