NeuroDrift · how-to · guide

The Shutdown Protocol

How to fall asleep in an age that steals sleep: an evidence-based descent corridor for an overclocked mind.

Method: A dimmer, not a switch → The landing → The shutdown sequence · techniques graded by evidence · calculators · bedroom audit · sleep under hard conditions

1:30 a.m. You're lying there, and the system is idling at full RPM: forty tabs, three half-finished ideas, tomorrow's conversation you've already lost out loud twice. For the first time in history, the tool on your desk (Claude, GPT) removes the ceiling: everything you've spent years shelving as "someday" is suddenly doable — today, by you alone. And the same ceiling that freed you by day won't let your mind sit down at night.

So you do the one thing that seems logical: you try to fall asleep — that is, flip a switch that doesn't exist. And the harder you press the button, the brighter the system burns. You're not alone: a 2025 AASM survey found 38% of adults admit scrolling in bed wrecks their sleep (among 18–24-year-olds, 46%), and one in four knowingly ranks screen time above sleep.

Sleep is a dimmer, not a switch: you don't flip it on, you turn it down. You don't hard-kill an overclocked system — you run it through a shutdown sequence.

This guide isn't "10 tips for better sleep." It's a protocol: a diagnosis of why the system won't shut off; a step-by-step descent corridor; an honest, evidence-graded breakdown of techniques; and separate routes for people falling asleep under sirens, with a newborn through the wall, or with an ADHD brain that fires its best ideas at exactly 11:47 p.m.

MethodThree metaphors, nested inside each other

Most people "try to fall asleep" — and that's exactly why they don't. Falling asleep has no button. It has a sequence. Three images describe the same mechanism at three levels:

A dimmer, not a switch — the diagnosis. The brain has no "sleep" toggle. The shift from active to asleep is a gradual dim, like a dimmer switch, not a click.
The landing — the shape. A plane doesn't drop from cruising altitude to the ground in one second. There's a glide-path: the descent starts long before touchdown. Your descent corridor is the last couple of hours.
The shutdown sequence — the steps. You power the system down process by process, in the right order: mind (offload) → light (dim) → body (cool) → control (let go).

"Don't press the button — turn down the dimmer."Each step below shuts off one process. Together they don't "sedate" you — they clear the obstacles, and sleep arrives on its own, because that's its default.

Layer 1 · DiagnosisWhy the system won't shut down

Sleep is the biological default: clear the obstacles and it arrives. So the question isn't "how do I fall asleep" but "what am I keeping switched on when it's time to power down." The 2020s brought a whole new panel of switches that keep the system in the red zone.

30.5%

of US adults sleep under 7 hours (2024)

CDC/NCHS 2024

12.4%

have a clinical insomnia disorder — not "1 in 3," as it's often repeated

van Straten 2025

38.5%

insomnia rate in Ukraine one year after the invasion — ≈double the baseline

Wang 2024

+25%

odds of short sleep for every extra hour of evening screen time

548k people, 2025

The new sleep thieves (the screen era and the raised ceiling)

Revenge bedtime procrastination — when the day got eaten by other people's demands, you steal "your" time back from the night. The one hour you control is the one you should be sleeping through.
Doomscrolling and feeds — it's not so much "blue light" as activation: anxiety, dopamine swings, "one more video" right when your mind should be winding down.
The raised ceiling of ambition — when one tool lets you do what used to take a whole team, the mind doesn't want to "close up shop": there are more ideas than hours, and the best ones show up on the pillow.
Hyperstimulation and overload — war, news, work, kids: the system runs at 100% CPU all day and has forgotten how to brake.

Layer 2 · The descent corridorThe shutdown protocol, step by step

Here's the glide-path. You don't have to do all of it — but the more steps you take, the lower the system arrives at the pillow. A practical mnemonic scaffold (popular, but it's a heuristic, not a law): 10–3–2–1–0 — 10 hours no caffeine, 3 no heavy food/alcohol, 2 no work, 1 no screens, 0 "five more minutes" in the morning.

