AtlasLibrary
Browse articles

131 articles

Change whether the decision comes up

Prevention Architecture

Read the articleMarkdown
A hand closes a drawer on a phone in one room, away from a worktable visible across the hall.
Moving a cue changes what the work setting asks you to resist.

What It Is

A donut on the desk creates a decision every time it attracts attention. If there is no donut in the room, that decision never arises. Prevention architecture changes the environment so that a response does not need to be resisted repeatedly.

The cost distinction is 0 willpower units for prevention versus 2-3 units for each act of resistance. Removing the recurring decision prevents those costs from accumulating.

A closer look

Repeated resistance and changed access

Repeated resistance and changed accessCompare Resist at every encounter; Change the encounter in advance. The relevant question is what the arrangement actually removes or changes, including any route around it.Resist at everyencounterChange theencounter inadvanceRepeated resistance and changed accessCompare Resist at every encounter; Change the encounter in advance. The relevant question is what the arrangement actually removes or changes, including any route around it.Resist at every encounterChange the encounter inadvance

The relevant question is what the arrangement actually removes or changes, including any route around it.

Read this diagram

Compare Resist at every encounter; Change the encounter in advance.

The Math

Someone who sees the donut 20 times in a day and spends 2 units resisting each exposure uses 2 units × 20 = 40 willpower units/day. Eventually the available units run out and the person gives in.

Not buying the donuts removes the exposures. There is nothing to see or decide about, so the cost is 0 willpower units/day. Both approaches aim at not eating the donut, but one requires repeated inhibition while the other changes the conditions that make inhibition necessary.

Why Resistance Fails

Each act of resistance reduces the capacity available for the next one. The prefrontal cortex depends on glucose and neurotransmitters, and repeated inhibition tasks deplete its metabolic resources.

Giving in at 8 PM can therefore follow a day of repeatedly using that capacity. Self-control research describes a resource that fatigues with use, like a muscle. Increasing the number of occasions requiring resistance increases the demands on that resource.

High Self-Control People Avoid Situations Requiring Self-Control

People with high self-control avoid keeping junk food in the house and avoid installing the social-media apps they would otherwise check. Their environment removes occasions for temptation before resistance becomes necessary.

The difference is in how the situation is arranged, rather than greater strength on every encounter. Someone who never installs the app does not need to repeatedly ignore its notifications.

Examples of Prevention Architecture

Phone Management

A phone on the desk remains available during work. Each noticed notification requires another decision not to check, costing 2-3 units. Repeated exposure makes the failure rate high: eventually the phone is checked.

A locked drawer in another room prevents notifications from reaching the work session and puts retrieval behind a 4-unit cost. Checking requires standing up, walking, unlocking the drawer, and retrieving the phone. The cost during work is 0 units, and the failure rate is near zero.

Food Environment

Snacks bought "for guests" still remain in the pantry. Resisting them costs 2-3 units per exposure and has a high failure rate.

If the snacks are not bought, there is nothing in the pantry to resist. The cost is 0 units and the failure rate is zero, because a person cannot eat food that is absent.

Digital Distractions

An open Netflix tab kept "just in case" leaves the option immediately available. Each temptation to click costs 2-3 units to resist, with a moderate failure rate.

A website blocker active during work hours makes Netflix inaccessible. The resistance cost becomes 0 units and the failure rate becomes zero, because the blocked site cannot be accessed.

State Machine Perspective

With a visible phone, the combination of work and the cue has a high probability of producing distraction:

  • In the old arrangement, work_state plus a visible phone leads to distraction_state, unless the person actively resists the transition.
  • In the engineered arrangement, work_state plus an absent phone has no transition to distraction_state through that cue.

Removing the trigger removes the transition from the decision tree. The person does not have to inhibit an option that is no longer available.

The Boltzmann Distribution

In statistical mechanics, systems settle into low-energy configurations. Applied to behavior, the action with the lowest activation energy is the one that executes.

A phone on the desk costs 0.1 units to check and 2 units to resist. The Boltzmann account therefore predicts checking. A phone in the drawer costs 4 units to retrieve, while continuing already-started work costs 0.5 units. The account then predicts continued work.

Changing the arrangement changes the relative costs for the same person. The explanation is thermodynamic rather than a change in the person's discipline.

Application Domains

Work Environment

Removing the phone from the workspace, blocking distracting sites at the router or application level, and closing unnecessary tabs eliminates competing cues. Visual clutter can be removed for the same reason. With the competing sources of attention absent, focus becomes the default.

Food System

Junk food can be kept out of the house, delivery apps removed from the phone, and meals decided in advance. These changes remove both access to temptations and recurring decisions about what to eat.

Social Media

Deleting an app removes access more completely than logging out. Browser extensions can block the associated sites, and accounts can be deactivated when necessary. Each mechanism makes a route to checking unavailable.

