The five hooks: how platforms are built to hold on to children and young people

When a child cannot put the phone down, it is usually explained as a failure of willpower: the child's or the adult's. Both explanations are wrong, and both lead professionals down a path where nothing works. What is actually happening is that five mechanisms, designed by people who were well paid for it and tested on millions of users, are working exactly as intended.

Why I am writing this

I have built such systems professionally.

For several years I worked in strategic communication and influence. Not short-horizon campaigns, but patient, long-term presence in the channels young people actually use, delivered through voices they already trusted. The language was not accidental. Every word was chosen, measured and chosen again. The goal was not a sale by Friday. The goal was to shift what a young person noticed, what they took for granted without checking, and what felt normal to them.

That work had legal advice. It had approval processes, rules about who we could reach, and long meetings about whether a single sentence was fair to the person who would read it.

That experience is why I am writing this. The arithmetic is uncomfortable: if it was achievable with a moderate budget, a small team and a long list of things we were not allowed to do — what is possible for actors with none of those constraints?

Almost none of them feel they are doing anything wrong. Neither did we. The reshaping is a by-product, and by-products do not come up in the approval meeting, because they are not in anyone's job description.

Why naming the mechanism is the whole point

Knowing how the trick works does not make you immune to it. That is worth saying plainly, because many professionals experience exactly this: they understand the problem, and still find the knowledge does not help them in the room with the child.

But naming does something else, and that is why this is worth half a day of a staff group's time.

Once the mechanism has a name, it stops being a personal matter. Without the name, the professional is left with two explanations, and both are bad: either there is something wrong with the child (poor self-control, weak boundaries at home), or there is something wrong with the adult who cannot manage it.

The third explanation is the correct one. Something is being done. It has a name. It was designed, tested and refined over years, and it works as intended.

That is the difference between a staff group discussing a child's character and a staff group discussing the system the child is standing in. Only one of those leads to measures that work.

The five mechanisms

Dozens of techniques are in circulation. Five of them do most of the work. The rest are variations. Learn these five and you can open any new platform in 2029 and find all five again within ninety seconds.

1. The slot machine

A machine that dispenses chocolate every time you pay is boring. You use it when you want chocolate, and otherwise you do not think about it. Same price, same chocolate, just unpredictable: a machine that pays out three times in ten is a slot machine, and people stand at those until their legs ache.

This is the oldest finding in behavioural psychology, and it is not disputed. An unpredictable reward is dramatically more compelling than a predictable one. It works on rats, on pigeons and on well-educated adults. There is no version of a human being it does not work on.

On the phone it is pull-to-refresh, which is a lever and was designed to feel like one. It is the red dot. It is opening the app for the fourth time in an hour because something might have happened. Usually nothing has happened, and that is exactly why they check again.

One important distinction: checking for anything at all is the slot machine. Waiting for a reply from one particular person is something else entirely: same thumb, same screen, completely different problem. Treat waiting-for-a-reply as addiction and you are wrong, the child notices immediately, and you have spent trust for nothing. The question that separates them is simple. Are you waiting for someone specific?

2. The social debt

A family goes on holiday somewhere with poor coverage. On day three the twelve-year-old goes quiet, then slightly frantic: there is a streak, a number next to a friend's name counting consecutive days of contact. It stands at four hundred and something, and it dies tonight.

The interesting part is that nobody likes the streak. Ask them both. They will say it is stupid.

It makes no difference, because the streak was never a reward. It is a debt. The same goes for read receipts, typing indicators and the little green dot that says active now. None of them are features for the user. They are obligations dressed as information, and each of them takes something that was free and attaches a cost to not doing it.

Before read receipts, a message was an offer: you took it when you had a moment. After read receipts, a message is an invoice with a visible timestamp, and everyone involved can see exactly how long it has gone unpaid.

The genuinely elegant part, the one I would have been professionally impressed by if it had crossed my desk, is that the debt is not owed to the company. It is owed to a friend. Stopping does not feel like leaving a product. It feels like letting someone down.

3. The identity mirror

Who chose the child's feed?