The "mind" step: offload, don't "stop thinking"

"Stop thinking about it" is an instruction the brain can't execute: trying not to think about a thought only spotlights it. The working alternative is offloading: 90 minutes before bed, write down on paper everything that's open (tasks, worries, ideas) — not to solve it, but so the system can let itself stop holding it in RAM. In a lab study (Scullin, 2018), people who spent five minutes before bed writing a tomorrow's to-do list fell asleep on average nine minutes faster than those who wrote about what they'd already done. Caveat: these were healthy young adults; the effect hasn't been re-tested in insomnia patients — but the move is cheap and safe.

The "light" and "body" steps: caffeine, temperature, timing

The two strongest levers you actually control are caffeine (when you last dropped an adenosine blocker) and core temperature (sleep onset rides on a drop in internal temperature — which is why a warm bath 1–2 hours out "tricks" the body into cooling harder afterward). Do your two numbers:

What you drank

Time of last dose

Lights-out

Metabolism

Caffeine's half-life averages ~3–5 h, but the range is 2–10 h depending on your CYP1A2 gene, smoking, pregnancy, and oral contraceptives. That's why "stop caffeine at 2 p.m. for everyone" fails: a slow metabolizer still has that espresso active at midnight. The calculator is a guide, not a verdict.

Layer 3 · TechniquesAn honest breakdown and comparison

There are dozens of techniques, and all of them get sold with the same confidence. The difference is evidence. Below is a breakdown by group and a sortable table showing what has a clinical foundation (strong), what helps many people but with thinner evidence (moderate), and what's mostly an expectation effect (weak) — though sometimes that's enough too.

One anchor above all: if insomnia is chronic (≥3 nights/week, ≥3 months), clinical guidelines put the first line not as a pill but as CBT-I (cognitive behavioral therapy for insomnia). The rest of the techniques are its building blocks, or first aid for a single rough night.

Sleep techniques — by evidence (click a header to sort)
Technique ↕	How it works	Evidence ↕	Time to effect ↕	Effort ↕	Who it's for

Evidence grades are checked against reviews and guidelines (see "Sources"). "Time to effect" for behavioral methods (CBT-I, sleep restriction) means weeks of consistent practice, not one night.

Layer 4 · EnvironmentThe room that shuts down for you

The cheapest lever isn't a technique — it's the room. Dark, quiet, cool, and one hard rule: the bed is for sleep (and sex) only. The moment you work, eat, and scroll in bed, the brain learns that bed = "the place where sleep doesn't happen." Score how much your bedroom is working against you:

Darkness

Noise

Temperature

Screens in the bedroom

Clock visible at night

Bed = sleep only

Layer 5 · Myths on the tableWhat pop culture sells you for nothing

Click a card to flip the myth into what the data actually shows.

Interesting, but the evidence is weak (still worth mentioning — works for some people)

Not everything under-proven is useless. The expectation effect is real, and if a safe gadget helps you — use it. The point is knowing the grade so you don't pay for the miracle with broken sleep.

4-7-8 breathing weak — no strong RCT specifically for insomnia; rests on small, unblinded self-reports plus the physiology of a long exhale. Fine as a wind-down ritual; the viral "fall asleep in 60 seconds" claim is false.
Yoga nidra / NSDR weak — 6 RCTs (n=244) give a signal for sleep onset; but the viral "+65% dopamine" claim comes from a single PET scan of 8 experienced practitioners and isn't about sleep at all. Good for daytime recovery after a short night.
Cognitive shuffle weak — conference posters and small student samples, no PSG. A plausible, low-risk try for people who don't like writing things down; not therapy.
Tart cherry (Montmorency) weak — there's a real objective signal for sleep efficiency, but from a few small trials. The viral "+84 minutes" claim comes from an 8-person pilot.
Lavender (aromatherapy) weak — signal in older adults, but study heterogeneity of 87–98% (a red flag), subjective outcomes, and an effect inflated by expectation.
Weighted blanket weak — pooled effect on insomnia is nonsignificant; the more reliable benefit is a modest drop in anxiety. Safe (~10% of body weight); use caution with children and breathing restrictions.
ASMR weak — in "responders" it genuinely lowers heart rate (−3.4 bpm), but that's a cardiac effect, not one on sleep latency; there's almost no RCT evidence for insomnia, and not everyone experiences ASMR.
Sleep stories / calm audio weak — same cognitive-distraction mechanism as imagery; helps with racing thoughts. Set a timer so it doesn't play all night.
The "military method" weak — a combo of PMR + imagery (each has separate support); but "fall asleep in 10 seconds" is a 1980s anecdote with no RCT, and it takes weeks of practice.
Warm milk weak — not enough tryptophan to cross the blood-brain barrier. The benefit is more ritual, warmth, and conditioning than pharmacology.
The "10-3-2-1-0" rule weak — the formula itself has never been tested; its value is that its parts (caffeine, alcohol, screens) each have support of varying strength. The effect depends on which part was your problem.

