4 min read

How to train your feed by leaving it

The most useful thing you can teach your algorithm is that you do not actually live there.

The big platforms all train their algorithms on what you click and how long you stay. The longer you stay, the more they show you the kind of thing that keeps you staying. Every signal you give them is a vote for the next thing they'll show you.

Most of the advice on "fixing your feed" is some version of consciously curating it: muting accounts, hiding posts, training the algorithm by tapping the right buttons. This works, sort of, but it's a losing game. The platform's incentive is misaligned with yours. You're trying to make it less compulsive; it's trying to make you more.

The simpler move is to leave more often. Not forever. Just for a few minutes. Open a different tab. Read a personal blog. Play a tiny game. Click a random link. Then come back when you actually have something to look up.

Every time you do this, you give the algorithm one piece of information: I'm not always available. Over weeks, that adds up. The feed quietly gets less aggressive about you, because the model has noticed you're not a reliable engagement source.

You can't really win against a recommendation system on its own turf. You can win by spending less time there. Same outcome, less effort.