Why is infinite scroll so hard to stop? Because stopping was engineered out of it. There is no ending to arrive at, the payoff comes on an unpredictable schedule so you cannot tell when you have had it, continuing costs nothing, and the next item begins before the current one finishes. None of those is about your self-control.
Four mechanisms, in order of how much they matter.
One: the stopping cue was removed
Every older format has an end built in. Chapters end, episodes end, albums end, a newspaper runs out of pages.
Those endings are not incidental. They are the moment where continuing becomes a decision. You have to actively start the next thing, and that small gap is where most people leave.
A feed has none. It does not run out, there is no page count and there is no last item. So the decision to stop has to be generated by you, from nothing, with no prompt, repeatedly, for as long as you are there. That is a much harder task than anyone gives it credit for, and it is the single biggest reason feeds hold people.
The version of this you actually experience is the sentence in just one more video.
Two: the payoff is unpredictable
If every item were good, you would get your fill and leave. If every item were bad, you would leave immediately.
What holds attention is neither. It is a mostly mediocre stream with an unpredictable good one in it. That is a variable ratio schedule, and it is the most persistent reinforcement pattern known. It is what makes slot machines work and it works for the same reason here: because you cannot predict when the payoff comes, there is no point at which leaving feels correct.
Nothing sinister is required for this to emerge. Any system optimising for time spent will converge on it, because it is what optimising for time spent looks like.
The felt experience is being about to leave and then seeing something good, which happens often enough to feel like coincidence and is not.
Three: continuing is free
The cost of the next item is one thumb movement of about two centimetres. Not a click, not a decision, not a load screen.
When the cost of continuing is effectively zero and the cost of stopping is a small act of will, the ratio decides the outcome. Stopping has to beat continuing hundreds of times in a session. Continuing only has to win once.
This is also why friction placed at the entrance works better than friction placed inside a session. Inside, you are competing with a two centimetre thumb movement. At the entrance, before momentum exists, a few seconds is enough to matter.
Four: the next item starts before this one ends
Autoplay, and the preview of the next item already visible under your thumb.
The effect is that there is never a moment of nothing. The seam between items, which is where an exit would live, has been paved over. Short video is the strongest version, where the next clip is often audible before the last has finished.
Take those four together and you have a format with no ending, no predictable satisfaction point, no cost to continue and no gaps. Then it is handed to a tired person at midnight.
What this means for fixing it
Two things follow directly.
Intent is the wrong instrument. You are being asked to win a repeated decision hundreds of times while the other side only has to win once. Any approach that depends on a perfect record is going to lose, and losing then gets read as a character flaw.
Put the seam back. The format removed the gap where a decision could happen, so any effective intervention has to reinstate one. Not a wall, which is a different thing and decays in the way described in why your app blocker stopped working. A gap.
The cheapest place to put a gap is the entrance, before momentum exists. That is what reCalm does: a shield over the apps you choose, one breath, then the app opens. Always. Six seconds is not much against a well-built feed, and it is the difference between an open you noticed and an open you did not. The category is described in what a scroll-breaker is.
Where the state comes in
The four mechanisms explain why leaving is hard. They do not explain why you arrived.
That part is usually a body that will not settle. Tired and wired at 1am, stuck at 10:04, alone in a quiet room. The feed is very good at holding whoever shows up, but something has to bring you to the door. Slowing the exhale addresses that half directly, and there is real physiology behind it, covered in does breathwork actually work.
Both halves matter. Deal only with the state and you still get held once you are in. Deal only with the entrance and you keep arriving.
The counter-design is the same trick in reverse: put an edge back where the feed removed one. The sizing of that edge is worked out in what should happen when your phone interrupts you.
Frequently Asked Questions
Is infinite scroll designed to be addictive on purpose?
Intent is hard to establish and mostly beside the point. A system optimised for time spent arrives at these four properties whether or not anyone set out to. The result is the same either way.
Why do I keep scrolling things I do not enjoy?
Enjoyment is not the mechanism. Unpredictable payoff holds attention regardless of whether the average item is any good, which is why people can be bored and still there.
Does turning off autoplay help?
It restores one of the four seams, so it does something, particularly on video. It leaves the other three intact.
Is short form worse than long form?
For this specific problem, generally yes. Shorter items mean more removed decision points per minute and a faster reward schedule.