Question 1

How do social media algorithms decide what you see?

Accepted Answer

Every major platform now uses a **ranking system** rather than showing posts in chronological order. The system predicts how likely you are to engage with each candidate post, scores them, and shows you the highest scoring ones.

**The general shape is consistent across platforms:**

**Candidate generation.** From potentially millions of eligible posts — accounts you follow, plus recommended content — the system first narrows to a manageable set of a few hundred or thousand.

**Prediction.** For each candidate, machine learning models estimate the probability of specific actions: that you will watch it, like it, comment, share, save it, follow the account, or hide it. Modern systems predict many separate outcomes rather than one score.

**Weighting.** Those predictions are combined using weights the platform sets. Shares and comments typically count for far more than likes, because they signal stronger engagement. Negative predictions — that you will report or hide it — subtract.

**Re-ranking.** Final adjustments apply diversity rules so you do not see six posts from one account, plus integrity demotions for borderline content.

**The signals that matter most** are usually behavioural rather than declared: **dwell time** (how long you stopped scrolling, even without interacting), completion rate on video, rewatches, and your history with that specific creator. What you actually do outweighs what you say you want.

**Two consequences worth understanding:**

**Optimising for engagement is not optimising for accuracy or wellbeing.** Content that provokes strong reaction performs well by these metrics regardless of quality, which is a structural rather than accidental problem.

**Passive consumption trains the system.** Watching something to the end tells it more than you may intend.

**What you can influence:** most platforms offer "not interested", follow and mute controls, and some offer a chronological or following-only feed — usually not the default.

Question 2

Do filter bubbles and echo chambers really exist?

Accepted Answer

The evidence is more nuanced than the popular account, and the honest answer is that both effects are real but smaller and more complicated than the standard narrative claims. The original argument:

Question 3

Why do people use social platforms as search engines now?

Accepted Answer

Because for a growing set of queries they answer the question better — and the shift is significant enough that search companies have publicly acknowledged it and redesigned results in response. What

Question 4

Why do posts with links get less reach?

Accepted Answer

Because a link takes the user off the platform, and platforms are optimised to keep people on it — so outbound links are ranked down, whether or not anyone says so explicitly. The incentive, stated

Question 5

What is an engagement pod?

Accepted Answer

A group of accounts that agree to interact with each other's posts — liking, commenting and sharing on a reciprocal basis — in order to signal popularity to a platform's ranking system. How they

Question 6

How do social platforms actually make money, and how does that shape what you see?

Accepted Answer

Almost entirely by selling attention to advertisers, which means the product being optimised is not your satisfaction but the time and engagement that can be sold — and nearly every design decision

Question 7

Why do online pile-ons happen?

Accepted Answer

Because the mechanics of social platforms convert individually reasonable actions into a collectively disproportionate one — and almost nobody involved experiences themselves as part of a mob. The

Related Questions