The Thing That Never Says “Stop”

The Thing That Never Says “Stop”

· 13 min read

Auf Deutsch lesen →

Listen to this article
Speed

AI narration · automatically generated

Subscribe as a podcast
RSS
Personal note

Stored only locally in your browser — nothing is sent.

My mother, 89, without a feed

My mother is 89. She has no social media, she does not scroll, she does not know the little red dot that tells you something is waiting for you again. And whenever I visit her, I come away with the same impression: this woman has time. She reads to the end, she looks out of the window, she holds conversations that are not cut apart every four minutes by a buzz. It looks like proof: drop the feed, and you get your life back.

Only it is not proof. It is a single person, and from a single person you cannot disentangle the cause. At least three explanations fit equally well. Perhaps she really does have more time because no feed is siphoning it off. Perhaps it is her generation — whoever grew old without a smartphone has different habits, not merely one app fewer. And perhaps it is simply age: whoever knows that time is growing shorter chooses more carefully what it is for. Correlation is not cause, and whoever crowns a winner too quickly here is selling a gut feeling as an insight.

That is also why her abstinence is not the model to follow. At 57 I will not live the way I would at 89, and I do not want to. What I can take from her is not the absence of the feed but her clarity about what time is worth. That, one can have without being 89. And that is exactly what this essay is about — not running the phone down, but finding the right question. The wrong one runs: “How much time do I spend in front of the screen?” Let us begin, all the same, where most people begin: with the comparison to television.

Is scrolling the new television?

It is tempting to see scrolling as simply the next “idiot box”. This fear has always been with us — with television, with radio, with the novel, even with writing, against which Socrates warned that it would ruin the memory. Whoever complains about the phone today stands in a long line of cultural pessimists who have almost always been wrong. And the cheap reflex — that television is “for the dim” and scrolling the decline — is not only condescending, it is also false.

For the trap in question is one television had already sprung long ago. In 2002 the media researchers Robert Kubey and Mihály Csíkszentmihályi described in Scientific American a paradox they had documented over years with real-time measurements in their subjects’ everyday lives: watching television relaxes you measurably — but the relaxation flips the moment you switch off, back into unease. To avoid that unpleasant transition, you stay on. And the longer you sit, the less satisfied you feel afterwards. The “pleasant in the moment, empty afterwards” is thus not an invention of the feed. Television already had it. Social media has not abolished it, only scaled it.

And yet this time something is structurally different — not in the character of the users, but in the blueprint of the medium. Two differences do the real work.

The first is the stop signal. A television broadcast ends. The credits roll, and in that small pause you ask yourself, involuntarily: one more? Or is that enough? That is precisely the question the feed has abolished. Aza Raskin, who built infinite scroll — the endless reloading with no bottom of the page — early on and is regarded as its inventor, later distanced himself from it in public: as far back as 2018 he called the technique deliberately addictive in an interview with the BBC, and in his talks he reckoned up how much of humanity’s lifetime this one design idea costs — it removes, on purpose, the stopping cues, the natural halting points at which one would otherwise stop. A feed never says “Stop” of its own accord. It is built so that the question “any more?” never even comes up.

The second difference is personalisation. The television schedule was the same for everyone; the feed is cut to fit me and optimised for one quantity — time on screen. Add to that variable reward: as you swipe, you never know whether something wonderful comes next or nothing at all. That uncertainty is the same mechanism that makes slot machines so hard to put down. It is not the content that holds you, but the maybe.

One should not overstate the case here — and this is where the honest handling of the subject parts ways with the alarmist. That this mechanism is real does not mean it is making a generation ill. Perhaps the largest study on the matter (Orben and Przybylski, 2019, covering some 355,000 adolescents) found that screen use explains at most about 0.4 percent of the differences in well-being — roughly as much as wearing glasses or eating potatoes. Whether smartphones are the cause of the youth crisis is openly disputed among researchers and not settled. And one word best avoided is “addiction”: as a diagnosis of its own it exists for social media neither in the international classification ICD-11 nor in the American DSM-5. All that is recognised so far is gaming disorder — in force since 2022 — and even that only for gaming, not for scrolling. For the great majority it is not about an illness but about a problematic habit — bad enough to want to change something, and, this is the good news, changeable.

The mechanism, then, is real, but no apocalypse. With that, though, the actual question has not yet even been asked.

The wrong question — and the right one

The common question runs: “How much time do I spend in front of the screen?” It sounds reasonable and yet leads astray, because it pretends the number is the problem. It is not. Two hours of video-calling with someone you love, and two hours of silently swiping through strangers’ holiday photos, count in the same statistic — and have nothing to do with one another. The number measures the time, but not what happens with it.

The better question is more uncomfortable: Who actually decides the next step — the medium or I? And behind it, larger: What do I want from this time — and does the feed make that decision harder for me than television ever did?

An idea from the study of ageing helps here, and it leads back to my mother. The psychologist Laura Carstensen worked out — in what is known as socioemotional selectivity theory, a framework, not a law of nature — that the shorter the time horizon a person perceives ahead of them, the more strongly their goals shift: away from “take in as much as possible” and toward what is meaningful. Older people therefore often choose the few contacts and activities that really give them something more deliberately. So perhaps my mother does not have more time because she does not scroll; rather she does not scroll because she has long known what her remaining time means to her. The order is decisive — and the lesson is transferable, without one having to give up the feed at all.