Most people would say the child did. They followed someone, liked something. That is how it feels from the inside too, and that is precisely the point.

What actually happened is that a system formed a hypothesis about who the child is, showed them something to test it, and measured what came back. Not a like. Not a follow. Something far smaller: a pause. Three quarters of a second of not scrolling is a signal, and it is a more honest signal than anything the child would have written, because nobody edits a pause.

Then it repeats. Thousands of times a week.

The child did not choose their feed. They confirmed it, unintentionally, one small hesitation at a time. And the result feels like their own, because in a sense they built it, and because it hits often enough to feel like being understood.

That is why this one is the most powerful of the five. For a thirteen-year-old who suspects nobody really sees them, an algorithm that seems to get it is not an irritation. It is a relationship.

This is also the mechanism most relevant to professionals meeting children in vulnerable phases. The narrowing usually looks entirely innocent right up until it does not.

4. The endless runway

Every medium before this one had an edge. A book has chapters and a last page you can feel coming with your thumb. Television had the end of the episode. A newspaper ran out. A record had a side, and someone had to get up and turn it over.

People stop at edges. We are not particularly good at stopping anywhere else, and never have been. The edges were not there for our benefit, incidentally: they were technical limitations. We built our self-control on top of them and mistook it for character.

The core innovation of the modern feed was removing the edge. That is the whole invention. No bottom to the page, no end of the episode, no page two. Autoplay starts the next one before you have finished deciding, and the decision you were halfway through quietly expires.

So when a child says I'll stop after this one, they are not lying. There is no "this one". There is no unit to stop after. You have asked someone standing on a moving walkway to stop at a particular tile.

This gives the single most useful question in the whole framework, and it is the one we recommend organisations use when separating screen use worth worrying about from screen use that is not:

The question is not how long. The question is: does it end by itself?

A film ends. A match ends. An episode ends. A feed does not. That is why general screen-time rules collapse in practice. They treat fundamentally different categories as if they were the same.

5. The targeting machine

Most people have had the moment: an advert appears for something the child has not mentioned to a living soul. The commonest explanation among adults is that the phone is listening.

It almost certainly is not. And while you are busy fearing the wrong thing, you are not looking at what is actually happening, which is considerably more impressive.

When you build audiences professionally, you are not trying to learn anything about individuals. That is slow, expensive and unnecessary. What you do is find a pattern that predicts behaviour: a combination of signals, most of them boring, none of them secret, that reliably appears before someone does the thing you want them to do. Then you find everyone else who fits the pattern.

You never need a name. You do not need a confession, a search or a single spoken word. You need enough people who resembled this person to have already done the thing, so the system can say: this one next, with roughly this much confidence.

So the child is not being listened to. They are being predicted. And here is the consequence that matters: if a system can predict what a child is going to want, it can also show them what makes them want it. Prediction and influence are the same job performed in two directions, and every technique in this article is available to whoever pays.

What this means for you as professionals working with children

Three things follow from the framework above, and all of them are organisational before they are individual.

1. Shared vocabulary before shared measures. A staff group without a common language for what is happening will reach different judgements about the same child. That is not professional disagreement. It is an absence of shared categories.
2. Stop measuring time, start measuring form. "How many hours" is almost always the wrong question. "Does it end by itself", "is she waiting for someone specific", "what has the feed concluded about this child": those are questions that give you somewhere to act.
3. Recognise that this applies to your own channels too. Organisations that communicate with children and young people use the same mechanisms, usually without having taken a position on it. This is where the question of digital neglect starts to get uncomfortable, and it is also where it becomes most useful.

On digital neglect

The term came up in feedback after a talk we gave for the Salvation Army in Norway. Their head of professional development raised the matter at the following Monday's management meeting with the question: are we contributing to digital neglect?

It is a better question than it first appears. Neglect is not only about what adults do to children, but about what adults fail to protect them from when they had the opportunity to understand. When the mechanisms above are known, documented and named, what does it mean not to familiarise yourself with them, in an organisation whose remit includes children and young people?

We do not claim the question has an easy answer. We think it is worth asking in the management meeting.