How it works1 of 11

Why you're seeing this

Nothing on a screen got there by accident. A company chose it, and it chose what keeps you looking.

How old is your kid?

How to explain it, by age

Same idea every time. What changes is how much of the machinery a kid can hold, so pick the version that fits and go up a band when it stops being interesting.

Ages 4-6

Skip the word algorithm. Try this instead: "Whoever made this show wants you to keep watching, so they made the next one start all by itself. That was their choice, not yours." Keep it to the fact that a person made a choice, because that is the whole idea at this age.

Try this

Watch something together and pause right when the next episode auto-starts. Ask: "Did you pick that? Who did?" Then let her pick the next one out loud, so the difference between being handed something and choosing it is something she has felt rather than been told.

Listen for

Any version of "somebody wanted me to." She does not need the mechanism. She needs to know a person is on the other end of it.

If she already has more access than most. If she already uses a tablet alone, add one rule with the explanation attached: "We turn off the part that picks for you, so you get to pick."

Ages 7-9

Now you can name it. "An algorithm is a set of instructions, like a recipe. People wrote a recipe that decides which video you see next. The recipe's job is to keep you watching, because the company earns money from ads while you watch." Seven-year-olds understand recipes and they understand rules, which is all this requires.

Try this

Sit with two different accounts, yours and hers, and search the same word. Compare the top five. The results differ, and she will notice on her own, which is worth a great deal more than being told.

Listen for

"Why did you get different ones than me?" That question means the idea has landed and you can stop explaining.

If she already has more access than most. If she already has her own account, look at the recommendations together and have her guess which of her past videos caused each one. She will be right more often than you expect.

Ages 10-12

Introduce the money. "These apps are free, so you are not the customer. Advertisers are the customer, and your attention is the product being sold. That is not a conspiracy, it is just the business model, and it explains almost every design choice you notice." Then name the choices: autoplay, infinite scroll, streaks, the pull to refresh, notifications timed for when you are most likely to open them.

Try this

Have her find the design feature in an app she uses that exists purely to extend a session. Then have her turn one of them off and use the app for a week. Ask her afterwards whether she used it less, and whether she missed it.

Listen for

Her spotting a design choice you did not point out. That is the skill transferring, and it is the point of the whole exercise.

If she already has more access than most. For a kid already deep in a platform, go one level further: ask who benefits from each recommendation, and whether the creator, the advertiser and the viewer want the same thing.

Ages 13+

Engagement optimisation has second-order effects nobody chose. A system tuned for attention will find that outrage travels further than nuance, that extremes hold longer than moderation, and that a body-comparison spiral is very engaging indeed. No engineer sat down and decided this. It emerged from the objective, and that is a more unsettling fact than a villain would be.

Try this

Have him deliberately train a fresh account for a week, engaging only with one narrow topic, then look at what the feed became. Doing it on purpose is how you see the mechanism you cannot see when it happens to you.

Listen for

Recognition that the system has no intent. The sophisticated version of this idea is not that the company is evil, it is that an optimiser does not care.

If she already has more access than most. Move to the labour and moderation layer: who decides what is allowed, who is paid to look at the worst of it, and what happens in the countries where those decisions are cheapest to make.

Why this concept goes first

A kid who can ask “why am I being shown this” has a tool that works on everything she will meet later: recommendation feeds, targeted ads, chatbots, whatever replaces them. Everything else in this curriculum is a variation on it.

It also has a property most media literacy lessons lack. It is checkable. She can run the comparison herself and see the difference with her own eyes, which means she is not taking your word for it. That matters more than it sounds, because a child who has verified one thing about how technology works is much harder to convince that the rest is magic.

The mistake to avoid

Do not turn this into a lesson about how the apps are bad. It backfires, and it is also not quite true. The point is that a system was built with a goal, and the goal is not the same as her goal. A kid who understands that stays curious. A kid who is told the internet is poison learns to hide her use of it from you, which is the outcome least compatible with anything else you are trying to do.

Misconceptions to address

That the app shows everyone the same thing. It does not. Two kids sitting side by side see different feeds, which is exactly why comparing notes works so well as a demonstration.
That it shows you what you like. Closer, but not right. It shows what holds attention, and the things that hold attention are not always the things a person would have chosen.
That someone at the company picked this particular video for you. A person wrote the rules and a machine applies them billions of times, which is why nobody at the company can fully explain any single recommendation.
Adults get this one wrong too. Plenty of grown-ups still treat their feed as a neutral window on the world rather than a product decision.

Running this with a class

Have every student open the same app on their own device and search the same word, then compare the first five results across the room. The differences do the teaching for you. For a class without devices, run it as a thought experiment with two invented students whose interests you describe, and have the class predict what each one gets shown.

The decision this connects to

Last reviewed August 2026