Your blog post
hyper-personalization
3/20/20251 min read


The concern that Spotify is feeding users repetitive and reductive recommendations is valid. Thats what happens when you outsource music reccs to tech. This hyper-personalization becomes incredibly impersonal and detached.
Spotify's ever-evolving algorithm is good but not a patch on Pandora's which was briefly available in the UK circa 2006 (sadly US only now). Pandora's recommendations are based on the "Music Genome Project" and does a pretty good job of capturing the essence of music at a more fundamental level, with any given song represented by approximately 450 "genes".
I once spent a day using Pandora to discover new music based on my fav artists (VH, Santana, Zappa, Todd, ABBA, Ozrics, Steely Dan) and genres (Metal, Prog, Funk, Fusion, Jazz Rock) and ended up buying about 20 albums in one transaction on Amazon. Powerful stuff.
This only works for discovering new music if you have broad, eclectic tastes. If you tend to operate in just a few genres like most people, you'll end up mining an ever disappearing seam.
So yes, Spotify is feeding users repetitive and reductive recommendations by using AI. Music discovery needs an informed, reliable, tasteful human filter in the process. Thats where the specialist press come in. I subscribe or recieve no less than 8 music mags every month. There's also the All Music new release email every Friday.
Here's my process for staying up with and discoveing new (including catalogue) music:
scan the music mag REVIEWS
only read those which score OVER 4/5 or 8/10
add those that sound interesting to your Amazon Wishlist
listen to them on Spotify
if theyre really good but them on vinyl
Ive been doing it for decades which is why Im blessed with constantly awesome sounds and the best record collection on the world. If its important enough to you, you will make the effort.
https://theweek.com/culture-life/spotify-wrapped-a-slave-to-the-algorithm