When the Feed Knows You Better Than You Do

There is a strange moment that happens to all of us now. You think of something — a band, a worry, a pair of shoes — and the next time you open an app, there it is. It feels like being read. Often it is just math catching up to a pattern you didn't know you had.

Prediction is not understanding

A recommendation system does not know you. It knows the thousand people who clicked the way you click. It bets that you are like them, and it is right often enough to feel uncanny. But the version of you it builds is an average of strangers, sanded down to whatever keeps you scrolling. The parts of you that don't convert into clicks are invisible to it.

The danger is the loop

The trouble starts when the prediction becomes the cause. The feed shows you more of what you lingered on, so you linger on more of it, so it shows you more. Your taste narrows and the system calls it accuracy. Over months you can become a tidier, flatter version of yourself — not because anyone forced you, but because the path of least resistance was always one tap away.

You can keep a door open

The fix is not to delete everything and move to a cabin. It is to deliberately do things the algorithm wouldn't predict. Read the book nobody recommended. Follow the person you disagree with. Sit with boredom instead of feeding it. Each of those is a small vote for a self the model can't yet see.

The machine is good at telling you who you have been. It has no idea who you might become — unless you keep handing it the data, in which case it will happily decide for you.