Presence Over Optimization: Give Every Metric an Expiry Date

There is a particular moment I have caught myself in more than once. I am somewhere worth being — a walk at dusk, the light doing something — and a part of my attention has quietly detached and gone off to do arithmetic. How much further to close the ring. Whether this counts as the workout or just as walking. What the number will be.

Nothing has gone wrong, exactly. I am still walking. But some fraction of me is no longer on the walk; it is upstairs in an office somewhere, keeping books on the walk. And the strange part is that the books are the reason I went.

I do not think tracking is bad. I want to say that early, because the easy version of this argument is a smug one, and I have no interest in telling anyone to throw out their scale. Measurement has taught me things about myself I would never have guessed and would have actively denied. What I have come to think is narrower and, I hope, more useful: measurement is not free. It costs attention, and attention is the same currency the experience itself is paid in. Most of the time the trade is worth it. Sometimes you keep paying long after you have bought the thing.

The measurement is made of the same stuff as the experience

Physicists have a version of this problem and so do software engineers. You cannot profile a program without the profiler consuming some of the machine's time. The instrumentation is part of the workload. In a well-built system the overhead is small enough to ignore, and in a badly built one your profiler is measuring, mostly, itself.

Attention works the same way and the overhead is worse, because attention does not parallelise. When you check the pace on your watch mid-run, that check is not free time carved out of a separate budget. It comes out of the run. When you notice, during a conversation with your kid, that this is a good moment and you should be present for it, the noticing has already relocated you slightly outside the moment. The observation and the observed are competing for one resource.

This is why the advice to be present is so annoying and so hard to act on. The instruction to attend to your experience creates a second process whose job is to check whether you are attending, and now there are two of you in the room and one of them has a clipboard.

The clearest case I know is meditation, which is the one domain where evaluating the session and having the session are flatly incompatible. Sit down, and the mind offers immediately to assess how it is going. Is this working. Was yesterday better. On the mornings I take that offer, nothing happens for forty minutes except an increasingly detailed report on nothing happening. The practice does not fail because I lack discipline. It fails because I brought a measuring instrument into a room that only has space for one thing.

What happens to a number once you start aiming at it

Two well-documented mechanisms explain most of how this goes wrong, and it is worth knowing both by name, because once you can name them you start seeing them everywhere.

The first is Goodhart's Law. Charles Goodhart was writing about monetary policy in 1975; the anthropologist Marilyn Strathern later gave it the phrasing everyone uses — when a measure becomes a target, it ceases to be a good measure. The mechanism is not that people cheat, though sometimes they do. It is that any metric is a lossy compression of the thing you actually care about, and the moment you optimise against the compression, all the pressure lands on the gap between the two. Any engineer who has watched a team told to raise its velocity number can complete this story without help. The number goes up. Nothing else does.

Your own life is not exempt. Steps are a compression of movement, and movement is a compression of a healthy body, and a healthy body is a compression of being able to live the way you want. Each step in that chain loses something. Aim hard enough at the first one and you get an evening spent pacing the hallway at 11 p.m. to reach a round number, which serves the metric perfectly and the original intention not at all.

The second mechanism is the overjustification effect, and it is the one that does the deeper damage. In a 1973 study, Lepper, Greene, and Nisbett gave preschool children who already loved drawing an expected reward for drawing. Afterwards, in free play, those children drew noticeably less than the ones who had never been rewarded. Adding an external reason for doing something did not stack on top of the internal reason. It displaced it.

A metric is an external reason. A very good, very sticky, quantified external reason, delivered daily. Which means that tracking something you do out of love carries a real risk that the love quietly moves out and the tracking moves in — and you will not notice on the day it happens, because the behaviour continues. What changes is why. Ask anyone who logged their reading for a year and then found, when they stopped logging, that they did not much want to read.

When the metric eats the thing

There is a clinical example so on-the-nose that it barely needs commentary. In 2017, sleep researchers at Rush and Northwestern described a pattern they named orthosomnia: patients arriving at sleep clinics because their trackers said their sleep was poor, whose sleep then genuinely deteriorated because of the anxiety and the effort of trying to produce better data. The devices were not especially accurate. It did not matter. The number had become the condition, and people lay awake at night working on it.

What makes that case useful is the shape of it, which repeats far outside sleep.

A person starts running because it clears their head, then starts tracking it, then finds that a run without the watch feels like it did not happen. The head-clearing is still available. It is just no longer the point.

A couple starts a shared budget spreadsheet to stop arguing about money, and eighteen months later the spreadsheet is what they argue about. The category is more precise. The conversation is worse.

Somebody photographs everything on a trip in order to remember it, and comes home with a complete record and a thin memory, having spent the week as the crew rather than the audience.

The common thread is not that tracking was pointless. In every case it started by solving something real. The thread is that nobody set an end date, so the scaffolding stayed up after the building was finished, and eventually people started to think the scaffolding was the building.

Scaffold or substitute: how to tell

Here is the distinction I have found most useful. A metric is either a scaffold or a substitute.

A scaffold helps you build something you cannot yet do unaided. It is legitimately valuable, often necessary, and its defining feature is that it is supposed to come down. Weighing your food teaches you what a portion actually looks like, and when you know, you can stop. A budget teaches you where the money goes, and when you know, the spreadsheet's job is largely done. Tracking sleep for a month tells you the true shape of your nights, which is genuinely worth knowing, because self-report on this is close to useless.

