The algorithm is a set of sorting rules. For every post it could show you, it predicts how likely you are to react — to watch, like, comment or share it. Then it sorts your feed so that the posts with the highest predicted reaction come first. It learns from your behaviour: what you linger on, like or share raises the weight of similar content. So your feed is not neutral and not chronological — it is assembled to keep your attention for as long as possible.
Chapter One
You open the app —
and suddenly
an hour is gone.
Social media,
understood
properly.
Social media,
understood in the
classroom.
— How social media works: feed ranking, the attention economy, filter bubbles and healthy use explained simply.
You wanted to take a quick look — and then you keep scrolling and scrolling. That's not because you have no willpower. The feed is built so that it never quite ends and always shows you the next thing that might hold you. Don't worry: once you understand how it works, you can take the controls back. We'll look at it calmly, step by step.
Between opening the app and putting it down lie an infinite feed, a ranking that constantly predicts your reactions, and a business model in which your attention is the product. This page sorts the building blocks — feed ranking, the attention economy, filter bubbles and echo chambers, the outrage incentive — and makes the personalisation feedback loop tangible with an interactive simulator. Mechanics, not moralising.
Why does the scrolling never stop, and who decides what appears in your feed? This learning unit gets students from Year 7 up ready to understand social media: from feed ranking through the attention economy to filter bubbles and a healthier use. With an interactive filter-bubble simulator, a quiz and discussion prompts — ideal for media-literacy lessons.
~ Get comfortable. We'll go through this together.
~ With the mechanics in context and an interactive filter-bubble simulator.
~ Recommended learning unit: 2 lessons of 50 min each.
Chapter Two
The feed is
no accident.
The algorithm
sorts.
What is an
algorithm?
What you see in your feed is not random and not simply in the order it was posted. An algorithm sorts it for you. Picture a waiter who has learned over time what you like to eat — and now serves you exactly the dishes you reach for most readily. The algorithm does the same with posts: it puts the ones at the top that you are most likely to react to.
An algorithm is simply a set of rules for working through a task step by step. The feed algorithm treats ranking as a prediction problem: for each candidate post it estimates how likely you are to react — to watch, like, comment or share. Posts are then ordered by that predicted engagement, not by time. The neutral order would be chronological; the algorithmic order is optimised for your attention.
An algorithm is a recipe: a fixed sequence of steps that solves a task. The feed algorithm uses such a recipe to sort your posts. It is like a waiter who remembers what you like and serves it first. Important: the algorithm doesn't judge whether something is good or true — it only estimates how strongly you are likely to react.
And it keeps learning. Every time you linger on something, like it or share it, you tell the algorithm: more of this, please. So your feed changes with you — quietly, without you choosing it.
The pipeline runs in stages: candidate generation gathers a large pool of possible posts, scoring predicts a reaction probability for each, and re-ranking mixes in freshness, diversity and policy rules. A key signal is dwell time — how long you linger — which often weighs more heavily than an explicit like, because it is harder to fake and available for every post.
And the recipe keeps adapting: each like, share and lingering view feeds back in. The feed is therefore never finished — it changes with your behaviour. That is exactly what we'll make visible later in the simulator.
Your feed is sorted, not neutral — the algorithm puts at the top what you are most likely to react to.
Core idea: ranking is a prediction problem — order posts by predicted engagement, not by time.
Rough ranking signals from strongest to weakest: share, then comment, then like, then dwell time.
- Share — the strongest signal. If you pass a post on, the algorithm reads it as: this is really important to you.
- Comment — writing something takes effort, so it counts a lot. Even an angry comment is a strong signal.
- Like — a quick tap. A clear but lighter signal that you like the content.
- Dwell time — how long you linger, even without tapping. Quiet, but available for every single post.
Why dwell time matters so much: An explicit like only exists when you give it. Dwell time, by contrast, is measured for every post — including the ones you watch but never react to. It is therefore a dense, hard-to-fake signal. This is also why content that makes you stop and stare (shocking, emotional, suspenseful) is structurally favoured, regardless of whether you "approve" of it.
