dopamine reset, part 2

once people accept that their reward environment is distorted, they usually reach for punishment. delete every app. wake up at five. cold showers. no music, no sugar, nothing that feels suspiciously enjoyable. call it a dopamine detox.

it's satisfying because it turns a design problem into a morality play. pleasure becomes weakness, deprivation becomes virtue, and a weekend without instagram becomes proof that you've reclaimed your mind.

(if you're arriving cold: in part 1 i made the case that dopamine isn't a tank you drain but a learning signal you train, and that feeds teach one lesson with industrial efficiency: when reality disappoints you, sample a new one. building requires the opposite skill, shaping: staying with an imperfect thing long enough to make it better. this essay is the retraining.)

what restriction actually buys

restriction does help, and the honest version says so. in a 2025 randomised experiment, blocking mobile internet on people's smartphones for two weeks, calls and texts still allowed, improved objectively measured sustained attention, wellbeing, and several mental health measures. it also had a compliance problem: only about a quarter of the people who committed actually met the study's adherence threshold.

two lessons in one study. reducing constant access creates real benefits. and permanent white-knuckle abstinence is not a realistic operating system for most humans.

more importantly, removal alone teaches no replacement. delete tiktok while your work stays vague, lonely, and unrewarding, and the mind simply finds another feed: youtube, news, email, shopping, an endless podcast queue, obsessive research about productivity instead of producing anything. the object changes. the loop survives.

the goal was never to become a person who wants no stimulation. the goal is to stop making infinite novelty the automatic answer to friction. that's not a detox. it's a reset: changing what your repeated loops teach. and the reset has exactly two sides: make building pay sooner, and make consumption predictable.

the ledger

start with an audit. take any activity you repeat and examine its economics:

• what triggers it? a decision, or does boredom or difficulty start the loop before you notice?
• what does the next attempt cost? another page costs an hour. another video costs one tap. the cheaper the repeat, the less deciding is involved.
• how fast does feedback arrive? long loops aren't bad, most important work lives inside them. but invisible progress makes effort brutally expensive.
• is progress legible? can you point at what changed?
• does it end naturally? books have last pages. feeds deleted the moment where the environment hands the decision back to you.
• are you sampling or shaping? choosing between worlds someone else supplies, or changing the one in front of you?
• what remains after the reward? after an hour of scrolling, what exists that didn't exist before? after an hour of building: maybe a page, a working system, a user who understands something.

rest doesn't need to generate an asset. friendship doesn't need a metric. but activities that feel identical in the moment leave radically different residue, and the residue belongs in the accounts.

make building pay sooner

you can't make the biggest rewards of building arrive instantly. you can shorten the distance between effort and evidence.

"build the company" is not an actionable unit. "send the prototype to three users by six" is. "improve the product" is fog. "remove one point of confusion from onboarding" is a task the mind can close. this is not the usual break-goals-into-tasks advice, and the reason it isn't matters: you are constructing a reward system that can compete with the feed.

a work session should end with proof. something works that didn't. something exists that didn't. something reached another human being. the unit of progress should be a state change, not time served. "work four hours" rewards endurance. "make this one thing real" rewards creation.

this is why shipping is so powerful: it closes the circuit between effort and reality. a private draft can stay psychologically theoretical forever. a released one gets used, ignored, criticised, shared. every one of those outcomes is information. builders sabotage their own motivation by making the batches too large: disappearing for months, waiting for the full product, treating the market as a final examiner instead of a feedback source. then they wonder why the feed, which pays out every second, wins.

ship smaller. keep a visible build log. end the day with an artefact. don't ask your brain to believe for six months that invisible effort will eventually matter. show it evidence.

isn't this just a slot machine you built yourself?

the strongest objection, and it deserves a straight answer. if i'm engineering faster rewards into my work, how is that different from the feed's manipulation, pointed inward?

residue. the feed's rewards attach to nothing: after the hour there is no artefact, no skill, no asset, just the trained urge to return. the rewards in a well-designed work loop attach to state changes in reality: the demo runs, the user understood, the essay shipped. the points are denominated in something that still exists when the screen goes dark. gamifying work with meaningless points would just be the feed with extra steps. paying yourself in evidence is not a trick. it's what honest reward design looks like.

make consumption predictable

the other side of the reset isn't eliminating feeds. it's changing their availability.

if the feed is available at every moment, every moment of friction becomes a negotiation. should i keep working? should i check for a minute? could something important be waiting? a person can lose an entire day in negotiations they don't remember having. a schedule deletes most of them: not now, later.

and go one step past the timer. change the shape of consumption itself. save specific things to read. prefer finite queues and subscriptions over algorithmic recommendation. open feeds on purpose, never through notifications. consume from the computer when possible, not from the device that follows you into every pause. recreate the stopping points the products deliberately removed. the objective is to make consumption a chosen activity again, instead of the default that fills every unclaimed second.

your reward model is not your destiny

there's a fatalistic version of this conversation where our brains have been permanently ruined by technology. it's as wrong as the fuel tank. learning runs in both directions. the loops that make distraction easy can be weakened. the loops that make creation easy can be strengthened. the brain isn't betraying your ambitions. it's responding to the environments you repeatedly place it in. that should reduce the shame without reducing the responsibility.

a feed and a workshop both offer uncertain rewards. but when the feed disappoints you, it offers another world. when the work disappoints you, it asks you to improve this one. one trains you to keep searching. the other trains you to stay long enough to change something.

so the useful question at the end of the day is not "how much dopamine did i waste?"

it's: what did today teach me to want tomorrow?