Layer 6 · Hard conditionsWhen "just go to bed earlier" doesn't work

The general protocol is the baseline. But there are life situations where it breaks against reality. Seven routes:

🛡️ War, alarms, hypervigilance, PTSD nightmares

IRT — rewrite the nightmare. The only AASM-recommended nightmare therapy: during the day you rewrite the nightmare's storyline into a neutral one and rehearse the new version. Effect at 4–8 weeks, holds for 6–12 months. · evidence: strong
CBT-I — the baseline against stress- and PTSD-driven insomnia; a digital version lowers the access barrier. · evidence: strong
Tactical breathing (box / extended exhale) — not a cure, but a "bridge": it quiets hyperarousal in the moment of an alarm. · evidence: weak
Prazosin — use with caution, not first line. After the large VA PACT trial (2018), its status is "possibly for select hyperadrenergic cases," not universal. · contested
Social support genuinely buffers the path from stress to chronic insomnia. · evidence: moderate

👶 Young kids (new parents)

Structural offloading of night care — a partner takes 1–2 feedings so the primary caregiver gets a consolidated block. Works from night one, buffers postpartum depression. The main lever. · evidence: moderate
CBT-I — once fragmentation has turned into insomnia in its own right. · evidence: strong
Anchor sleep — a fixed 4–5-hour sleep block at the same time amid the chaos. · evidence: weak
A realistic frame: mothers lose ~1 hour of sleep/night in the first months and don't fully recover for years — this is a structural load, not a hygiene failure. "Sleep education" alone doesn't give you back the sleep. · evidence: moderate

🌗 Overtime / shift work

Damage-control package: a prophylactic nap before a night shift + caffeine in the first ~2 hours of the shift + bright light while working + blackout and earplugs for daytime sleep. · evidence: moderate
Anchor sleep — a stable sleep core amid rotations. · evidence: weak
Psychological detachment / wind-down — 30–60 min transitioning from work to rest, a "worry dump," no email or Slack. · evidence: moderate
Honestly: no tactic normalizes the physiology of night shifts — it's partial adaptation, not alignment. Ongoing night work is a recognized cardiometabolic and mood risk. · evidence: moderate

⚡ ADHD

Timed low-dose melatonin — 0.3–0.5 mg (adults) / 1–3 mg (kids) taken several hours before desired sleep shifts phase (DLMO +44–88 min). It's a clock, not a sedative. · evidence: moderate
Morning bright light + avoiding evening light — 2 weeks of morning light shifts the rhythm; paired with melatonin, the effect is largest. · evidence: moderate
CBT-I — for the insomnia and rumination component. · evidence: strong
Context: up to 80% of adults with ADHD have sleep disorders, with the rhythm shifted ~90 minutes — often circadian delay, not "poor discipline." · evidence: moderate

🏙️ City noise (and light pollution)

