The Compulsion Loop

The loop

The pattern is simple enough to remember: Trigger -> Reach -> Open -> Consume -> Drift -> Regret -> Repeat. Each stage has its own feeling. The trigger is often discomfort or empty space. The reach is quick and familiar. Opening creates immediate possibility. Consumption supplies novelty. Drift is the loss of the original intention. Regret arrives when the session becomes visible.

The loop becomes powerful because no single step feels dramatic. You do not experience a major decision. You experience a chain of tiny permissions.

Common triggers

Boredom, difficult work, task transitions, fatigue, notifications, waiting, loneliness, conflict, environmental cues, and unclear next steps all create openings. The trigger does not have to be intense. It only has to make the phone or browser feel like the easiest next movement.

Many people try to intervene at the regret stage. That is the most expensive point, because the session has already consumed energy and time. Move the defense earlier. Put the phone across the room before uncertain work. Close the feed before the transition. Define the next action before the break ends.

Interruptions that work

The useful question is not, 'How do I become someone who never wants to check?' The useful question is, 'Where can I make the next automatic step less automatic?' A delay, a logout, a removed shortcut, a blocked time window, or a written next action can be enough to bring choice back into the loop.

The loop hides inside reasonable actions

The first action in the loop often has a reasonable costume. You check the message because someone might need you. You open the browser because you need one answer. You pick up the phone because the task is paused anyway. The problem is not that the first action is always irrational. The problem is that it hands control to a surface designed to keep offering next actions.

That is why the loop should be mapped with honesty rather than contempt. If the trigger is uncertainty, the intervention is a clearer next task. If the trigger is fatigue, the intervention may be movement, food, or shutdown. If the trigger is a notification, the intervention is notification design. Different triggers require different defenses.

Intervene by stage

At the trigger stage, write the next physical action before the pause opens. At the reach stage, move the device outside immediate reach. At the open stage, require a written purpose. At the consume stage, use a visible endpoint. At the drift stage, close the input without trying to understand the whole day. At the regret stage, use the recovery protocol.

The earlier stage is usually cheaper. A phone in another room requires less inner argument than a feed already open at midnight.

Smallest useful first step

For one high-risk loop, write a single sentence beginning with 'When I notice...' Keep it physical. 'When I notice I am reaching during a task transition, I put the phone on the kitchen counter and write the next action on paper.'

A good interruption is boring, visible, and repeatable. It should not require inspiration.

Field application

Map one real episode in plain language. Do not write the noble version or the self-attacking version. Write the sequence as if you were documenting a system: what time it was, what task came before it, what feeling or cue appeared, what you opened, how the session extended, and what finally made it visible.

Most people discover that the regret stage is loud but late. By the time regret arrives, the loop has already received enough time and novelty to become sticky. Earlier stages are quieter. A notification, a pause before hard work, a browser left open, or a phone beside the bed may not feel like the problem, but those are often the parts that can be changed.

Choose the earliest repeatable intervention. If the trigger is uncertain work, define the next action. If the reach is automatic, change the device location. If opening is the problem, require a written purpose. If consumption extends, add a stopping condition. If regret appears, use recovery instead of self-attack.

• Best intervention: before the app is open.
• Second best: before the session loses its endpoint.
• Always available: interrupt the next ten minutes.