A substitute is what a scaffold becomes when nobody takes it down. It no longer teaches you anything — you have long since learned what it had to teach — but it has taken over the job of telling you whether the day was good.

Four questions separate them, and they are more diagnostic than they look:

• Has it taught you anything new in the last two months? A scaffold still under construction produces surprises. A substitute produces confirmations. If you can predict the number before you look, you already have the information, and you are collecting it for a different reason than the one you started with.
• Do you check during, or after? Checking afterwards is record-keeping. Checking during is the metric taking a seat inside the experience, and it is the single strongest signal that the overhead has stopped being small.
• Has the shape of what you do bent toward the number? Walking a lap of the block at night to round up. Choosing the shorter book. Ending a conversation because the session is over. This is Goodhart arriving in person.
• What happens when the two disagree? This is the real test. You had a genuinely good run and the watch says it was slow. A meditation that felt spacious and the app says eleven minutes. If the number wins — if you feel worse about a good experience because it recorded badly — then the metric is no longer describing your life. It has replaced the part of you that was supposed to judge it.

That last one is the one to watch. Everything else is calibration. That one is a transfer of authority.

The things that never had a number

There is a further problem, quieter than the others. Instruments do not measure what matters most; they measure what is easy to instrument. And whatever gets counted gets attention, so a life organised around metrics slowly tilts toward the countable — not by anyone's decision, but by the drift of where attention goes.

Nothing measures the quality of an hour spent with someone you love. Nothing measures whether you were actually there for it. There is no metric for the specific ordinary satisfaction of a problem that finally came apart cleanly, or for a Tuesday that was fine, or for having been kind in a moment when it cost something and nobody saw.

These are not lesser goods that we tolerate being unmeasured. For most people, asked honestly, they are the whole account. And they are systematically underweighted by any system that assigns numbers to the other things, because the numbers are vivid and daily and the unmeasured things are quiet.

A month with one thing left alone

Here is the practice, and it is deliberately small. Not a digital detox, not deleting everything. One domain, one month.

Pick the wrong one on purpose. Go through what you currently track and choose the one you would least like to stop measuring. That reluctance is the whole signal. The domain you are relaxed about giving up is not the one holding anything.

Actually stop, do not just look away. Take the watch off for the run rather than wearing it and resolving not to check. Turn off the ring notifications. If the device records passively and you cannot disable it, at minimum do not open the app for thirty days — but genuinely off is better, because a number waiting to be read is still exerting force.

Keep doing the thing. This matters. The experiment is not stopping running; it is running without the record. If the activity itself collapses the moment nobody is counting, that is not a failure of the experiment. That is the result, and it is the most useful thing you could have learned this month.

Replace the number with a sentence. Once a week, write one line about how that domain is actually going. Not a rating — a sentence. Sleeping badly on Sundays for some reason. Runs have been better since I stopped caring about pace. This is not measurement; it is a different faculty, and for most of us it has gone soft from disuse.

At the end, ask three things. Did the behaviour hold up without the metric? Did the experience change — did anything come back that you had not noticed was missing? And are you reaching for the tracker now because you need information, or because you want to know how you did?

If the honest answer is the second one, you have found out something worth the month.

A version of this that generalises: give every new metric an expiry date the day you start it. Six weeks, twelve weeks, whatever fits. When the date arrives, the default is off, and continuing requires a reason you can say out loud. That single change turns tracking back into what it should have been — a tool you pick up for a job, rather than a permanent tenant who has forgotten they were once a guest.

Questions worth asking

Is this an argument against tracking?
No. It is an argument against tracking with no end date. Measurement is unusually good at one specific job — correcting a wrong belief about your own baseline — and self-report is unreliable enough that you often cannot do that job without it. The failure mode is not starting. It is never stopping.

What if the metric is the only reason I do it at all?
Then keep it, and know what you have. There is nothing shameful about needing external structure; a great deal of useful behaviour runs on it. But be clear-eyed that a habit held up entirely by a number is a habit resting on a foundation you do not control, and it is worth occasionally testing whether anything else is holding it. That is what the month tells you.

Is not tracking just a different kind of optimising?
It can be, and this is the trap directly in front of anyone who takes this seriously. Making unmeasured time into a discipline you are good at, and quietly scoring yourself on how present you were, reproduces the whole problem with better branding. The only defence I know is to notice when a practice has acquired a scoreboard, and to be willing to abandon the practice rather than defend the scoreboard.

How do I know if a month was long enough?
The first two weeks are mostly withdrawal and will tell you very little. The signal, if there is one, tends to arrive somewhere in week three, usually as something unexpected returning — noticing the route rather than the pace, finishing a chapter without checking how many pages are left. If nothing has come back by day thirty, that is real information too: the metric was probably fine, and you can pick it up again with a clearer conscience.

The old traditions were unambiguous about this, which is worth sitting with for a second. They do not say measure your practice; they say do it, and stop looking. That is not anti-intellectualism. It is a very old observation about attention — that some things are only available to a mind that has stopped standing beside itself, taking notes on how it is doing.