4 ms·
From my understanding, Spotify's recommendations have a major flaw - they're based off other users' listening data. What this means is that if a song isn't list
by Miner49er 8y ago
From my understanding, Spotify's recommendations have a major flaw - they're based off other users' listening data. What this means is that if a song isn't listened to by very many people it'll never be recommended, and many songs on Spotify don't even have a single stream, let alone however many it takes to get the data to make it a recommendation.
- drexlspivey 8y agoNot true, it's actually the exact opposite. See my other comment https://news.ycombinator.com/item?id=19503994 https://news.ycombinator.com/item?id=19503994
- Miner49er 8y agoNo, that is simply research they've done. They haven't implemented it AFAIK. https://www.businessinsider.com/how-spotify-discover-weekly-works-2015-9 https://www.businessinsider.com/how-spotify-discover-weekly-...
- hbosch 8y agoLuckily, the author outlines the fact that many of Spotify's playlists are curated by humans and when you submit an album to Spotify, you highlight a single track for a human to listen to. It is very possible to catch the ear of a human at Spotify and be placed on playlists, which is a spark that ignites more listenership.
- deegles 8y agoHow does one land a job as one of these playlist curators?
- plorkyeran 8y agoThey have a few openings listed on their job site, such as https://www.spotifyjobs.com/job/content-programming-manager-kids-and-family-oy3n8fwg/ https://www.spotifyjobs.com/job/content-programming-manager-...
- Lewton 8y agoI've had songs with less than 7000 listens recommended to me on my discover weekly
- sjapkee 8y agoNope. I find many artists who have 3-8k listeners per month through the daily mix and song radio.