The same four weeks, with the parts that actually decide whether it works. What to set up, what each day tends to feel like, and what to do when it goes wrong.
Before day one
The plan is fixed and the content is yours. Before anything else, answer these five in writing. Not in your head, because in your head every answer is "sometimes, it depends," and that's not a thing you can reclaim.
Where are you when you stop being awake, what is the last thing you look at, and what time is it really? Not your intended bedtime. The actual one.
How do you move between places, how long does it take, and what is playing right now while you do it?
Name three specific gaps under five minutes that you reliably fill. Getting dressed. Waiting for coffee. The walk from the parking lot.
After a hard day, what is the first thing you reach for, and how long does it last before you notice you're doing it?
Which hours does the device have to be in your hand for money reasons, and where exactly does that stop being true? Be honest about the second half. It's the harder question and the answer is usually earlier than you'd like.
Keep the answers somewhere physical. You'll want to compare against them in five weeks, and the phone isn't the place to store a document about your phone.
Days 1–3
Measure, change nothing
Record your daily screen average, your top three apps by time, and the clock time you actually fell asleep. Three days, written down, no behavior change of any kind.
The instinct is to skip this because you already know it's bad. Skip it and you lose the only thing that will keep you going in week three, which is being able to look at a number from before and a number from now and see that something moved. Feeling better isn't evidence. You'll feel better and worse for reasons that have nothing to do with this.
"I already started, I couldn't help it."
Fine. Record three days anyway, starting now, with whatever you've already changed left in place. A slightly contaminated baseline beats none.
The number is worse than you expected.
It usually is, by one to three hours. Screen time reports undercount, because they don't capture the device being on across an activity you're also doing. Write down the real number and move on. It's the last time it will be that high.
Week 1
← Horizontal · the sleep window
Not face down. Not on silent. Out, in a different room, from the time you get home until the morning.
An alarm clock on the nightstand, bought and working before night one. Move the charger to wherever the device is going to live. If you're on call for real reasons, turn on the setting that lets repeated calls from the same number ring through, and leave the device close enough to hear it. Reachable isn't the same as in reach.
You'll lie there. Falling asleep takes longer, and the absence is loud. This is the most common place to quit and it's also the least meaningful data you'll collect all month.
It gets easier and you start sleeping better than you did. Mornings are the surprise: you'll notice how much of your morning mood was set in the first ninety seconds of the day, and that it now starts somewhere else.
It stops being a thing you're doing and becomes how the room is. That's the whole target for the week.
You share the room with someone who isn't doing this.
Then it's a conversation, not a rule you impose. Ask for what you need rather than prescribing what they should do. If they keep their device in there, you still move yours; a room with one device in it's meaningfully better than a room with two, and you don't get to make it a condition.
You woke up at 3am and went and got it.
One night isn't a reset, per rule three. Two in a row on the same window means you repeat the week. Put it back and go again.
You have to be reachable overnight.
Some people do. Then the device goes as far from the bed as it can go while still being audible, screen down, in a drawer. The point was never distance. It's that reaching it requires standing up and deciding to.
Week 2
← Horizontal · the transit window
Keep week one running. Add this: however you get from one place to another, do it with nothing playing.
If you drive, the device goes somewhere you can't reach from the driver's seat before you start moving, not after. If it pairs to the vehicle, calls and maps still work with it out of reach, which is the entire reason that's acceptable. If you ride or walk, the headphones stay in the bag, not around your neck.
Roughly ten minutes of real restlessness at the start of each one, then it thins. You'll reach for something that's not there, more than once, without deciding to. Notice it and keep driving.
Something starts happening in the quiet. Problems you've been carrying resolve themselves at stoplights. You'll want to write things down at the other end, which is a good sign and worth acting on.
You put a podcast on. It felt fine.
It always does, and it's the single failure this whole plan exists to prevent. A podcast in a reclaimed window is a vertical step with no horizontal move, which is the dashed arrow. Restart the week. This rule isn't negotiable and it's the one that decides the outcome.
Your mind goes somewhere you don't want to be.
This is the most important thing that happens in week two and it's the reason a decade of noise was there. If what surfaces is heavy, that's worth saying to a person, not worth covering back up. The training is real, but it isn't a substitute for the conversation.
Your commute is three minutes.
Then transit is a weak window for you and something else is your real one. Swap in the longest recurring window from your inventory and keep the ordering. The sequence matters more than the specific window.
Week 3
↓ Vertical · every screen you own
Weeks one and two stay running. This is the first vertical move and it's the one that makes every later move permanent, because up to now nothing has changed about what the devices are capable of. You've been reclaiming time in front of an unchanged machine.
Do all three screens in the same week. This isn't thoroughness for its own sake. Leave one open and the appetite reroutes into it within about a week, and you'll conclude the plan failed when what actually happened is that you moved the supply.
You can't delete the device your income runs through, so you separate the tools from the feeds instead: remove the browser or restrict it to a named list of sites you actually work with, remove the app store so removals stay removed, cut every notification that's not a human addressing you by name, and take the top rung out while you're in there. The full build is on the whiskey taster page, including why the passcode has to end up with somebody who isn't you.
