Explain it in two minutes
What is a filter bubble? And how to explain it to a teenager.
Short version: a feed is a feedback loop. It shows more of what you engaged with, so engaging narrows what you are shown, which narrows what you can engage with next. No one has to have chosen the narrowing for it to happen — it is the loop running exactly as built, and it is not a failure of your teenager’s judgement or of your parenting. Below: the mechanism in plain language, an honest account of what the research does and does not show, how to raise it without it reading as a lecture, and a demo that closes the bubble in a single tap and then lets you push it back open.
Ages 14–18 Read it in three minutes Free demo, plays in your browserWhat is a filter bubble?
A filter bubble is what you are left seeing once a recommendation system has finished learning what holds your attention. The term was coined by the writer Eli Pariser in 2011, and the thing it names has not changed since: a feed decides what to show you next mostly on the evidence of what you engaged with before.
The load-bearing word is loop. What you engage with decides what you are shown; what you are shown is the only thing you can engage with next; and that engagement decides the round after. Each single pass is small and entirely reasonable. Run it a few thousand times and the range of things reaching your teenager has moved a long way from where it started, without anyone — including them — ever making that decision.
The reason it is hard to notice from the inside is a property of the interface, not anything sinister. A feed shows you what it selected. It does not leave a gap on the screen where the things it stopped showing used to be. From the seat your teenager is sitting in, a feed that has narrowed and a feed that is showing them the whole world look identical.
“Filter bubble” and “echo chamber” get used interchangeably and they do overlap, but they are worth separating. A filter bubble is what the ranking selects for one person. An echo chamber is the room you end up standing in: a space where the same view keeps arriving from what look like different directions, so it starts to feel less like an opinion and more like the weather. In practice the two words get swapped for each other constantly — the demo below does it too — which is worth saying out loud when you run it with your teenager.
Why does my teenager only see one kind of content?
Because the system did the job it was built to do. Nothing has gone wrong with their phone, and nothing has gone wrong with them. A recommender is optimised for engagement — time on a clip, replays, the length of the pause before a scroll. It is not optimised for range, because range is not a number it measures. Narrowing is not the aim; it is what falls out of the aim.
That distinction is worth getting right before you raise any of this with them. There is a popular version of the story in which a company decided to trap your child in one worldview. Usually that is not what happened, and it does not need to have happened: a ranking that maximises one number will find the content that maximises that number and serve more of it, and if that content all points one way, the feed points one way. This is also why “they are doing it to you on purpose” is the weakest argument to bring to a sixteen-year-old. They will find the hole in it in about four seconds, and then they will stop listening to the rest.
The signals are behavioural, not stated. Nobody asked your teenager whether they wanted more of this. They watched one to the end, went back for a second, and the loop took that as the answer. The same is true in reverse: scrolling quickly past something is a signal too. The feed narrows from both ends at once — pulling in what held them, quietly letting go of what did not.
And the loop has no idea what the subject is. The same mechanism that fills a feed with one football club fills it with one politics, one body, one narrow idea of what a normal life looks like. It is not choosing those things because it prefers them. It is counting.
Do filter bubbles really cause polarisation?
Here we should be careful, because the honest answer is: the big version is contested. The strong claim — that personalised feeds are a principal cause of political polarisation across a whole society — is genuinely disputed among researchers. It is an open question, not a finding. The counter-argument runs like this: people use several platforms, they see what their family and colleagues post, and some of the disagreement they meet online they would not have met offline at all. Anyone telling you this part is settled is ahead of the evidence.
What is not in dispute is the mechanism at the level of one person’s feed. An engagement-optimised recommender shows more of what you engaged with and less of what you did not. That is not a research finding to be argued over; it is what the ranking is for. You can watch it happen in the demo below, in a single tap.
One well-documented thing about people, rather than software, is worth adding, because it explains why a narrow feed feels convincing rather than merely narrow: repeated claims get rated as more likely to be true than unfamiliar ones. That is the illusory-truth effect, first reported in the 1970s and replicated many times since. It is not a mark of being fifteen, or of being gullible — the classic studies were run on adults, and repetition shifts what feels true for careful, well-read ones too. A feed that keeps repeating one framing is doing that to everyone in the house, and saying so out loud takes the accusation out of the conversation.
