Chapter
Creating a Playlist to Sing to in the Car
1:06:48 - 1:14:31 (07:43)
A product manager discusses the process of creating a playlist and identifying a product need for something like "songs to sing in the car". They curate a list of thousands of popular songs and come up with a concept for a playlist.
Clips
Product managers define the concept behind algorithmic decision-making, such as creating a playlist of the best songs to sing in the car, that machines cannot do on their own.
1:06:48 - 1:10:43 (03:55)
Summary
Product managers define the concept behind algorithmic decision-making, such as creating a playlist of the best songs to sing in the car, that machines cannot do on their own.
ChapterCreating a Playlist to Sing to in the Car
EpisodeGustav Soderstrom: Spotify
PodcastLex Fridman Podcast
Building a taste vector for music is a difficult task as music preferences are hard to specify concretely.
1:10:45 - 1:12:05 (01:20)
Summary
Building a taste vector for music is a difficult task as music preferences are hard to specify concretely. It requires a lot of data and is not a stationary problem as it changes over time. Handcrafted heuristics and latent factors in playlist data can help in building taste vectors.
ChapterCreating a Playlist to Sing to in the Car
EpisodeGustav Soderstrom: Spotify
PodcastLex Fridman Podcast
The process of finding the principal components to accurately predict mood for Spotify or other mood-based products is challenging and involves dimensionality reduction, which may not be easy in the engineering sense.
1:12:05 - 1:14:31 (02:26)
Summary
The process of finding the principal components to accurately predict mood for Spotify or other mood-based products is challenging and involves dimensionality reduction, which may not be easy in the engineering sense.