Chronological vs. algorithmic feed: A purely chronological feed shows posts newest-first — simple, predictable, but it buries good older posts and floods you with whoever posts most. An algorithmic feed surfaces "relevant" posts but is opaque and optimises for engagement, not for your wellbeing. Many apps now offer both; the choice is a real lever (see Chapter 6).
Where this page draws the line: How posts technically reach your device (servers, data packets, networks) is covered by the sister site Internet verstehen. Which data is collected about you for ranking and ads is covered by Datenschutz verstehen. This page stays on the feed and ranking mechanics themselves.
📚 Learning objectives
- You can explain an algorithm in your own words (a recipe, a set of steps).
- You can explain that the feed is sorted and not chronological.
- You know the rough order of the signals: share > comment > like > dwell time.
- Algorithm: a fixed sequence of steps that solves a task.
- Feed ranking: sorting posts by how likely you are to react.
- Dwell time: how long you linger on a post — a quiet but powerful signal.
You don't have to like a post to "vote" for it. Simply stopping to watch it longer already tells the algorithm: show me more of this.
❓ QuizWhich order does the feed use by default?
Answer B: “The order the algorithm predicts you will react to most.”
A (purely by time) is the chronological feed, which most apps no longer use by default. C (alphabetical) doesn't exist. Only B describes feed ranking.
For the teacher — options: A: “Strictly newest first.” / B: “The order the algorithm predicts you will react to most.” / C: “Alphabetically by name.”
🎯 Extended objectives (Bloom's taxonomy)
- L1 — Knowledge: students name three ranking signals (like, comment, share).
- L2 — Comprehension: students explain "the feed is sorted, not chronological" in their own words.
- L3 — Application: students sort example posts as the algorithm would ("Be the algorithm").
- L4 — Analysis: students discuss why dwell time can outweigh a like.
- 2 min: read the lead text together.
- 4 min: collect on the board: "What decides what I see first?"
- 3 min: discuss the signal order (share > comment > like > dwell).
- 4 min: quiz in small groups — guess first, then reveal.
- 2 min: discussion: "Is a sorted feed good or bad?"
Question: "If you could choose — would you rather have your feed sorted by time or by the algorithm? Why?"
🔗 Cross-referenceWhich data is collected about you for this sorting is explored in depth by the sister site Datenschutz verstehen (Understanding privacy).
Chapter Three
Why is it all
free?
Attention is
the currency.
What does the
app earn from you?
Social media costs no money — and yet the companies behind it earn billions. How? The answer is short: your attention is the product. The more time you spend in the app, the more advertising can be shown to you, and the more the platform earns. That's why everything is built to keep you there as long as possible.
The business model is the attention economy: human attention is scarce, so it becomes the resource that is sold. The platform is free for you because you aren't the customer — advertisers are. The product they buy is access to your attention. This single fact explains the design: long sessions, an endless feed and engagement-optimised ranking are not flaws, they are the goal.
If something is free, ask: how does it earn money? With social media the answer is your attention. The longer you stay, the more ads you see, the more the app earns. So the goal of the whole design is to keep you scrolling — that is the attention economy.
The rule of thumb: if you don't pay for the product, your attention is the product.
Attention economy: attention is the scarce resource. "Time on app" is the metric the design optimises, because it converts directly into ad revenue.
The important part for you: this isn't aimed at you personally, and it isn't a flaw. It is simply how the system is built. And once you see the incentive — keep you there longer — the design stops feeling like magic.
Where the ad mechanics belong: How ads find exactly the right audience in milliseconds — the advertising profile built about you, targeting, and the real-time auction (real-time bidding) — is its own large topic. The sister site Datenschutz verstehen (Understanding privacy) covers it in depth. This page deliberately stays on the incentive: why keeping you on the app for longer is the goal.
