You are signed in. Close

You are signed out. Close

Thank you. Your subscription has started. Close

Thank you. You will get an email when a new comic comes out. Close

We could not confirm your payment. If you paid, use the link "Sign in" at the bottom of the page. Close

Something went wrong on our side. Please try again in a moment. Close

Previous episode Embeddings: how a video feed turns you and every video into numbers

How TikTok and YouTube decide what you see next

Candidate generation: how a video feed finds a few hundred videos among millions

A video app has millions of videos. It has a fraction of a second to pick your next one. It cannot score every video, one by one, in that time. So where does it start?

  1. 1 The recommender picks your next video in two steps. The first one is quick and rough: out of millions of videos, it keeps only a few hundred.millionsa few hundred

    The recommender picks your next video in two steps. The first one is quick and rough: out of millions of videos, it keeps only a few hundred.

  2. 2 It has to be fast: YouTube's 2016 system gave this step tens of milliseconds, less than a blink. So which few hundred, and how does it find them that fast?tens ofmillisecondsa blink

    It has to be fast: YouTube's 2016 system gave this step tens of milliseconds, less than a blink. So which few hundred, and how does it find them that fast?

The rest of this episode is for subscribers

Subscribe to read every premium episode. You can cancel when you want.

Paid yearly: $99 instead of $119.88

Next episode Ranking: how a video feed decides which video comes first

Share this comic

More comics