Prevention ArchitectureA prepared environment can reduce the number of times the same temptation needs to be reconsidered.

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.

Each act of resistance takes some willpower, and another is due the next time the donut is noticed. Prevention is a single act, done before any of those moments arrive. Removing the recurring decision keeps the acts of resistance from accumulating.

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.

Counting the Decisions

Someone who sees the donut 20 times in a day faces 20 separate decisions about it. The donut is still there in the evening only if all 20 go the same way, and the later ones are made by a person who has already refused many times that day. It takes one decision going the other way for the donut to be eaten.

Not buying the donuts removes all 20. There is nothing to see or decide about. Both approaches aim at not eating the donut, but one requires repeated inhibition while the other changes the conditions that make inhibition necessary.

Resisting each time, or removing it onceInteractive example

Times a day it catches your attention

20

Resisting 20 times a day is 140 separate acts of resistance in a week. Keeping it out of the room is one decision.

Why Resistance Fails

A day of refusing the same thing wears a person down. Each refusal is made with whatever the earlier ones have left, and the series has to break only once.

Giving in at 8 PM comes at the end of a day of such refusals. Self-control tires with use in the way a muscle does, so every added occasion for resistance leaves less for the occasions after it.

High Self-Control People Avoid Situations Requiring Self-Control

People who appear to have high self-control are often people who have arranged to need little of it. They keep no junk food in the house, and they never install 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. A work session contains many of them, and it takes only one decided the other way for the phone to be checked.

A locked drawer in another room prevents notifications from reaching the work session. Checking now requires standing up, walking, unlocking the drawer, and retrieving the phone, and none of those steps happens by reflex. During work there is nothing to resist.

Food Environment

Snacks bought "for guests" still remain in the pantry. They have to be passed over every time the pantry is opened, and passing them over has to work every time.

If the snacks are not bought, there is nothing in the pantry to resist. A person cannot eat food that is absent.

Digital Distractions

An open Netflix tab kept "just in case" leaves the option immediately available. Each time attention drifts to the tab, clicking has to be decided against again.

A website blocker active during work hours makes Netflix inaccessible for those hours. With nothing to click, there is no decision to repeat.

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, the Boltzmann distribution says that the lower a configuration's energy, the more often a system is found in it. Behavior can be pictured the same way: of the actions available at a given moment, the one with the lowest activation energy, meaning the least effort to start, is the one that most often happens.

With the phone on the desk, checking it takes a glance and a reach, while leaving it alone takes an act of resistance. Checking is the easier of the two, and the picture predicts checking. With the phone in a drawer in another room, fetching it takes a walk and a key, while continuing already-started work takes very little. The picture then predicts continued work.

Changing the arrangement changes the relative costs for the same person, whose discipline is the same in both cases. That is as far as the comparison goes. The effort of walking to a drawer is not a physical energy, so the picture gives an ordering of the options and no numbers for them.

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 no longer has to be defended against them.

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. One example is a rule against outside breakfast: it eliminates the 90-minute deliberation over dim sum, because the morning no longer has a food decision in it.

Another is an eating window that closes at 2 PM, which removes eating from the options after 2 PM. Dinner and snacks no longer require repeated adjudication in the evening, because the question was settled when the window was set.

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 puts more effort into each refusal and leaves the number of refusals unchanged, so the series can still break at any one of them. Prevention changes the environment so that the recurring resistance is unnecessary, and an arrangement that asks for nothing each day can stay in place.

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 20 moments of resistance a day for a year, which is 20 × 365 = 7,300 separate refusals. 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. Many everyday temptations do meet all three conditions, the phone, the pantry and the open tab among them, and for those prevention is the first thing to try.

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, and the change carries forward through the later links. That is a better place to act than the final eating decision, which arrives late at night when the least willpower is left.

"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 take those decisions over, so that after the one-time setup they no longer have to be made by hand.

Related Concepts

• 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.