Get noise below ~40–45 dB — earplugs, sealing windows. · evidence: strong (thresholds)
Stable broadband masking — white/pink noise or a fan smooth out the peaks that wake you. · evidence: moderate
A dark bedroom (≤1 lux) — city life means light pollution; blackout curtains or a mask. · evidence: moderate
Remember: noise does damage below the threshold of conscious waking — micro-arousals and cardiovascular reactions. Quiet isn't "oversensitive exaggeration." · evidence: strong

🕰️ Age

Realistic expectations — nighttime cortisol rises, wakings increase, and things plateau around age 60. Managing expectations is itself therapeutic. · evidence: strong
Morning light + a consistent wake time — circadian anchoring against phase drift. · evidence: strong
CBT-I — works in older age too. · evidence: strong
Tart cherry — a low-risk "food-first" try (a pilot showed an effect in over-50s, but the sample was tiny). · evidence: weak

🌡️ Hormones (menopause)

Hormone therapy (HT) — when it's specifically vasomotor symptoms (night sweats and hot flashes) breaking your sleep; needs a prescription and an individual risk assessment. · evidence: moderate–strong (for VMS)
CBT-I — works for menopausal insomnia: it won't stop the hot flashes, but it fixes the insomnia. · evidence: strong
Gabapentin — an evidence-based alternative for VMS. · evidence: strong (within meta-analysis context)
A cool bedroom + light bedding — directly counters night sweats. · evidence: strong
Avoid sleeping pills as a first move — a network meta-analysis found hypnotics give no significant benefit for menopausal sleep. · evidence: strong

HonestlyWhen the protocol breaks — and the limits of the science

Honesty & limits

The objective-subjective gap is the field's main fault line. Even CBT-I shows large effects on self-report, but on objective measures (polysomnography) sleep barely changes. It's entirely plausible that insomnia is partly a disorder of sleep perception. Keep in mind: most "sleep improved" findings in technique studies are subjective measures.

The loudest pop-science numbers are migrated or invented, and we didn't let them in here: "caffeine cuts deep sleep by 20%" (actually from teenagers; ~1% in adults), "yoga nidra +65% dopamine" (one scan of 8 practitioners, not even about sleep), "tart cherry +84 minutes" (an 8-person pilot), "53% practice revenge bedtime procrastination" (fabricated — the real source had 177 people and no such numbers).

The whole "breathing / supplements / gadgets" family is softer than the marketing. The AASM gives relaxation only a conditional recommendation on low-quality evidence; it recommends against melatonin, valerian, and tryptophan for chronic insomnia. The effects are real but small — minutes, in small and biased trials. "There's evidence" ≠ "it's proven."

When the protocol isn't enough — see a doctor. If insomnia is chronic (≥3 nights/week for over 3 months), guidelines put CBT-I as the first line, ideally with a specialist. Loud snoring, witnessed breathing pauses, or heavy daytime sleepiness are signs of possible apnea: no technique here substitutes for diagnosis. Persistent early waking paired with a depressed mood is a reason to check for depression.

What the science actually knows solidly (the load-bearing column, not the decoration): CBT-I beats pills over the long run; caffeine and alcohol genuinely damage sleep architecture; light timing and temperature act on measurable physiology; melatonin-as-chronobiotic works for jet lag (you'd need to treat ~2 people to help one). The rest is decoration of varying strength: use it, but know the grade.

FinaleYou don't "fall asleep" — you finally let the system land

1:30 a.m., the same room. The difference isn't that you learned some secret trick. It's that you stopped pressing a button that doesn't exist and started turning down the dimmer: offloaded your mind onto paper, dimmed the light, cooled the room, let go of control. The system flew the glide-path and landed on its own — because sleep, once you clear the obstacles, is its default. You didn't beat the night. You just stopped holding the switch down.

FAQFrequently asked questions

How many hours of sleep do I actually need?

Adults need 7+ (individual need runs ~6–10). "Exactly 8" is a cultural round number, not a clinical threshold. Go by how you feel during the day, not by a number: chasing a specific figure breeds orthosomnia — anxiety about sleep that ruins the sleep itself.

I wake up in the middle of the night — is that bad?

