How algorithms work.
The feed that feeds you isn’t showing you the world. It’s showing you what holds you.
“The algorithm” sounds like weather — something that just happens to you. It isn’t. A recommendation algorithm (the YouTube feed, the TikTok For You page, the timeline) is a machine doing one job, on purpose, every second. Here’s the job, in five plain steps.
1 · It optimizes one thing
Engagement — watch time, clicks, scroll-depth, time-on-app — because engagement is what sells ads. Not truth. Not your wellbeing. Its only goal is keep this person here longer.
2 · It decides by your signals
What you watch, how long you linger, what you replay, skip, like, or share — plus what people like you did (that’s “collaborative filtering”). From all that it predicts what will hold you, ranks everything, and serves the top.
3 · Nothing is random
Every single thing in your feed was chosen and ranked to maximize that prediction. When something “just showed up,” it didn’t — it was placed. (That’s why we don’t call it random here.)
4 · It learns in a loop
It watches your reaction and narrows toward what holds you. That’s the rabbit hole — and it’s why outrage and novelty win: they’re engaging, not true. The loop can wall you into a bubble without ever telling you.
5 · The catch
Because it optimizes engagement, it will happily feed you a lie if the lie holds you longer. Engagement is not truth, and it is not good-for-you. (This is the Tell — and the reason Spot the Lie exists.)
How to take it back
You can’t turn the machine off, but you stop being only its product the moment you read the meter. Notice you’re being ranked-to. Seek instead of only receiving — go get a thing on purpose. Doubt what’s handed to you, and check it. The feed predicts; only a human chooses — and only a human scores whether it was worth the minutes it took.
The algorithm decides what you see.
You still decide what you believe — if you remember you’re deciding.