Short video on a television is the ladder's newest move and the strangest one, because broadcast was the format that used to end on its own. Delete the feed apps from the set entirely. In whatever you keep, turn off autoplay-next, which is the single setting that converts an episode into a feed. If a service has no way to stop it, that service is a feed and it goes.
The browser is the entire problem here, same as on the phone. Block the feeds at a level you can't casually undo, log out of everything that's a feed, and delete the saved passwords for them so getting back in costs something. No feed apps at all; if it's worth having on the computer it's worth going to deliberately.
You'll reach for things that aren't there, ten or twenty times a day, without deciding to. That count is the most honest measure of how automatic this had become, and it's more informative than any screen time report you'll ever look at.
The reaching drops off sharply. What replaces it's a flat, bored, faintly irritable feeling. That's the recalibration, it's supposed to be unpleasant, and it's the first honest reading you've had of your own baseline in years.
You set a daily limit instead of removing things.
Then you didn't do week three. An allowance holds the baseline precisely where it is, and the baseline is the entire target. A countdown also makes the remaining minutes denser rather than lighter, because you're now racing it.
You did the phone but not the television.
This is the most common version of a half-finished week three, and it reliably produces a strange result: phone screen time falls, everything else stays the same, and you can't work out why nothing feels different. The supply moved to the biggest screen in the house. Finish the week.
You reinstalled something.
Very common and not the end of anything. Remove it again the same day, and this time deactivate the account rather than just the app. If it returns twice in one week, hand the passcode to someone else, because willpower has now been tested here and has a record you can look at.
A different app immediately got worse.
Expected. The appetite reroutes before it reduces, and it will keep doing that until the baseline moves. Don't chase it this week; chasing turns the plan back into the whack-a-mole you already tried. Week five handles it structurally.
Week 4
← Horizontal · transitions and one bare task
The three transition windows from your inventory, taken back. Then the part that matters most: one task a day, chosen the night before, done with nothing in your ears.
A routine one you always soundtrack. Cleaning something, a drive you already do, a stretch of physical work, the gym if music isn't load-bearing for you there. Not the hardest thing on your list, because the difficulty of the task isn't the variable being tested. The absence of accompaniment is.
Longer than it should feel. You'll finish and be surprised how little time passed, which is the reference point moving in the direction you want.
The first week where other people tend to notice something. Usually that you're less snappish in the evening, or actually present in a conversation. Take that as the signal, more than any number on a screen.
You forgot to pick the task.
Then it doesn't happen, which is why it gets chosen the night before. Choosing in the moment is a decision made by the version of you who's already tired, and that version doesn't choose this.
Silence at work isn't possible.
Then it isn't the work task. It's one task, once a day, and it can be four minutes long. Scale it down rather than skipping it. A short rep taken daily beats a long one skipped.
Week 5 on
↓ Vertical · formats that end
The second vertical move and the last one. From here, media has a stopping point built into it. An album. An episode. A chapter. A film. A thing that's over when it's over, without you having to be the one who decides.
This isn't abstinence, and it was never supposed to be. Stage-one media had this property as a matter of physics and infinite formats removed it on purpose, because a stopping point is a place where attention leaves. Restoring it's most of the work.
You're also not going back to the top rung. Not because it's forbidden, but because you've spent four weeks moving your baseline and returning to the densest thing available will move it straight back in a couple of days. That's not a moral claim. It's just how the ratchet runs.
How long does this last?
Indefinitely, and the framing of the question is the thing to watch. If you're counting down to when you can stop, you're treating this as a sentence rather than a place to live, and sentences end with a return to exactly where you started.
Are you allowed to listen to anything, ever?
Yes. An album while you work is fine. The rule was never silence everywhere. It's that specific windows stay uncovered, permanently, and that what fills the rest has an ending in it.
When it breaks
Every previous attempt collapsed at the first slip, because there was no defined response to one and the slip became a referendum on whether you're the kind of person who can do this. Here's the response, so it doesn't have to be a referendum.
One bad day.
Not a reset. Write it down, note which window, continue. A single day carries no information.
Two consecutive bad days on the same window.
Repeat that week from the start. Don't carry on to the next one with a broken window underneath it, because the weeks are cumulative and week three doesn't hold without one and two.
A whole week off the rails.
Restart at week one, not at the week you fell from. This feels like punishment and isn't. Later weeks depend on the earlier ones being live, and starting from the middle is how you get a fourth failed attempt instead of a first finished one.
A hard event knocked it over. A death, a crisis, a bad stretch.
Then stop the protocol on purpose rather than failing it by accident. Put it down, name a date, pick it up at week one on that date. Deliberately pausing isn't the same as collapsing, and treating it as the same is how people conclude they can't do this.
You finished and slid back three months later.
Normal. The economic pressure pushing density upward doesn't stop when you finish, so the baseline drifts back up unless the windows stay uncovered. Run it again. The second time takes about half as long.
None of this is touching it.
Then it may not be a habit problem, and a framework is the wrong tool. Compulsive use that survives a real structural change, especially when it's tangled up with sleep, mood, or something you'd rather not say out loud, is worth putting in front of a doctor or a therapist. That's not a failure of the plan. Some of this is out of reach of a plan.
If you ran this, or part of it, the week you stopped is the most useful thing you could tell me. Especially if you stopped. That's the information the plan is missing.