No. A few brief wakings per night are normal. The problem starts when you check the clock and try to "force" sleep. Don't check the time, don't grab your phone; if you're not back asleep in about 20 minutes, it's better to get up, sit in dim light, and return to bed once you're actually sleepy (stimulus control).

Is melatonin a sleeping pill?

No, it's a chronobiotic: it shifts your internal clock rather than "knocking you out." Its territory is jet lag, shift work, and a late rhythm. As a "fall asleep" pill it barely works (~7 minutes gained), and the AASM recommends against it for chronic insomnia. Small doses (0.5 mg) work no worse than large ones (5–10 mg) — and large doses add morning grogginess.

When's the last time I can drink coffee?

There's no single cutoff for everyone: caffeine's half-life ranges from 2 to 10 hours depending on genetics. Work out your own remainder in the calculator above; the general rule is to stop ≥6–8 hours before bed, earlier still for slow metabolizers. And remember the dose: a small coffee (~100 mg) is far more forgiving than a large one (~400 mg).

When is it time to see a doctor instead of reading guides?

If insomnia persists at least 3 nights a week for over 3 months, that's chronic, and guidelines put the first line not as a pill but as CBT-I (ideally with a specialist, or through an evidence-based app). Separately, watch for warning signs: loud snoring, breathing pauses noticed by a partner, or heavy daytime sleepiness — signs of possible apnea, which no relaxation technique treats.

SourcesWhat this stands on

van Straten et al. Insomnia prevalence (strict DSM ~12.4%). J Sleep Research, 2025
CDC/NCHS. Short sleep among US adults (30.5%, 2024). Data Brief № 559
Youngstedt et al. No objective secular decline in sleep. Sleep Med Rev, 2016
Hafner et al. (RAND). Economic cost of insufficient sleep (up to $411 billion). RAND Europe, 2016
Wang et al. Insomnia in Ukraine one year after the invasion. Psychosomatic Medicine, 2024
Edinger et al. AASM guideline: CBT-I is the only STRONG recommendation. JCSM, 2021
van Straten et al. CBT-I effects (severity g=0.98; TST only 0.16). Sleep Med Rev, 2018
Furukawa et al. CBT-I vs pharmacotherapy (remission 41% vs 28%). Psychiatry Clin Neurosci, 2024
Mitchell et al. CBT-I on objective measures (PSG). Sleep Med Rev, 2019
Scullin et al. Pre-bed to-do list → −9 min to sleep onset. J Exp Psychol Gen, 2018
Phillips et al. Melatonin suppression by light (~25 lux). PNAS, 2019
Brown, Brainard, Cajochen et al. Consensus lighting thresholds. PLoS Biology, 2022
Haghayegh et al. Warm bath 1–2 h out → shorter SOL. Sleep Med Rev, 2019
O'Callaghan et al. Caffeine half-life (range 2–10 h). 2018
Meta-analysis of acute caffeine (TST −34.67 min; deep sleep ~1%). Sleep Med Rev, 2025
Gardiner et al. Alcohol suppresses REM. Sleep Med Rev, 2024
Sateia et al. AASM guideline: against melatonin/valerian/tryptophan for chronic insomnia. JCSM, 2017
Herxheimer & Petrie. Melatonin for jet lag (NNT≈2). Cochrane, 2002
Chang et al. Screens before bed (melatonin −55%). PNAS, 2015
Frontiers. Each +1 h of screen time → +25% odds of short sleep (548k). 2025
Kroese et al. Bedtime procrastination (N=177). Front Psychol, 2014
Morgenthaler et al. AASM recommends IRT for nightmares. 2018
Richter et al. New parents' sleep (mothers −1 h/night). Sleep, 2019
Depner et al. Weekend catch-up sleep doesn't save metabolism. Current Biology, 2019
AASM Sleep Prioritization Survey (doomscroll 38% / 46%). 2025

How to use this. Run your own numbers through the calculators, build your corridor in the checklist, match a technique to your symptom — and start with one night. This is a guide, not medical advice: for chronic insomnia, daytime sleepiness, or loud snoring, see a doctor. · Читати українською →