How Algorithms Decide What You See

How recommendation algorithms rank posts and videos to maximise engagement, the shift from friends' posts to algorithmic feeds, and concerns about filter bubbles.

Social media feeds are shaped by algorithms.

How they work

Algorithms predict what you’re likely to engage with: watch, like, share or comment.
They use data on your past behaviour and similar users.

From friends to algorithms

Early feeds showed posts from friends in time order.
TikTok’s “For You” page showed the power of algorithmic recommendations from anyone, and other platforms followed.

Goals

Algorithms usually optimise engagement and time spent, which drive ad revenue.

Concerns

Filter bubbles: seeing only content that matches your views.
Outrage and extreme content can get more engagement.
Opacity: users don’t know why they see content.

Research

Studies on filter bubbles show mixed results; exposure is complex.

Regulation

The EU’s Digital Services Act requires large platforms to explain recommendations and offer non-personalised feeds.

The recommended video

After watching one cooking video, a user's feed fills with recipes, as the algorithm predicts more engagement.

Thinking feeds show posts neutrally

Algorithms rank content to maximise engagement.

Key takeaways

Algorithms predict and rank what you'll engage with.
TikTok showed the power of algorithmic feeds.
They usually optimise for engagement.
Filter bubbles and opacity are concerns.