The over-optimisation trap

The sleep score that made sleep worse

In 2017 a team of sleep researchers led by Kelly Glazer Baron published a short paper in the Journal of Clinical Sleep Medicine¹ describing a pattern they had started seeing in clinic. People were arriving for treatment of sleep problems they had diagnosed themselves, entirely on the basis of tracker data. They felt fine, or had felt fine, until the device told them they had spent too long in light sleep.

The researchers named it orthosomnia. Ortho, meaning correct or straight, in the same family as orthorexia. What they described was a perfectionistic pursuit of ideal sleep in order to optimise the next day, and the pursuit itself was generating the anxiety that kept people awake.

The detail that stayed with me was this. Baron noted that these clients were believing the device over their clinician, and over their own experience of the night. The estimate had become more real than the evidence.

And it is an estimate. No consumer wearable measures sleep directly. It infers sleep from movement, heart rate, and skin temperature, then dresses the inference up as a stage-by-stage breakdown. They are reasonably good at telling sleep from wake. They are considerably shakier on the stage percentages that people fixate on.

Being told you slept badly is enough to make the day worse

In 2014, Christina Draganich and Kristi Erdal ran an experiment at Colorado College³ that I think about often. They wired students up to a machine and told them it could measure the quality of the previous night's sleep. The machine did nothing at all.

Students were then given a figure. One group was told they had spent 28.7 percent of the night in REM sleep, comfortably above average. The other group was told 16.2 percent, well below. Then both groups sat cognitive tests.

The group told they had slept badly performed significantly worse on attention and arithmetic tasks. The group told they had slept well performed better. And critically, how the students had rated their own sleep before being given the fake number predicted nothing at all.

Think about what that means in practice. A number arrives on your phone at six in the morning telling you that you slept poorly. Your alertness, your mood, and your patience for the day can drop in response to the number itself, regardless of what your body actually did in the night. The score is not just reporting on the day ahead. It can help create it.

And measuring can drain the pleasure out of the thing

There is a second line of research, from a different field entirely. Jordan Etkin, a consumer psychologist at Duke, ran six experiments on what she called the hidden cost of personal quantification.⁴ She had people track ordinary enjoyable activities: walking, reading, colouring.

Measurement worked, in a sense. People who wore a pedometer walked more than those who did not. But they enjoyed the walking less. Drawing attention to the output turned the activity into work. And when the measuring stopped, so did some of the engagement, because the intrinsic reason for doing it had been quietly replaced by an external one.

The finding held even among people who chose to be measured, which is the part that matters for us. Opting in does not protect you. The walk that used to be a walk becomes a step target that you either hit or fail.

What your body has been telling you the whole time

Here is what I actually use, on myself and in the clinic. None of it costs anything. All of it is available to you right now, this morning, without an app.

One honest complication, because it matters for anyone reading this in pain. A 2024 systematic review in The Journal of Pain⁵ found that people living with chronic pain tend to report being highly aware of their internal sensations while actually being less accurate at reading them. More attention, less accuracy, and less trust in the body's signals. So the instruction is not simply to listen harder. If you are in persistent pain, monitoring more closely can be its own trap. The aim is a short, regular, low-drama conversation with the body, and then getting on with the day.

The daily readouts

These are the ones you already have access to several times a day and mostly ignore.

Bowels

Something regular, formed, soft, easy to pass, with no straining and no sense of unfinished business. The commonly used clinical range is three times a day to three times a week,⁶ and roughly 95 to 98 percent of healthy adults sit inside it. Consistency matters more than the number. Someone who goes every second day comfortably is fine. Someone who goes daily with real effort and hard stool is not.

Urine

Pale straw, at regular intervals through the day, without needing to get up repeatedly overnight. A systematic review of urine colour as a hydration marker⁷ found correlations with laboratory measures ranging from moderate to strong. It is not precise, and B vitamins will throw it off entirely, but as a free daily check it holds up. Consistently dark means drink more. Consistently clear all day usually means you are drinking more than you need.

Falling asleep

You get to sleep without a fight, most nights, without needing to negotiate with your own head first. Notice that this is the felt experience of the transition, not a duration. Nobody needs a device to tell them whether getting to sleep was a struggle.

Waking

You wake up feeling reasonably like yourself. Not radiant, not thrilled about the alarm, but restored enough to begin. This one is worth protecting from the score in particular. It is the reading that a tracker most easily overwrites.

Afternoon energy

No hard floor at three o'clock. A gentle dip is human. Needing sugar or caffeine to get to dinner is a signal.

Morning stiffness

Stiff on waking is normal, particularly after forty. It should ease within twenty to thirty minutes of moving. The length of time it takes to feel like yourself is one of the more sensitive readouts you have. When that window stretches out over weeks, something has changed.

The slower readouts

These move over weeks and months rather than days, which is exactly why they are more trustworthy than a nightly number.

Appetite

Hunger that arrives on its own, rather than by clock, boredom, or mood. And a sense of enough that you can actually feel.

Patience

Whether small frustrations stay small. Temper is a nervous system readout far more than it is a character trait, and it is usually the first thing to go when a system is under load.

Breath at rest

Through the nose, low in the body, quiet, without frequent sighing or yawning to catch up. Sit still for thirty seconds and just notice. If the breath is high in the chest and you are pulling for it while doing nothing, that is information you can act on today.

Getting off the floor

Down to the floor and back up without planning the route first, and without needing both hands. Claudio Gil Araujo's sitting-rising test⁸ followed around two thousand adults aged fifty-one to eighty. Each single point of improvement in the score was associated with a meaningful survival advantage. It is a composite of leg strength, hip mobility, and balance, all in one movement. It is a marker, not a verdict, and it improves with practice at any age.

Standing on one leg

Ten seconds, barefoot, near a wall or bench, eyes open. Also from Araujo's group, published in the British Journal of Sports Medicine in 2022.⁹ Do it with support within reach, and treat a wobble as a thing to train rather than a thing to worry about.

Walking and talking

You can hold a conversation while walking at your normal pace, on a slight rise, without breaking sentences to breathe.

When a number does earn its place

I am not asking anyone to throw a device in the bin. Measurement is a tool, and like any tool it works when there is a job for it. The test I use is simple.

Is there a question attached?

Measuring to answer something specific is useful. Does coffee after two in the afternoon change how long it takes me to fall asleep? Does a walk after dinner change how I feel the next morning? Am I actually getting stronger on this lift, or does it just feel harder? Those are real questions and a record answers them.

Is there an end date?

This is the one most people skip. Run the experiment for two or three weeks, get your answer, then stop. Open-ended measurement with no question is not information gathering. It is surveillance, and you are both the guard and the prisoner.

Does the number make you do something, or feel something?

A number you act on is working. A number you have a feeling about is working on you. If your first response to the score is relief or dread, the device has stopped being an instrument and become an examiner.

What does not belong in a self-check

Everything above is self care. It is not diagnosis, and there is a category of signal that should never be filed under a Sunday check-in.

Five minutes, once a week, then put it down

The whole point is that this takes almost no time and does not follow you through the week.

There is a version of health that is a project, permanently under review, never quite finished. And there is a version that is closer to a working relationship with a body you happen to live in, where you check in, notice what needs attention, and get on with the week.

The second one is quieter. It also tends to last longer, because nothing about it requires a battery, a subscription, or your enthusiasm.