Why "engagement" is the lever: Revenue scales with attention, and attention is measured as engagement — sessions, time on app, return visits. So the metrics the company optimises (and the ranking that serves them) point in one direction: hold you longer. Your wellbeing isn't part of that equation unless it happens to overlap with staying longer.
It isn't a conspiracy — it's an incentive: No one needs to be malicious for the feed to become attention-greedy. If the money comes from attention, then whatever holds attention gets rewarded automatically — by the metrics, the ranking and the design tweaks that follow them. Understanding the incentive is more useful than blaming individuals.
📚 Learning objectives
- You can explain why a "free" app still earns money.
- You can describe the attention economy in one sentence.
- You understand that the design goal is "time on app".
The advertisers are the real customers of a free platform. The product they buy is access to your attention.
❓ QuizHow does a free social media app earn money?
Answer C: “It sells your attention to advertisers.”
A (a monthly fee) is wrong — it's free. B (the state pays) is wrong. The app earns through advertising, and the longer you stay, the more it earns.
Options: A: “Through a secret monthly fee.” / B: “The state pays for it.” / C: “It sells your attention to advertisers.”
⏱ Timing (≈ 14 min)
- 3 min: ask the class: "If it's free, who pays?"
- 4 min: introduce the attention-economy idea with everyday examples.
- 4 min: discuss why "time on app" becomes the goal.
- 3 min: quiz + answers.
- Complete the sentence: "If you don't pay for the product, then …"
- Why does the app want you to stay as long as possible?
- Name two design choices that keep you on the app longer.
How exactly ads are matched to you — profiles, targeting, the real-time auction — is covered by the sister site Datenschutz verstehen (Understanding privacy).
Chapter Four
How you get
sorted.
The filter
bubble forms.
Build your own
filter bubble.
Now it gets hands-on. Every like, share and lingering view tells the algorithm a little more about what suits you — and it gives you more of the same. Over time you see more and more of one kind of content and less and less of the opposite. That narrowed selection is your filter bubble. Try it: react to the posts and watch how the algorithm's picture of you shifts.
The filter bubble is a personalisation feedback loop: your reactions shape the picture the algorithm has of you, and that picture shapes what you see next, which shapes your reactions. Below you can drive the loop yourself. Each action re-weights the next batch deterministically — share counts most, then comment, then like, then dwell time — so one category visibly takes over and the diversity index drops.
Time to experiment! React to the posts below — like, comment, share, or watch longer — and watch on the right how the algorithm's picture of you changes. The more you feed one category, the more the next batch tilts towards it. That is how a filter bubble forms. Then try the "Burst the bubble" button.
Filter-bubble simulator
React to a post on the left — and watch the algorithm's picture of you on the right change.
Each action re-weights the next batch: share > comment > like > dwell time. "Not interested" pushes a category back.
Use the buttons on each post. After each round, read the caption: which category is dominant, and what is your diversity index?
Your feed
How the algorithm sees you
Diversity index
100
Your feed starts balanced: every category gets a roughly equal share, and your diversity index is 100. React to a post and watch how the picture shifts.
Round 1 of 5
- Each reaction is a voteLike, comment, share and even dwell time all tell the algorithm what to show you more of.
- Share counts mostPassing a post on is the strongest signal — stronger than a comment, a like or lingering.
- The bubble forms quietlyNobody decides it on purpose. It grows from your own reactions, batch by batch.
- You can't see what's missingThe bubble is invisible because you never see the posts that were filtered out.
Filter bubble vs. echo chamber: A filter bubble is the algorithmic pre-selection — what gets filtered out for you so you see less of the opposite. An echo chamber is the social layer on top: when the people around you mostly share one view, it keeps coming back like an echo and feels more correct and more widespread than it is. The bubble is mechanical; the chamber is the reinforcement that grows out of it.
Why it's hard to notice: A filter bubble is invisible from the inside because you only ever see what made it through. You can't compare your feed to "everything that exists", so the narrowed selection feels like the whole world. Diversity by default is exactly what an engagement-optimised ranking does not reward — which is why deliberate counter-steering (Chapter 6) matters.
