We tend to imagine persuasion as a debate. Somebody makes a claim, you evaluate it, and eventually your mind changes.
That model misses a quieter form of influence: deciding what repeatedly enters the room.
An algorithm does not need to make you believe that crime is rising. It can simply put ten dramatic crimes in front of you this week. It does not need to persuade you that everybody is beautiful, wealthy and productive. It can fill your idle moments with people selected because they are unusually good at looking that way.
Nothing in those examples has to be false.
That is the interesting part.
Attention comes before judgement
You cannot evaluate information you never encounter. Before arguments compete on truth, enormous systems compete on selection.
What gets ranked high enough to be seen? What is omitted? What is repeated? What arrives while you are bored, lonely, angry or half asleep?
This is why media literacy should involve more than fact checking. A perfectly factual stream can still produce a distorted model of reality if selection is systematically unrepresentative.
Ask about the frame
When a stream of information produces a strong impression, do not only inspect each item. Inspect the set.
What category is overrepresented? What denominator is missing? Which boring events would have to be visible for the picture to become representative?
The algorithm does not have to win an argument with you.
Sometimes it only has to choose the agenda.