Some weeks are not comparable to other weeks.

The failure mode of every tracker is the same: it keeps measuring you against your ordinary self during a period that is not ordinary, and the chart becomes a record of a hard month presented as a decline.

In a relationship product that is worse than useless. The weeks a couple most needs to look back on kindly are exactly the weeks a naive average renders as failure.

What it does

A life context is a named stretch of time with a start and, eventually, an end. While it is running, it eases the target your week is read against — the bar, not the arithmetic.

It names itself on the dashboard, so the softer reading is never a mystery.

What it does not do, on purpose

It changes no score. Not one point, on any contribution, ever.

Two reasons, and the second is the real one.

The first is mechanical: weighting the formula during a context would make that week incomparable to every other week in your history. The frozen target stays frozen and the easing happens when the week is read, so the record stays one thing throughout.

The second is that a coefficient on somebody’s effort during a hard stretch is a judgement about whose effort counts more, written into the one part of this product that is supposed to be mechanical. Deciding that the person doing the night feeds should have their contributions multiplied — or that the one at work should — is precisely the argument the couple is having. An app does not get a vote in it.

It is asked against the week’s own dates

A context that started in March leaves February exactly as you reviewed it. A baby arriving on the Wednesday is the reason that whole week went the way it did, so the week is asked by overlap rather than by whether the context had started on the Monday.

Your history does not get rewritten when you name something, and it does not get rewritten when you end it either.

One at a time

A couple has one running context, not a stack of them. Two overlapping explanations for a week is a way of explaining every week, and a target that is always eased is not a target.

Naming a context is a paid feature. Reading one and ending one are not — a couple whose plan lapsed still sees why their January read the way it did, and can still close the stretch when it is over.

Nothing chases you to close it

There is no notification asking whether the hard thing is over yet. When it is, you end it, and the weeks after it read normally again.

If the numbers are the wrong tool for this stretch entirely, quiet mode puts them away without you losing a thing.