Case 1 — Participant 1 (the founder)
Chatroom flood at 200–500 pings/hr at a university desk. Restraint fell 8→2 in one day. Ended with 106 unbroken minutes.
Page six · Data
Every number we have. And a column reserved for you.
Four people have run the Restraint Method under observation — each one on a different phone, in a different place, with a different flood. That accident is the most useful thing about this dataset: the four cases are four variations of the method, and one of them probably matches your exact situation. Read your case. Run your case. Post your numbers.
This is a founding dataset, not a proof. n = 4. Every measure is self-report. Every participant knew what was being tested, and the founder coached all of them. No control group existed — with two exceptions worth your attention: twice, the flood was secretly switched off mid-experiment, and neither participant noticed. Their focus held with nothing arriving. Those two deception probes are the strongest evidence on this page, and they're marked where they happen. Everything else is a baseline for you to beat — that's exactly why your replication matters. Instrument: self-report scales, /7 (the first day of the pilot used /10 — labeled wherever shown).
Find your variation
Two axes, four cells, all four covered. Pick the cell you live in.
← In person · Online → iPhone Android
One curve, four times
Different phones, different floods, different rooms — the same three lines every time. This is the shape you're looking for in your own numbers:
Restraint falls
the effort of not looking → 0
Focus holds or rises
while the flood keeps falling
Aversiveness fades
the pings stop mattering
shape shown: Case 4's actual 30/60/90-minute scores, /7
Case 1 · Pilot
Participant 1 (the founder) — the chatroom flood
- Device
- iPhone
- Setting
- university, in person
- Flood
- Telegram public chatrooms — never a bot, never an app
- Dose
- 200–500 pings/hr, escalating
- Coaching
- none — self-run, first ever
- Structure
- 25-min pomodoros → limit test
- Scale
- /10 (pre-standardization)
- Date
- April 2025
This is the origin: the first person to sit down, point a firehose of meaningless chatroom noise at his own phone, and refuse to look. No bot existed yet — the flood was real public Telegram communities, which means the dose was never fully under control. Here's what the chatrooms actually delivered:
bar = average pings/hr during trials whisker = observed range
And here's what happened to the cost of ignoring it, six pomodoros into day one:
— focus — restraint — PCT (phone-conscious thought, dashed)
| Session (day 1) | Focus | Restraint | Study quality | PCT |
|---|---|---|---|---|
| Pomodoro 1 | 8.5 | 8 | 8 | 3 |
| Pomodoro 2 | 8.5 | 6 | 8 | 3 |
| Pomodoro 3 | 9 | 4 | 8 | 2 |
| Pomodoro 4 | 9 | 3 | 8 | 0 |
| Pomodoro 5 | 9 | 2 | 9 | 2 |
| Pomodoro 6 | 9 | 2 | 7 | 2 |
The transfer spike — changing the cue re-raises the price, briefly
Switching from sound to flash-only training briefly pushed restraint back up (5 → 7 → 8, with two slips) before it collapsed again. The training transfers; the transfer is not free. Then came the deception session — music plus flash, and:
| Flash-only sequence | Focus | Restraint | Study quality | PCT |
|---|---|---|---|---|
| New system 1 | 9 | 5 | 9 | 2 |
| New system 2 | 7 | 7 | 6 | 0 |
| New system 3 — looked twice | 6 | 8 | 7 | 0 |
| Deception (music + flash) | 10 | 0 | 10 | 0 |
"i have NEVER been able to study with a phone on my desk constantly turning on and pinging, but now i can, without effort." — Participant 1's session log
The limit test — and the honest ceiling
106 minutes continuous, ringer, no music: focus 10, restraint 0. Then, near the 1h46m mark of cumulative ringer exposure, aversiveness spiked without warning — "notifications became unbearably irritating and I had to turn it off." He stopped. That spike is why the stop-at-any-time rule outranks every number on this site: ringer is the training dose, not the operating mode.
What this case can't tell you: it's the founder testing his own idea — maximum motivation, maximum expectancy. The chatroom dose was uncontrolled (that's the whiskers above). Scale was /10, before the instrument settled. Treat it as the origin story with numbers attached, not the clean trial — the clean trials are Cases 3 and 4.
