How Social Media Algorithms Work
Ranking Signals, Engagement Loops, and the Recommender Systems Behind Your Feed — A TLDR Primer
Your feed feels like it knows you better than your friends do. This TLDR primer explains exactly why — no computer science degree required.
If you've ever wondered how TikTok decides what you see, why Instagram keeps showing you the same three accounts, or why one YouTube video turns into three hours of rabbit-holing, this guide walks you through it in plain English. You'll learn what a ranking algorithm actually is (spoiler: it's just a scoring system for millions of posts), the concrete signals platforms track about you and your content, and how collaborative filtering and neural networks combine to guess what you'll tap next.
It also covers the parts platforms don't put in their help center: why optimizing for engagement naturally produces filter bubbles and outrage spirals, what shadowbanning and content moderation really do behind the scenes, and how to read your own feed more critically once you understand the machinery behind it.
Written for high school and early college students — and just as useful for a parent, teacher, or curious adult who wants the real explanation instead of a hot take — this book is concise and to the point. No padding, no jargon left undefined, no chapter you have to skim through to find the one sentence you needed. Just the ideas, explained clearly, with real examples from real platforms.
If you want to actually understand the algorithm instead of just complaining about it, start here.
- Explain what a social media 'algorithm' actually is and what problem it solves
- Identify the key ranking signals platforms use to score posts
- Describe how collaborative filtering and content-based recommenders work at a high level
- Understand engagement-optimization feedback loops and their consequences (filter bubbles, virality, misinformation)
- Reason critically about your own feed and the tradeoffs platforms make
- 1. What a Feed Algorithm Actually IsFrames the algorithm as a ranking problem: given millions of candidate posts, score and order them for one user.
- 2. Ranking Signals: What the Algorithm Sees About You and a PostBreaks down the concrete features platforms use — user signals, content signals, context signals, and interaction signals — with real examples from TikTok, Instagram, and YouTube.
- 3. How Recommender Systems Learn: Collaborative Filtering and Neural ModelsExplains the two main families of recommender algorithms — collaborative filtering and content-based — and how modern platforms use neural networks and embeddings to combine them.
- 4. The Feedback Loop: Engagement Optimization and Its Side EffectsShows how optimizing for engagement creates filter bubbles, rabbit holes, virality spikes, and amplification of outrage or misinformation — and why this is a structural feature, not a bug.
- 5. Moderation, Shadowbans, and the Rules Layered on TopCovers the non-ML rules layer: content moderation, demotion, shadowbanning, and how platforms adjust the algorithm for safety, ads, and legal compliance.
- 6. Reading Your Feed Critically: What This Means for YouPractical takeaways: how to interpret what your feed shows you, how to influence it, and what open questions researchers and regulators are debating.