Nine minutes: the real gap between one more video and being asleep

It is past one in the morning, the room is dark, the phone is the only lit thing in it, and you have already told yourself twice that this is the last one.

What follows is arithmetic, not a lecture. Every number below is either attached to a named study or built in front of you out of numbers that are. Where a figure does not exist, I say so and leave the gap open rather than filling it with something that sounds about right.

Nine minutes is a sum, and you are owed both halves of it

Nobody has measured the gap between deciding to stop and being asleep, because that is not a thing science measures. So the number in the headline is not a finding. It is a sum with two parts, one of them sourced and one of them yours.

The first part is 4.86 minutes. That is the entire measured effect of blue-light blocking glasses on how long people took to fall asleep, pooled across three double-blind crossover randomised trials in a 2025 systematic review in Frontiers in Neurology. Total sample: 49 people. The effect was not statistically significant. Nothing else moved either.

Forty-nine people is a small pile of people, and "not significant" is not the same as "proven to do nothing". The honest reading is narrower than the debunk you will see elsewhere: at the most generous interpretation of the best available trials, the optical half of the bedtime problem is worth under five minutes, and might be worth nothing.

The second part is whatever you actually watched. I have put four minutes in the slot, which is eight short videos at thirty seconds each, because it makes the two halves roughly equal and the headline round. That is a placeholder and I am telling you it is a placeholder. Five plus four is nine. If your own stopwatch says nine, the headline got lucky. If it says ninety, the headline was being kind to you.

A panel of sleep scientists would not put their names to the blue light story

In May 2024 the National Sleep Foundation published an expert consensus statement in Sleep Health on screen use and sleep across the lifespan. Two of its results deserve to be read slowly.

The panel did not reach consensus that screen use impairs sleep health in adults. It did not reach consensus that blue light from screens before bed impairs sleep in adults. Those are the two sentences that the entire evening filter industry is built on top of, and a room full of sleep researchers could not agree on either one.

The same panel did reach consensus on three other things. For children aged 5 to 12, screen use impairs sleep health. For the same children, the content of pre-sleep screen use impairs sleep health. And across the board, behavioural strategies can reduce the negative effects of screen use on sleep.

That is not a finding that screens are harmless at midnight. It is a finding that the mechanism everybody repeats is the one that failed to gather signatures, and the mechanism that survived is about what is on the screen rather than what is coming out of it.

It is not the light coming out of the phone. It is what the phone is for.

What survives is duller, and much harder to sell you a fix for

Two mechanisms are left standing, and neither of them is optical.

The first is displacement. The hours were spent awake. This one needs no study and no panel, because it is subtraction. If you meant to be asleep at midnight and you were asleep at half past one, the ninety minutes are gone and no filter recovers them.

The second is cognitive and emotional arousal from the content. Fear, anger, excitement and anticipation delay sleep. That is the mechanism the panel endorsed in children, and it is the one that is mechanistically plausible in adults even where consensus did not land.

Both are worse news than blue light, commercially speaking. You can sell a tinted lens for light. Nobody has worked out how to sell a filter for anticipation.

The bill is real, it accumulates, and you cannot feel it arriving

In 2003, Van Dongen, Maislin, Mullington and Dinges published a study in Sleep in which 48 healthy adults aged 21 to 38 were randomly assigned to 4, 6 or 8 hours in bed for 14 consecutive nights. Restriction to 4 or 6 hours produced significant, cumulative, dose-dependent deficits on every cognitive task they measured. Six hours a night for two weeks produced vigilance deficits comparable to around two nights of total sleep deprivation.

The result that matters most at one in the morning is the last one. The participants' own ratings of how sleepy they felt did not track their objective decline. They rated themselves as slightly tired while performing markedly worse. There is no internal gauge. You cannot audit this by how you feel, because how you feel is the instrument that stops working.

Now the arithmetic, and note what it does and does not claim. That study manipulated time in bed. It says nothing about phones, and I am not going to pretend it does. But the phone is one route to six hours. Lose ninety minutes a night for a fortnight and you have spent 21 hours, and landed somewhere between the six-hour arm and the eight-hour arm of a study whose six-hour arm looked like two consecutive nights without sleep.

