Infinite Scroll: The Interface Built to Have No Stopping Point

Every page used to end. You read to the bottom, and the bottom asked you a question: do you want more? Maybe the question looked like a button marked “Next,” or a row of little page numbers, or the last line of a chapter. Whatever its form, it was a real question, and answering it took a small act of will. Infinite scroll deleted that question. It replaced the end of the page with more page, arriving automatically, forever, and in doing so it quietly removed the most ordinary moment in the history of reading: the moment where a person notices they have finished and decides whether to continue. I want to write about that specific piece of interface design — where it came from, what it actually does to a decision, and why a missing button turned out to matter more than almost anything else on the screen.

What the pattern actually is

Mechanically, infinite scroll is unglamorous. As you approach the bottom of a list, the software fetches the next batch of items in the background and appends them to what you are already looking at. The scrollbar, if you can still see one, shrinks a little and resets. Nothing announces itself. There is no flash of a new page loading, no URL change, no pause. The content just keeps going.

The thing it replaced is pagination: content broken into numbered chunks, each one a separate page you had to request. Pagination had a cost you could feel. You clicked, the screen went white, you waited, the new page arrived, and you often lost your place. On a phone, over a patchy connection, that was genuinely miserable. Pagination also had a benefit nobody thought to name, because it was invisible until it was gone. The end of a page was a boundary. Boundaries are where people make decisions.

The designer usually credited with inventing the auto-loading feed is Aza Raskin, who built it around 2006 — a prototype that simply kept the results coming instead of making you click through them. He is the son of Jef Raskin, who started the Macintosh project at Apple, so he came from a tradition that treated removing friction as the whole job of interface design. To be fair to the history, he is not the only claimant: a Microsoft team filed a patent covering similar continuous-scrolling behaviour in the same period, and versions of the idea occurred to more than one person at once, as usually happens with ideas. But Raskin popularised it, showed it around, and it spread from prototype to near-universal default in under a decade.

He has since said, publicly and repeatedly, that he regrets it. Not in a coy way — he has testified about it, spoken about it in interviews, and helped build an organisation devoted to arguing against the design patterns he helped normalise. He has offered rough estimates of how much human time the pattern consumes, framed in units of entire human lifetimes per day or per month. Those figures vary between tellings and should be read as back-of-envelope illustrations rather than measurements; nobody has a reliable global scroll-hours dataset. The point of them is rhetorical, and the rhetoric is his own: he thinks he made something whose costs he did not see.

The reason he did not see them is worth sitting with, because it is not stupidity. It is the ordinary blindness of a designer solving the problem in front of him. The problem in front of him was clicking “next” is annoying. That is a true problem. The solution was elegant. What he could not see from inside the problem was what the solution would be worth to companies whose revenue was a function of time-on-screen — because in 2006 that industry barely existed yet.

Pagination had a cost you could feel and a benefit nobody named. The click was friction. The bottom of the page was a decision.

Why removing a stopping cue is so powerful

Here is the part I find genuinely interesting, and it has nothing to do with willpower.

Human beings are not good at continuously evaluating whether to keep doing something. We are much better at evaluating at breakpoints. Think about how you actually stop doing anything: you finish the chapter, the episode ends, the song runs out, the plate is empty, the road reaches a junction. These are all moments where the activity structurally pauses and hands control back to you, and it is at exactly those moments that you check in with yourself and ask whether you want more. Psychologists have studied this in contexts far removed from phones — the classic demonstration involves soup bowls that secretly refill, and people eating substantially more from them without feeling any fuller, because the cue they were unconsciously using to decide they were done was the bowl is empty, not an internal signal.

Infinite scroll is a self-refilling bowl. It never hands control back. The activity has no structural pause at which the question “shall I stop?” naturally arises, so the question has to be generated entirely from inside you, against the pull of whatever is currently on screen, with no help from the environment at all. That is a much harder cognitive task than clicking or not clicking a button, and we ask it of ourselves dozens of times a day without noticing the difficulty.