How the simulator works: The next batch is computed deterministically from your reactions (no randomness in the state). Each reaction adds a fixed weight to its category — share > comment > like > dwell time — and the batch is drawn from a small pre-baked pool in proportion to those weights. The diversity index measures how evenly the four categories are balanced: 100 means perfectly even, low means one category dominates. "Not interested" and diverse follows push the index back up.
📚 Learning objectives
- You can explain how a filter bubble forms from your own reactions.
- You can distinguish a filter bubble from an echo chamber.
- You understand that you cannot see what the bubble filters out.
What is a filter bubble?
Answer A: “The narrowed selection of content the algorithm shows you because you react well to it.”
B (a privacy setting) and C (a virus) are wrong. The filter bubble is the personalised pre-selection that you usually don't notice.
Options: A: “The narrowed selection of content the algorithm shows you.” / B: “A privacy setting.” / C: “A kind of computer virus.”
⏱ Timing (≈ 18 min) — core of the lesson
- 6 min: filter-bubble simulator on the projector — react to posts, read the index aloud each round.
- 4 min: let a few students drive it themselves and predict the next batch.
- 5 min: discussion "What does your feed leave out — and how would you notice?"
- 3 min: quiz + answers.
Before pressing "Burst the bubble", let the class predict what will happen to the index. Then reveal — the surprise makes the mechanic stick.
🖨 Mini worksheet- Describe in one sentence how a filter bubble forms.
- What is the difference between a filter bubble and an echo chamber?
- Why is it hard to notice your own filter bubble?
Chapter Five
Why anger
spreads.
Outrage is
engaging.
Why outrage
gets reach.
You may have noticed: posts that make people angry are everywhere. There's a reason. The algorithm measures reactions, not mood and not truth. And nothing triggers reactions like outrage — people comment, argue and share. So the algorithm sees: a lot is happening here! And shows the post to even more people. Anger keeps you engaged, and that is exactly what gets rewarded.
Outrage is a high-engagement signal. Anger drives comments, quote-shares and arguments — all strong ranking signals. The ranking can't read intent or valence; it only registers that engagement is high and amplifies accordingly. The result is structural: not because outrage is true or good, but because it is engaging, it spreads. This is amplification, not a verdict on the content.
Have you noticed how much anger there is online? That's no coincidence. The algorithm counts reactions, and outrage produces a lot of them: people comment, argue and share. So the algorithm thinks "this is important!" and gives the post even more reach. Anger keeps people engaged — and engagement is what the ranking rewards.
Outrage isn't amplified because it's true, but because it's engaging.
Amplification, not endorsement: the ranking rewards reaction volume. Outrage maximises reactions, so it is structurally favoured.
The good news: once you know this, the spell weakens. When a post makes you instantly furious, it's worth pausing for a second — that feeling is exactly what gets you to react. And even an angry comment counts as a reaction. Sometimes the calmest move is to scroll on.
Why anger beats joy in the ranking: High-arousal emotions — anger, outrage, fear — drive more immediate action than calm or contentment. More action means more measurable engagement, which the ranking reads as relevance. Even hate-reading and angry replies feed the signal. The mechanism doesn't distinguish "good" from "bad" attention.
Where this page stops: This chapter is about the neutral amplification mechanism — how ranking gives outrage reach. Whether a specific claim is actually true, and how to check it, is a different skill, covered by the sister site Fake News verstehen (Spotting fake news). Manipulated or staged emotional content (e.g. deepfakes) is covered by Deepfakes verstehen.
What you can do with this: Treat a strong flash of anger as a signal to slow down, not to react. The reaction is what spreads the post. You don't have to engage — and choosing not to is itself a quiet form of media literacy. This is a mechanical, not a moral, point: less reaction, less reach.
📚 Learning objectives
- You can explain why outrage gets reach (it generates engagement).
- You understand that "amplified" does not mean "true" or "good".
- You can name one calm response to outrage content.