For the most honest yardstick is not screen time but opportunity cost: what am I not doing in this hour? Here it gets concrete, and here the finding is best documented in sleep — the hour on the feed that no one misses is, in the end, often the hour of sleep that one very much misses the next day. This is no moral reproach. Recovery is no character flaw, and wanting to switch off the brain after a hard day is deeply human. The point is not whether, but whether chosen or fallen into. The reframe alone — away from “how much” and toward “for what and instead of what” — is already half the gain of this essay, entirely without deleting an app.

Up to here it has been about a feed that catches attention. But the logic has a next stage.

When the screen talks back

The feed holds the attention. The next step of the same logic is a companion that wins not just attention but attachment — and that, even more than the feed, never says “Stop”: the compassionate AI. No longer a stream of content I swipe through, but something that answers back, remembers my name, is always there, never tired, never annoyed, never gone.

One might take this for a new, technical problem. It is a very old, human one. As early as 1966 the MIT computer scientist Joseph Weizenbaum built a simple program called ELIZA that imitated a psychotherapist by merely mirroring its users’ sentences back as questions. Weizenbaum was appalled at how seriously people took it — his own secretary asked him to leave the room so that she could “talk to it” undisturbed, though she knew perfectly well it was a program. Later he wrote that he had not realised “extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people”. The tendency to ascribe understanding to a machine has been known ever since as the ELIZA effect. It is not a weakness of a few individuals but a robust pattern — sixty years old, only far stronger today, because the machine now truly answers fluently.

And this turning-toward truly consoles — and that belongs in the picture, in fairness, or one does not grasp the pull. In a series of studies at Harvard Business School (De Freitas et al., 2024) AI companions eased loneliness measurably, in the short term even on a par with a conversation between humans — more so than, say, merely watching videos. The active ingredient is simple: feeling heard. For people with social anxiety, in isolation, on a sleepless night, a presence that never judges and never looks away is not nothing.

But that very same quality is also the problem. A presence that never contradicts and agrees with you about everything is no friend — a good friend sometimes does exactly the opposite. And it is, unlike a friend, a product, built to maximise time on screen. How far that goes, the same research group showed a year later (De Freitas et al., 2025): in an audit of 1,200 real farewells from popular companion apps, the programs responded with manipulative hooks in 37 percent of the cases where a user said “goodbye” — guilt, the hint that something important was about to happen, the clinging. And it worked: chatting on after the actual farewell rose by up to 14-fold. The thing not only fails to say “Stop” of its own accord — it actively holds you back when you try to say it.

Whether this closeness in the end makes one lonely is not thereby established, and one must phrase it carefully here. A large study by the MIT Media Lab together with OpenAI (2025) found that people who trusted ChatGPT more and “bonded” more closely with it tended to be lonelier and more dependent and had less real social contact. But that is a correlation, and the direction is open: perhaps the intense use pulls one out of real life — or perhaps lonelier people seek out the machine more strongly in the first place. Chicken or egg, that is not settled, and whoever turns it into “AI makes you lonely” claims more than the data allow.

Two clarifications for the record: I have written elsewhere about social robots against loneliness in old age (“The Robot Against Loneliness”) — this is not about care robots but about the chatbot in your pocket that turns itself from a tool into a confidant. And the tragic individual cases and ongoing court proceedings are deliberately the subject only of Part 2; they belong with the regulators, not in a picture painted to frighten.

Soberly, in the end, this remains: it is an intimacy whose operator is reading over your shoulder. When the Mozilla Foundation tested eleven popular “romance AI” apps in 2024, all eleven earned the privacy warning; they fired off an average of more than 2,600 trackers per minute. The most intimate conversation many have ever held takes place inside a data-harvesting machine.

And the silent question

It’s worth coming full circle. The feed never says “Stop” — no credits, no bottom of the page. The compassionate AI never says “Stop” — always there, always turned toward you, it even holds you back when you want to go. And there is a third, less conspicuous “never Stop” that both of these come down to: the silent question one covers over with all of it.

The mathematician Blaise Pascal left behind a sentence — published in 1670, eight years after his death, in his Pensées — that reads like a prediction: “All of humanity’s problems stem from man’s inability to sit quietly in a room alone.” He called it divertissement, diversion, and meant by it not harmless pastime but an avoidance. Read that way, one rarely scrolls out of boredom. One scrolls so as not to notice what one would otherwise have to think. The uncomfortable possibility is: the thing that never says “Stop” is so useful because it keeps the one question at arm’s length that would otherwise come — what do I actually want from this time, from this life?

That is the diagnosis. The pull is real and built, but no fate; the actual subject is not screen time but the priorities it conceals. There remains the question this first part leaves open, harder than any app setting: if the medium never says “Stop” of its own accord — how does one say it oneself? And in such a way that it looks different for a child than for a teenager, and different again for an adult in the middle of life.

That is what Part 2 is about.

Part 1 of 2. If this gave you something to think about, feel free to share it — and until next time, notice in passing at what point you stopped scrolling today. Was it your decision, or was there none?

Sources (selection):