So the thing to say to your teenager is the smaller, sturdier claim — which happens to be the more persuasive one anyway: this is what your feed is doing, right now, to you. Not “this is wrecking democracy.” A claim they can check on their own screen tends to land. A claim they cannot check gets filed with the other things adults say about phones.
How do I talk to my teenager about social media without lecturing?
Make it a system they can inspect rather than a habit they are being told off for. Teenagers are very good at spotting the moment a conversation about technology is actually a conversation about them, and that is the moment the shutters come down. You are not asking them to justify anything. You are showing them a machine and betting they will find it interesting, which they usually do.
Two openings that tend to survive first contact. The first is to ask them to explain it to you — “how does it actually decide what to show you next?” — and mean it, because they spend far more time inside that system than you do, and being the expert in the room changes what kind of conversation it is. The second is to compare. Sit next to each other and scroll your recommended feeds side by side on the same app — not your search results, the feed itself. Ask them to narrate why they think each thing showed up for them and not for you. Where the two feeds diverge, that is the ranking showing its hand; where they overlap, that is worth noticing too. Nothing you say will make the point as well as their screen sitting next to yours.
Then hand over the part they control, because it is the only part a sixteen-year-old will take. Deliberate engagement is a signal too. Follow three accounts that do not fit the pattern, use the “not interested” control on the things they were scrolling past anyway, and then look again next week to see what has moved. It is an experiment they run themselves, on their own phone, with the result in front of them — and if less has changed than they expected, that is a finding too. It is also the honest version of the advice: you cannot step outside the loop, but you can decide what goes into it.
One last thing — keep two conversations apart. “Here is how the ranking works” and “here is what I think about what you have been watching” are both fair conversations. Held together, the first becomes the delivery mechanism for the second, and they will only hear the second. If something in that feed genuinely worries you, say it plainly and on its own, another day. That conversation deserves its own evening.
Now let them close the bubble themselves
Six topics, drifting. Tap the one that catches your eye — the way anyone taps a feed — and watch what a single tap does: the counter drops from six to one, the topic you touched swells to fill the middle, and everything you did not touch shrinks and dims at once. That is the demo doing in one tap what a real feed does over thousands of small ones. Then it hands the loop back: tap one of the small, dimmed ones and pull it in again. It runs in the browser — no download, no signup.
1Tap one thing that catches your eye, exactly as you would in any feed. There is no wrong tap here — that is rather the point.
2Watch the counter marked “Topics still in your feed” drop from six the moment you tap — not gradually, at once. The topic you touched grows to fill the middle; everything else shrinks and dims. Nobody chose that.
3When it collapses, tap one of the small, dimmed ones. It comes straight back — that is the same loop running the other way, and it is the only part of it a teenager actually controls.
Every one of these runs entirely on your device — no ads, no tracking, and no way for it to reach the internet at all. That is how they ship to a child.
More questions, answered properly
- Why is the sky orange?And how to explain it to a five-year-old.
- Why does the Moon change shape?Moon phases, without the eclipse mix-up.
- Why do we have day and night?The Earth spins — that is the whole answer.
- Why does my child melt down when screen time ends?It is the design, not your child. Ages 5–7.
- Why can’t my child put their phone down?Two mechanisms, and what actually changes it. Ages 11–13.
What is WeKids?
WeKids turns the questions your child actually asks into something built for them — a short video or a hands-on activity, made for the age they actually are, in an app of their own. The demo above is that kind of thing, on a subject that happens to be the feed itself.
You get the other half: gentle-parenting nudges that cite the research they came from, so you are the one deciding what reaches your child rather than an engine optimising for their attention. That is also why this page tells you which parts of the filter-bubble story are settled and which are still being argued about — a nudge that cites its evidence has to be willing to say when the evidence is thin.
Their app is not a feed. The day’s plan is finite — it finishes, and it says so — with no autoplay and nothing scrolling on underneath, because keeping a child in there longer was never the thing we were trying to do.
And when a question is better coming from you than from an app — the tender ones, about death or bodies or fairness — it says so, and helps you start instead of answering over your shoulder.
Get WeKids
Start here — WeParents
Then, on your child’s device — WeKids
Free to start. No ads to children, ever.