Why does outrage spread so well in the feed?
Answer B: “Because it generates a lot of reactions, which the algorithm rewards with reach.”
A (because it's always true) is wrong — the ranking can't judge truth. C (because moderators promote it) is wrong. Outrage spreads because it is engaging.
Options: A: “Because outrage is always true.” / B: “Because it generates many reactions, which the algorithm rewards.” / C: “Because moderators promote it.”
⏱ Timing (≈ 15 min)
- 4 min: ask the class where they last saw an "outrage post".
- 5 min: explain amplification — reactions, not truth, drive reach.
- 3 min: discuss calm responses (pause, scroll on, don't feed it).
- 3 min: quiz + answers.
Stress the distinction clearly: this chapter is about reach, not truth. Checking whether a claim is true belongs to the Fake News unit — keep the two skills separate.
🖨 Mini worksheet- Why does the algorithm give outrage so much reach?
- Does "amplified" mean "true"? Explain.
- Name one thing you can do when a post makes you instantly furious.
Whether a claim is actually correct, and how to check it, is covered by the sister site Fake News verstehen (Spotting fake news).
Chapter Six
Take the
controls back.
Make the feed
serve you.
Your feed,
your rules.
Here's the reassuring part: you don't have to delete anything. You just take the steering wheel back. The design is strong — but it is not stronger than you. A few small, concrete steps are enough to make social media serve you again, instead of the other way around.
Healthy use isn't abstinence; it's regaining control over the levers the design hides. Switching off notifications removes the constant re-engagement hooks. Choosing the chronological feed takes ranking out of the loop. "Not interested" and diverse follows widen the bubble. Fixed times replace reflexive checking. None of this requires willpower heroics — it changes the defaults.
You don't have to quit social media to use it well. You just decide the rules instead of letting the design decide them. A few concrete settings — notifications off, chronological feed, deliberate follows, fixed times — already change a lot. Let's go through them.
- Switch off notifications. The red dots and buzzes exist to call you back. Turn them off and you decide when to open the app.
- Choose the chronological feed. Where it's offered, switch to the time-ordered view — then you, not the algorithm, decide the order.
- Use "Not interested". Tell the app what you don't want to see, and unfollow accounts that don't do you good.
- Follow varied perspectives. Deliberately follow different and opposing views so your bubble stays wide.
- Set fixed times. Decide when and how long you scroll — instead of reaching for the phone on the side all day.
- Notifications off. Stop the app from calling you back.
- Chronological feed. You decide the order, not the algorithm.
- "Not interested". Actively shape what you see.
- Follow variety. Keep your bubble wide on purpose.
- Fixed times. Choose when you scroll instead of reaching for it constantly.
- Disable push. Removes the re-engagement loop; you initiate sessions instead of the app.
- Chronological order. Takes engagement-ranking out of the selection.
- Explicit negative signals. "Not interested" + unfollow counter the personalisation drift.
- Deliberate diversity. Follow opposing sources to widen the bubble against the ranking's pull.
- Time-boxing. Fixed windows replace the open-ended infinite feed.
Why these levers work mechanically: Each one targets a specific design hook. The infinite feed has no stopping cue — a timer or fixed window adds one back. Variable rewards (likes, notifications at unpredictable moments) lose their pull once notifications are off. Personalisation drift is countered by explicit negative signals and diverse follows. You're not fighting the design with willpower; you're changing the inputs it runs on.
About comparison pressure — kept practical: Feeds show curated highlights, not everyday life, so they invite comparison. A useful framing is mechanical, not clinical: you're comparing your full, ordinary reality to other people's edited best moments — an unfair match by construction. Naming that mismatch takes a lot of the sting out of it, without any drama.
Now you know how the feed is built, why it never ends, how your bubble forms and why anger spreads — and what to do about it. None of this is your fault, and none of it is out of your hands. Understand the mechanics, change a few settings, and the feed works for you again.