Layer on top of this the fact that the feed is not a fixed sequence. What loads next is unpredictable — mostly dull, occasionally excellent. That intermittent, variable payoff is the most reliably compelling reward schedule we know of, which is why it shows up in slot machines and why it shows up here. I won’t belabour the neuroscience, because that is its own argument and I have made it at length in the piece on doomscrolling. The narrow claim here is about the interaction between the two things. Variable rewards make you want the next item. The absence of a stopping cue means nothing ever asks whether you still want it. Either one alone is manageable. Together they produce a session that ends not when you decide it should, but when something external interrupts you — a phone call, a dead battery, someone saying your name.

Notice the tell: almost nobody remembers deciding to stop scrolling. Ask yourself how your last long session ended. If you can’t recall a decision, that is not a personal failing. That is the design working exactly as designed.

Where it lives now

Everywhere, essentially. The vertical video feeds are the purest expression — there, infinite scroll fused with autoplay, so you don’t even have to scroll to receive the next item, only to reject the current one. Social timelines, shopping apps, news aggregators, dating apps, food delivery listings, music recommendations, search results on some platforms, job boards. Comment sections. The pattern escaped social media a long time ago and became simply what a list looks like.

Its spread wasn’t a conspiracy. It was A/B testing. Once you can measure time-on-app, and once your revenue is a function of that number, the version without a stopping point wins the test every single time. No one has to decide to make the product compulsive; they only have to keep shipping the variant with the better metric. The metric does the deciding. That is the mechanism I keep coming back to in writing about how technology gets captured: not villainy, but the steady, impersonal pressure of an incentive that no individual inside the company needs to endorse for it to prevail. The commercial logic underneath it — why your attention is the product being sold — is the subject of the attention economy, and it is that logic, not the scroll bar, that explains why the pattern became universal rather than merely popular.

The honest counter-argument

I don’t want to write a one-sided piece, because the case for infinite scroll is real and I think people who campaign against it are often too quick to wave it away.

On a phone, pagination is bad. Small targets are hard to hit with a thumb. Page loads over mobile data are slow, and each one is a chance to lose your scroll position and your patience. If you are browsing a shop for a pair of shoes, the continuous list is faster, calmer and less annoying than clicking through eleven pages. There is a reason the pattern won on merit before anyone weaponised it: for browsing tasks, where you are scanning for one thing among many, it genuinely is the better interface.

It is also worth saying that pagination was never neutral either. Publishers loved splitting a single article across nine pages precisely because it multiplied their ad impressions — a dark pattern that infinite scroll partly killed. The choice has never been between a manipulative design and a pure one. It has always been between designs with different costs.

The design decides what happens when you do nothing. And what happens when you do nothing is what happens almost all of the time.

So the useful question isn’t “is infinite scroll evil?” It is: what is this list for? A list you are searching should flow continuously; you are going to leave the moment you find the thing. A list designed to be consumed indefinitely, with no destination, is a different object wearing the same clothes. The tragedy is that the interface makes them look identical, and most of us never learn to tell them apart.

Defaults are where the power is

What I take from this history is smaller and more general than a complaint about feeds.

The design decides the default. The default is what happens when the user does nothing — and doing nothing is what people do, overwhelmingly, most of the time. With pagination, the default was stop, and continuing required an action. With infinite scroll, the default is continue, and stopping requires an action. Not one item of content changed. Not one word of the terms of service. Just the direction of the path of least resistance, flipped. And that flip was worth billions, because the population-level behaviour of a few billion people is determined almost entirely by which way that path slopes.

This is why arguments about technology that begin and end with personal responsibility make me impatient. Yes, you can act against a default. People do it constantly — setting timers, greying out their screens, deleting apps from the phone and keeping them on the laptop, all the deliberate practices I find worth defending in the piece on digital minimalism. But every one of those is a private tax you pay to undo a choice someone else made for you, and the tax falls hardest on people with the least slack: the exhausted, the lonely, the young, the unwell. Designing a default that only the well-resourced can resist isn’t neutral. It is a transfer.

Raskin’s regret is decent and I take it seriously, but regret is not a mechanism. The feature is still there, in almost every app on your phone, because it still works and it still pays. What would change it is not designers feeling bad. It is treating a default as what it actually is — a decision made on millions of people’s behalf, by someone they will never meet, about how their evening ends.

The bottom of the page used to ask you a question. Somebody decided you shouldn’t be asked. That is the whole story, and it is worth remembering that it was a decision, not weather — because things that were decided can be decided differently.