The three in the morning check is a separate charge on the same account

Waking in the night and reaching for the phone runs into sleep inertia, which is the transitional state of reduced arousal and impaired cognition immediately after waking: slower reactions, worse short-term memory, slower reasoning. Its severity depends on which stage you were pulled out of, worst from slow-wave sleep, on how sleep-deprived you already were, and on the time of night. Snooze alarms make it last longer, apparently because each forced awakening restarts it.

Here is where I stop. There is no reliable published figure for how much longer it takes to fall back asleep after checking a phone during a night waking. It was looked for. It is not there. Anyone who quotes you a number for that is quoting nothing, so time your own: note the clock when you put it down, note the clock at the next waking, and after four or five nights you will have a figure that applies to exactly one person, which is the only person it needs to apply to.

One more was never one, and the reason has an actual history

The phrase for this is revenge bedtime procrastination, and it did not start as psychology. It started as the Chinese bàofùxìng áoyè, 報復性熬夜, "retaliatory staying up late", which spread on Weibo around 2016 in response to 996 working culture: nine in the morning to nine at night, six days a week. It reached English through a tweet by the journalist Daphne K. Lee on 25 June 2020, describing people with little control over their daytime hours refusing to sleep early in order to reclaim some sense of freedom at night.

It is a popular coinage, not a clinical diagnosis, and nobody should write it as one. But the origin explains the failure of every plan you have made in the dark. The late hour is not a lapse in willpower. It is the only hour of the day nobody else has claimed. Any fix that treats you as weak will lose, because you are not being weak, you are being outbid for your own evening.

The only thing with consensus behind it is the boring one

Go back to the panel's third result: behavioural strategies and interventions can reduce the negative effects of screen use on sleep. That is the one they agreed on. In a randomised trial of 105 families, restricting screens in the hour before bed for toddlers produced small to medium improvements in objectively measured sleep efficiency and night awakenings. Toddlers are not you, and the effect on measured attention was nil. But it is real evidence that the behaviour is the lever even in the places where the optics are not.

Which leaves a plain problem. The mechanism is displacement, the trigger is content, the deficit is invisible to you, and the decision has to be made by a version of you who has already decided twice tonight and lost.

1. Get your own number first. One night, note the clock at the moment you decide to stop and the clock when the phone goes face down. Everything after this depends on that gap being yours rather than mine.
2. Put the decision earlier than the moment. A rule made at nine in the evening is made by someone with a functioning gauge. A rule made at one in the morning is made by someone whose gauge, per the 2003 study, is reading wrong.
3. Make the rule cost something to break. Not shame. A delay, a walk, a stated price, anything that puts a gap between the urge and the tap.

Which is roughly the whole reason a small dog is involved

Scrappy does three things and refuses to do a fourth. Quiet hours, a daily limit, and a limit on how many times you open a thing. Quiet hours is the one this article is about: you set the hour while you still have judgement, and after that a dog is standing in front of the app.

He escalates when you push. Polite first, then surprised, then firm, then shouting, which is one word and the word is "No.", and after that wounded. He never tells you screen time is bad for you, because he has read the same consensus statement and it did not say that. He made a deal with you earlier in the evening and he is keeping it on your behalf.

You can buy your way past him with a treat, which buys five minutes, and a treat is earned by walking roughly 2,000 steps. Treats are never sold, and never will be. That is worth putting next to the number at the top of this article: five minutes of extra scrolling costs you two thousand steps, and it is more time than the best trials could reliably attribute to an entire evening of blue-light avoidance.

Each app gets a resistance setting. Negotiable takes the treat. Firm makes you sit still for ten seconds first. Absolute means he will not take the treat at all, from you, at that hour, for that app. There is one emergency pass a month with no scolding attached, because a rule with no exit is a rule you delete at 2am. Turning off the removal lock takes a full day, which is deliberately longer than any urge lasts.

No account, no sign-up, no server of ours. Screen Time is drawn by a sandboxed Apple extension that cannot phone home. The free tier is one whole rule, uncrippled, plus every safety rail, and quiet hours is a perfectly good rule to spend it on.