How the reading works

The headline is two splits

One on each axis, because they are different questions and they usually disagree.

The doing — the time and effort of carrying a task out. The carrying — everything around it: noticing it is coming, working out the options, deciding, and checking it got done. Each axis carries three marks: an even split, what your hours suggest, and what happens.

The even split is a reference mark, not a target, and the app never says it is one. It is there because a reading with no fixed point on it is hard to read. What your hours suggest is derived from your own answers about paid work, commuting and other obligations, rather than handed down.

The gap is the reading

The two numbers that matter are the distances: between what your hours suggest and what happens, on each axis. A split on its own invites an argument about the split. A gap invites a question about what would close it — which is the only useful question in the room.

Expect the gap on the carrying to be the larger of the two, and expect the contrast between the axes to be the more interesting finding. I'll do it, just tell me what needs doing is an even split of the doing sitting on a one-sided split of the carrying, and it is the single most common flashpoint in these households.

Nothing is shown before it can be shown honestly

Two rules that sound like edge cases and are not:

• A task nobody has spoken for does not count. An untouched row is drawn faint and outlined, at no split at all, and it stays out of the reading. A greyed-out even bar would show balance where the truth is silence.
• Before anything is known about hours, there is no mark for what they suggest. The app shows what happens and says plainly that the other half is missing. It does not put an even split in the missing mark's place, and it does not nag for the answer.

What the app will never print

A points total. Not yours, not your partner's, not a category's. The model computes in points internally and every user-facing surface is barred from quoting one.

This is a deliberate trade and it costs something real: you cannot see how your own week adds up, which is a fair thing to want. The reason is that the total would be a number the instrument cannot stand behind — what a point is, and what it is not has the whole argument, including what is lost by withholding it. Shares, differences between shares, and counts are what the app is allowed to say.

Moves, when there is something worth moving

Where a split sits well away from its reference, the reading offers up to three moves: rows that could change hands, and what the split would read if they did, as shares — the carrying goes from 77–23 to 69–31. They are arithmetic, not advice: every candidate is scored by how much of the gap it closes against what it would cost the person taking it on, the ones that land in a window when that person is not there are dropped, and so are the ones they dread and you don't. The set is chosen one move at a time, each against the split the earlier ones leave.

The carrying's moves do not wait for anybody's hours, because hours barely change what either of you is expected to carry; until hours are in, they are worked out against the even split, which the reading marks as a reference and never as a goal. The doing's moves wait for both of your hours, because there the hours are what the reference is made of.

When an axis is already close to its reference, the app offers nothing on it and says so in a line — the carrying is close to even — never that it is done, or good. A set of moves too small to change the printed split by a few percentage points is not shown either: a difference that size is as likely to be about which rows happen to be filled in as about the household.

The most useful of these is usually the one about time. A large share of what one person carries is time-flexible — it does not need anybody home before half six. That reframes an argument about physical presence, which nobody can win, into one about which work moves, which is solvable.

What it is not

Not a chore app: no to-do list, no assignments, no mark as done, no reminders pointed at one person. Not a shared calendar. Not therapy, and it makes no clinical claim about anybody's health. Not a scoreboard, and there is no version of it where one partner scores the other.