What is the principle that Tiktok Algorithm uses to engage & Spotify Algorithm to personalize ?
Patrick Fagan of Capuchin Behavioural Science explains the Principle of Thin Slices
The underlying principle of thin slices is that if you saw a very thin slice of cake, you could tell quite accurately what the whole cake looks like. And then the same way, if you see a thin slice of someone’s behavior, like their Facebook profile, the Spotify plays, the picture of the bedroom, you can tell quite accurately what they’re like as a person based on that thin slice.
If you can get a thin slice like a digital footprint, you’ll be able to tell what that person’s personality is like, and therefore predict other things about them.
Spotify is using the principle of thin slices to send personalized recommendations and notifications. And it’s the best practice of personalization, better than Netflix. There is all sorts of amazing things they do. They extract features from songs like the tempo, the speed, and based on that one can decipher the genre quite easily. They can pull out not just the genres, but also the mood. Is it for ambience, is it upbeat?
Tiktok engages , by doing the same thing as Spotify, it extracts features. Tiktok is able to extract visual features. There are different clusters of popular content. One that is very popular is basically muscular shirtless man. The other clusters are dogs, groups of people dancing etc. We have found about over 100 of these clusters. Tiktok gives you what you want. And so it can if you’re watching something and you’re engaging with it, Tiktok will take all the features from that content, and the algorithm will go and look for things that are similar and feed it to you.
Tiktok focuses on grab our attention and engaging us emotionally. So cute puppies and kittens has worked very well. Any kind of anything that’s a bit surprising or weird also gets you to engage with the content and keep watching.
Listen to Patrick Fagan’s Full Conversation on Youtube talks about on application of neuroscience for understanding minds, optimizing messaging and nudging behaviour.