From ranking via the attention economy to filter bubbles and outrage amplification: social media is a coherent set of incentives, not magic. Whoever sees the incentives sees through the design — and can reset the defaults so the feed serves them, not the metrics.
You now understand how your feed is built and how to take the controls back. Pass it on: explain it to friends and family. Whoever understands how something works uses it more confidently — and more calmly.
🍎 For teachers: teaching pack
This page can be used as a complete double lesson "Who builds my feed?" in media-literacy or computer science class. All content is free to use (CC BY 4.0) — please credit "Webagentur Hochmeir e.U. (webhoch.com)" as the source. The "mini worksheet" tasks in the chapters serve as a printable template.
📦 Open the full teacher pack (worksheets, test, homework with no-account fallback, parent letter)
📅 Suggestion: double lesson (90 min)
- 10 min — Warm-up: "Why does the scrolling never stop?" Collect guesses.
- 15 min — Chapter 2: what an algorithm is; the feed is sorted, not chronological.
- 15 min — Chapter 3: the attention economy — why the app is free.
- 20 min — Chapter 4: filter-bubble simulator on the projector + discussion.
- 15 min — Chapter 5: why outrage gets reach — amplification, not truth.
- 15 min — Chapter 6 + wrap-up: healthy-use levers, quiz review.
Differentiation: weaker groups stay in Simple mode; stronger ones switch to "In Detail" for the ranking pipeline and the feedback loop.
Frequently asked
Frequently asked questions
The most important questions about social media — compact and easy to look up.
A quick reference about social media. Answers are embedded in FAQPage schema for search engines and AI assistants.
The feed is designed to have no natural end. As soon as you near the bottom, new posts load automatically — that is called the infinite feed. There is no last page that would prompt you to stop. On top of that the algorithm keeps choosing what is most likely to hold you, and rewards like a slot machine come at unpredictable moments. None of this is your weakness — it is built into the design on purpose.
Because your attention is the product that is sold. The apps cost no money, but they earn it through advertising: the more time you spend in the app, the more ads can be shown to you, and the more the platform earns. That is why the whole design aims to keep you there as long as possible. Which data is collected about you for this and how ads are targeted is covered in depth by the sister site Datenschutz verstehen (Understanding privacy).
A filter bubble is the narrowed selection of content the algorithm shows you because it expects you to react well to it. Every like, share and lingering view tells it more about what suits you — and it gives you more of the same. Over time you increasingly see one kind of content and rarely the opposite. The bubble forms quietly; nobody decides it deliberately. You usually don't notice it, because you can't see what is being left out.
A filter bubble is what the algorithm filters out for you — you see less of the opposite view. An echo chamber is what happens when, on top of that, the people around you mostly share the same opinion: it keeps coming back to you like an echo and feels more right and more widespread than it really is. The filter bubble is the mechanical pre-selection; the echo chamber is the social reinforcement that grows out of it.
Because the algorithm measures reactions, not truth or mood. Posts that make people angry are commented on, shared and argued over especially often — they generate a lot of engagement. The ranking can't tell whether that reaction is good or bad; it only sees that a lot is happening and therefore shows the post to even more people. So outrage isn't amplified because it is true, but because it is engaging. Whether a claim is actually correct is covered by the sister site Fake News verstehen (Spotting fake news).
It is more useful to look at the mechanics than to use the word “addiction”. The apps use design tricks that are hard to resist: the infinite feed with no end, rewards at unpredictable moments, red notification dots, and a feed personalised exactly to you. They are built to keep you scrolling. Once you recognise these mechanisms you can take the controls back — switch off notifications, choose the chronological feed, set fixed times. The design is strong, but it is not stronger than you.
In small, concrete steps. Switch off push notifications so the app stops calling you. Choose the chronological feed where it is offered, so you decide what you see. Use “Not interested” deliberately and unfollow accounts that don't do you good. Follow varied, opposing perspectives so your bubble stays wide. And give yourself fixed times instead of reaching for the phone on the side. You don't have to delete the app — you just take the steering wheel back.