Time-Based Prevention

A precommitted time rule can remove a recurring decision. A rule against outside breakfast eliminates the 90-minute deliberation over dim sum and leaves zero willpower spent on that morning food decision.

An eating window that closes at 2 PM removes eating from the options after 2 PM. Dinner and snacks no longer require repeated adjudication, leaving zero willpower spent resisting evening eating.

Without a Rule, the Judge Always Says Yes

An invitation to dinner presents immediate benefits: a friend's excitement, social warmth, and food. The costs are distributed across the month's spending, diet, and body. Those costs are absent from the situation in which the individual invitation is being judged.

"A judge ruling on present evidence will always say yes. This isn't weakness — it's what any judge would rule given that evidence set."

After another expensive dinner out that he mostly did not want, Will described the conflict with his standing goals for the gym, diet, and less eating out:

"I don't have any rule. So it's always like, I need a judge, and my judge is always saying, sure why not."

The failure concerns the boundary around the decision. The preference is already known, but each invitation is reconsidered using mood, social pressure, and immediate pleasantness. Each dinner is judged as an isolated instance, "and the macrostate never gets representation in court."

A precommitted policy lets the decision be made earlier, with the month's aggregate information available through tracking. "The rule is the trend line speaking." An exception budget then reserves fresh judgment for genuine edge cases, so discretion is limited rather than eliminated.

Under this diagnosis, "sure, why not" indicates a missing rule rather than a newly revealed preference. The rule removes the recurring case instead of demanding a stronger judge every time an invitation arrives.

A known policy can also reduce interpersonal conflict. The other person can know the boundary in advance, so the interaction no longer depends on one person policing the other or the other person defying them.

"A judge with no law rules on vibes, and vibes always vote for tonight."

Why This Works When Trying Harder Doesn't

Trying harder spends more of a finite resource on inhibition, which eventually fails through depletion. Prevention changes the environment so that the recurring resistance is unnecessary and the arrangement can remain sustainable indefinitely.

The intervention removes the environmental source of the impulse. It does not require a new display of moral effort at each exposure.

The Planning Phase

Prevention requires initial work. Temptation sources have to be identified and removed, blocking mechanisms installed, and systems established to maintain the arrangement.

That one-time investment eliminates the corresponding daily resistance. The comparison is 1 hour spent setting up website blockers against a saving of 2-3 units × 20 instances × 365 days = 14,600+ willpower units/year. The initial setup can therefore repay its cost across repeated avoided decisions.

Limitations

Prevention requires an opportunity to change the conditions:

  • The temptation must be discretely removable.
  • The person must control the relevant part of the environment.
  • Exposure must not be socially required in a way that cannot be avoided without consequences.

When temptation is unavoidable, the environment is outside the person's control, or the social context requires exposure, prevention fails. Habit building, reframing, and other strategies then become necessary. The estimate here is that prevention remains available and optimal in 80% of cases.

Intervention Design Questions

Intervention design questions search for a mechanism that can prevent a failure. "How do I have more willpower?" treats willpower as an input to increase. "What forcing function prevents this?" instead searches for an implementation that changes the situation.

"What's the earliest intervention point?" follows the causal sequence backward. Late-night eating might follow hunger, boredom, a lounge state, missing the gym, waking late, and a sleep time of 1am. Changing sleep time intervenes at the earliest modifiable point, with effects that propagate forward and prevent the entire chain. It is computationally superior to waiting until the final eating decision, when willpower is already depleted.

"What category of forcing function applies?" identifies which kind of intervention fits. Accessibility changes remove the object from the environment; temporal prevention limits its availability; commitment devices supply external accountability; environmental redesign changes the default options. Classifying the problem narrows the search to relevant mechanisms.

"How do I make this automatic?" identifies the decisions still requiring manual resistance. A default script or prevention arrangement can remove their repeated 2-3-unit costs and reduce the cost to 0 units after the one-time setup.

  • Willpower is the resource conserved by removing recurring resistance.
  • Self-Control uses prevention as a primary strategy.
  • Discipline describes how successful prevention appears from the outside.
  • State Machines represents the removal of a transition.
  • Activation Energy describes how prevention raises the cost of an unwanted action.
  • Focus becomes easier when distracting cues are absent.
  • Question Theory supplies questions that search for forcing functions.
  • Commitment provides precommitted policy at the decision level.
  • Tracking supplies the aggregate information represented by a rule.
  • Epistemic Contamination applies prevention to information exposure by quarantining sources before their influence has to be resisted.

Key Principle

Changing what is available can remove a decision that would otherwise recur. The useful intervention depends on the source of the choice: an accessible object, an open website, or an invitation being judged without a standing policy.

Return to the libraryBack to the beginning