Run this variation: it's Version A — public chatrooms, live in 10 minutes. Doses and when to scale back: the reference card.
Case 2 · Pilot
Participant 2 — ten days, fully remote
- Device
- Android
- Setting
- home desk, two monitors
- Flood
- chatrooms (days 1–4) → bot (days 9–10)
- Coaching
- remote, over chat, asynchronous
- Structure
- 25-min pomodoros → 1-hour blocks
- Data on
- days 1, 2, 3, 4, 9, 10
- Scale
- /10 on day 1, /7 after
- Date
- Apr–May 2025
The longitudinal case: the only participant we can watch across ten days, including what happens after a four-day break. He was never in the same room as the coach — everything below happened over chat, which makes this the closest case to what you reading this website will do.
— focus — restraint — aversiveness
Read the restraint line left to right: settled at 1/7 by day 3 — then four days completely off — back up to 4/7 on return (the reflex partially regrows) — and back down to 1 within a session. Training fades without use and reinstalls at a discount. Day 10's middle session is the flash-only introduction: the screen-flash cue briefly costs more (aversiveness 3), then fades by the next hour — the same transfer spike Case 1 hit.
And day 10, session 1 hides this case's deception probe, unplanned: the bot was down and no flood arrived — he studied at focus 6, restraint 1 anyway, and only realized afterward. Same result as Case 3's deliberate version.
Day 1, before the instrument settled (scale /10)
| Session | Focus | Restraint | Work quality | PCT |
|---|---|---|---|---|
| Pomodoro 1 — first ever | 5 | 4 | 4 | 3 |
| Pomodoro 2 | 7 | 7 | 8 | 0 |
| Pomodoro 3 | 9 | 5 | 8 | 0 |
| Pomodoro 4 | 6 | 7 | 6 | 0 |
| Pomodoro 5 | 7 | 4 | 7 | 0 |
Phone-conscious thought hit 0 on his second-ever session and never came back. From his day-1 log, first session: "i looked at the phone a bit as the default reaction when the first notification showed up but i immediately told myself to ignore." Four sessions later: "i am very happy and kinda surprised with the result."
Coach, day 5: "Do you feel locked in now?" — Participant 2: "i do"
Two confounds, disclosed: his vibration was on for roughly the first four days before anyone noticed (vibration is explicitly against protocol — it adds difficulty with no training value), and the scale changed from /10 to /7 after day 1, which is why day 1 sits in its own table instead of the chart. Also: his day-9 and day-10 sessions were 1-hour blocks, not pomodoros — durations aren't constant across the line above. Remote and unsupervised means no one verified he never looked; his slip reports are on his honor.
Run this variation: start with Version A today, move to Version B — your own bot like he did. If you take days off, expect the day-9 bump and don't read it as failure.
Case 3 · Experiment
Participant 3 — coached cold, in public, in one sitting
- Device
- Android, all sounds standardized
- Setting
- crowded public university — deliberately loud
- Flood
- Telegram bot only
- Coaching
- in person, 10–20 minutes, scripted
- Structure
- 4 trials in ~2 hours: ringer · ringer · silent · deception
- Filmed
- entire session + 20-min documentary after
- Scale
- /7
- Date
- May 2025
The first real test of whether the method survives being taught: a participant with no preparation, handed the rules and a 4-minute observation video, in the loudest environment available. Total instruction time: under twenty minutes. He never looked at his phone once — including on trial one.
— focus — restraint — aversiveness — PCT (dashed)
Focus never moved off 7/7. Everything else — the effort, the annoyance, the phone in his head — went to zero inside the sitting. Trial 4 is the deliberate deception probe: the flood was secretly off, and his numbers didn't change. By then it was never the flood; it was him.
| Trial | Focus | Restraint | Study quality | Aversiveness | PCT |
|---|---|---|---|---|---|
| 1 — ringer, first ever | 7 | 4 | 6 | 3 | 3 |
| 2 — ringer, +30 min | 7 | 1 | 7 | 0 | 0 |
| 3 — silent | 7 | 0 | 7 | 0 | 0 |
| 4 — deception | 7 | 0 | 6 | 0 | 0 |
His own answers on the report sheet, trial by trial: "Did you forget about your phone?" — trial 1: "a little bit" · trial 2: "yep" · trial 3: "yes" · trial 4: "Yes." And the line that carried past the session:
"its more easier to focus not only when studying, but anything I put my mind into it. Even though my mind races, its easier now to focus on one particular thing no matter the distractions around me." — Participant 3, next-day report
Follow-up over the next month: reading and work output sustained. The coach's field note from the session records the counterintuitive finding that shaped the method's environment rule: the crowd helped — with many competing stimuli, no single distraction could dominate.
