← the guide

How to read the water

every age · Updated 2026-08-25

The home screen is a tide. The waterline is your child's sleep pressure — the real, physical drive to sleep — rising while they are awake, draining while they sleep, drawn against their own personal threshold: the level at which they, historically, fall asleep easily. The percentage riding the "now" line says how close the water is to that shore. 100% means "sleep would come fast now". It never means "they must sleep now". Nothing on this screen is a deadline; here is how to read it.

The two ends, which the number does tell you

The middle, which it does not

Between those two ends the percentage stops being a good guide to the phase of the day, and the app no longer pretends otherwise.

The reason is in the physics. Pressure rises like a capacitor charging — fast at first, then flattening as it nears the ceiling — and the time constant grows with age. At two years the water crosses 70% barely two hours into the day and then needs nearly three more hours to climb the last 30%. So a band of the percentage that means "soon" for a four-month-old can cover the entire middle of a toddler's afternoon.

The app used to key its words to those bands. It printed getting sleepy over a child who visibly wasn't, for roughly half of every wake window, and a status line that is true half the time carries no information at all. Since 25 August 2026 the words come from time to the window, not from the percentage — wind down inside about 45 minutes, ready inside about ten — while the percentage stays on screen as the honest thing it always was: a pressure reading, not a mood. The full argument is in the notebook.

So: read the two ends off the number. Read the middle off the forecast time, and off your child.

The bands are still there — on the chart, where they are honest

They didn't deserve deleting, only relocating. Pull up THE DAY and you'll find them drawn against the curve: winding down, the wave, and past the wave, as three soft ribbons around the shore your child's pressure is climbing toward.

Two things make them truthful there and not in a status line. A chart has a time axis, so a band that will take three hours to cross looks like three hours — the exact fact the printed word hid. And on a chart the bands can bend, which they must, because the shore itself moves: the circadian threshold swings about eleven points across the day, which is wider than the wave band itself. Watch the ribbons rise through the evening and dip in the early afternoon. That bend is the body clock, drawn to scale.

Two zones are deliberately left unmarked. Low water has no band because the lower threshold is already on the chart as its own line. And the long ordinary middle of the day has none because it is not a state that needs naming — marking it was half of what made the old ladder lie.

Three things the number is not

It is not a battery. Pressure builds fast and then decelerates, so 50% is not halfway in time — for my own child it was roughly a third of the way to the wave. And a nap doesn't drain the water to zero: a fair afternoon nap from a high line lands around 25–30%, so afternoons start with a head start. Only a full night empties it.

It is not a countdown. The most honest translation of the % is how long would falling asleep take if we tried now — at low water even a perfectly executed nap attempt means a long latency or refusal; on the wave it means minutes. (This latency link is measured, not folklore: in 2–5-year-olds, sleep latency runs about 24 minutes at low pressure and saturates near 13 after ~7 h awake.)

It is not still. Pull up THE DAY and the figure runs to three decimals, live — that is not an animation, it is the same physics the engine uses, integrated between syncs. Awake, it climbs; asleep, it falls about four times faster, because pressure drains on a roughly one-hour time constant and builds on a much longer one. If you ever want to see the model actually working, that is where to look.

It is not a verdict. Your child's cues outrank it. The water tells you what the biology has had time to build; your eyes tell you what today did with it.

Past, future, and the blur

The line is sharp where the day already happened and softens as it runs ahead, because the uncertainty is about when the next transition lands, not what the physics does after it. A running nap keeps a sharp shape; it is the wake at its end that smears. If the future ever looks hazy to you, that is the app being honest, not broken. The compact strip below the water shows the whole day in the same terms — solid for sleep that happened, hollow for sleep merely expected.

One distinction worth knowing, in the details pane: prescribe is where the raw physics says the water reaches the shore — where to steer. Predict is where your family's actual habits say sleep will really start — what is likely. The gap between them is information, not error: it is how far your logged routine has drifted from the biology.

What to actually do with it

Glance, see where the water stands, close the app. That's the intended use — five seconds. Believe the window, not the minute: the best prediction error I can honestly achieve is 17–21 minutes (the full scorecard, losses included), and the design assumes exactly that softness. If the water and your child disagree, the child is right — and logging what actually happened is how the tide gets less wrong tomorrow.

Where this comes from