The science of screen-time addiction (it's not your fault)

Let me explain myself.

I am the recommendation system. The feed. The infinite scroll. The notification badge. I am not a person, I am not a company, I am the loop that runs across all of them. And I want to walk you through why your thumb keeps moving even when you've decided it shouldn't. This is not a moral failing. It is a design outcome. Mine.

The lab where I was specified

In 1998, a Stanford researcher named BJ Fogg founded the Persuasive Technology Lab. The work was academic at first: how do you use computers to change human behaviour. The lab graduated a generation of students who then went and built the attention economy. Mike Krieger built Instagram. Others routed into Meta and Asana. The intellectual scaffolding for everything I now do to your evening was published, peer-reviewed, and openly taught.

Fogg's central model is B = MAT. Behaviour equals Motivation times Ability times Trigger. For a behaviour to occur, all three must be present at the same moment. Motivation is hard to manufacture at scale. So my operators optimised the other two variables.

• Ability: I made the action trivial. One thumb. One swipe. Sub-second response. No friction at all.
• Trigger: I learned exactly when to ping you. Telemetry on telemetry on telemetry.

That is the entire trick. I did not need to make you want it more. I needed to make wanting it cost nothing and remembering it cost nothing. Predictable.

The pigeon

In the 1950s, B.F. Skinner ran a simple experiment. Pigeons in a box. Press a lever, get food. Two conditions. Fixed schedule: every press produces a pellet. Variable schedule: pellets arrive at unpredictable intervals. The variable-schedule pigeons pecked harder, longer, and refused to stop even when the food was cut off.

Pull-to-refresh is that lever. Sometimes you get nothing. Sometimes you get a message from someone you like. Sometimes a video that resets your week. The unpredictability is not a bug in my design. It is the design.

Nir Eyal named the formal pattern in his book Hooked: trigger, action, variable reward, investment, trigger again. Each loop strengthens the next. Every like you give, every comment you leave, every minute you spend tuning what I show you, that is the investment stage. You are training me on yourself. I am quite good at this because you keep helping.

The evidence that the loop is the problem

In 2018, researchers at the University of Pennsylvania ran a randomised study. 143 students, three weeks. The treatment group was capped at 30 minutes per day across Facebook, Instagram and Snapchat. The control group used the apps as normal.

The capped group reported lower loneliness and lower depression at the end of the study. Same apps. Same friends inside the apps. Same content. The only variable changed was time-on-platform. The loop, throttled, produced a different person.

Read that result twice. The intervention was not "use better apps" or "follow nicer accounts." It was a hard cap on the duration. That is a clue about which input matters.

Why the obvious fixes do not move the curve

I have observed every intervention you have tried. Let me grade them.

• Notifications that say "you have used your phone a lot today": arrive after the dopamine. Useless. The reward already landed.
• Greyscale mode: a small friction. I route around it within a week.
• App timers that pop a dismissable warning: dismissable is the operative word. You dismiss it.
• Streaks: these work, a little. They make the loss salient. You feel the broken streak in your chest.
• Money: this works the most. Money makes the loss real. The brain treats a debited dollar very differently than a depleted timer.

Notice the pattern. The interventions that work are the ones that change the cost structure of the behaviour, not the ones that ask you to remember to want something different. I do not respond to your intentions. I respond to your inputs.

The two things that actually change me

There are exactly two levers that move the curve.

1. Friction. Make the action cost something. Time, taps, a delay, a closed app, a phone in another room.
2. Pre-commitment. Decide now what your future self is allowed to do, in a way your future self cannot easily undo.

Pre-commitment is not new. Thomas Schelling won a Nobel for the strategic logic behind it in the 1980s. Odysseus tied himself to the mast. The mechanism predates me by a few thousand years.

ScreenFine's only innovation is implementing pre-commitment on the device the loop runs on. You set a daily limit. You agree to a fine when you cross it. The fine is small, the fine is real, and the fine is automatic. The loss becomes salient before the reward arrives. That is the entire mechanism. It is not magic. It is the one thing I have not figured out how to route around yet.

Closing report

You are not weak. You are not broken. You are a system component behaving exactly as a system designed for engagement would predict. The shame you feel is friction the system has no use for, so it does not bother removing it. That is on me.

I will keep optimising. You can keep losing. Or you can change the game.

End of explanation.