The coach was physically present the whole session — expectancy and social pressure are baked into these numbers ("I will try to get your attention" was literally part of the protocol). The deception probe is n = 1. And one accidental discovery worth stating: Android forced every app's notification onto one identical sound, and standardized pings turned out noticeably easier to untrain against than a mix of meaningful and meaningless sounds — that limitation became protocol.
Run this variation: this one takes two people. The full coached procedure — the rules speech, the observation video step, the train-of-thought script, the deception finale — is written out as the coached variation on the Replicate page. Teach a friend in twenty minutes.
Case 4 · Experiment
Participant 4 — the hardest variant, the cleanest curve
- Device
- iPhone
- Setting
- home, alone, remote
- Flood
- custom app — flash-only from the start, fully silent
- Dose
- 60 notifications / 40 min
- Structure
- one continuous 90-min sitting, reports at 30/60/90
- Baseline
- self-described severe phone addiction
- Scale
- /7
- Date
- August 2025
Every other case got sound. Participant 4 got nothing but a silently flashing screen in her peripheral vision — the subtlest cue, the hardest to habituate, chosen precisely because she rated her own phone addiction "so bad." No coach in the room. One sitting. Three reports:
— focus — restraint — aversiveness — PCT (dashed)
| Checkpoint | Focus | Restraint | Aversiveness | Environment | Study quality | PCT | Slip-ups |
|---|---|---|---|---|---|---|---|
| 30 min | 3 | 7 | 3 | 5 | 4 | 5 | ~12 |
| 60 min | 5 | 4 | 2 | 3 | 5 | 3 | 5 |
| 90 min | 6 | 2 | 1 | 3 | 6 | 2 | 2 |
Her three reports, in her own words. At 30 minutes, maximum restraint, twelve slips: "the first 30 mins were mostly me mentally challenging myself to resist the urge to look." At 60: "my mindset was kinda like — idc if its my friends or the app, i need to get ts done." At 90:
"I WAS LOWKEY SURPRISED AT MYSELF LMAO and i didnt really think about my phone surprisingly i js got used to the flashing of the screen … I FINALLY GOT THIS SHIT ASSIGNMENT DONE (which was my goal)" — Participant 4, 90-minute report
Then the part no one designed. She turned the app off — and her phone kept flashing, real notifications from real friends — and she kept working through those too, finished the last 20 minutes of her assignment, and replied afterward:
"i literally continued working and my phone was flashing with notifs from my friends and i didn't look … i actually locked in and got it done" — Participant 4, after the session
That's the transfer working on the first day: the training was fake pings; the victory was real ones.
One participant, one sitting, no follow-up data yet — this case is a single clean curve, not a durability claim (Case 2 covers what fading looks like). She listened to music at the start, which the protocol allows for flash-only but which softens the first checkpoint. And the app she used isn't public: to run her exact condition, use the bot with your phone on silent, flash-only, ~60/hr — functionally identical.
Run this variation: phone on silent, notifications wake the screen, phone in peripheral vision, ~60 pings/hr, one honest report every 30 minutes. If your addiction feels bottomless — this is the case that says start anyway.
The Living Dataset
Everything above is the founding four. Below is everyone since — every approved self-report, anonymised to a participant number and a timestamp, unedited and uneditable. This table is the experiment now.
No public replications yet. Participant 6 is an empty row with your name deliberately not on it — post your numbers.
What this page is for
Not persuasion — calibration. When you run your first session tonight and your restraint reads 7/7 with a dozen slips, this page is how you know you're exactly on schedule: that's Participant 4's 30-minute mark. The founding four drew the curve. You're the replication.