“Music recommendation algorithms were supposed to help us cut through the noise, but they just served us up slop.“
See also the linked blog post about the revival of interest in student radio. Lots of good news in there for us, and ideas about how we might be able to connect across generations to rebuild organic music community.
Anybody remember when Netflix had social features and a movie recommendation system that actually worked?
It was akin to Google’s original ‘ambient findability’ but applied to user likes instead of links.
It worked by looking at a movie you liked, then looking at all other watchers who liked the same movie, and surfaced movies they liked, collectively.
That but for music:
“Other listeners who liked Horse Rotorvator by Coil also liked Jasmine Nightdreams by Edgar Winter and Das Grosse Wunschkonzert by Heino”
Only better.
Intriguing.
There are some fairly minor privacy issues that would need to be addressed in the design. Especially when doing this in a decentralised network of services running Bandwagon, Mirlo, etc. Some way of building associations between songs/ albums that can’t be easily deanonymised to see what music you listened to and when, etc.
But this seems totally tractable, and I love the idea of building a human-centric recommendation system. Rather than the deeply manipulative stuff Spitify does with recommendations (see the article).
From the article on TheVerge;
“Bandcamp Daily has been a cornerstone of music discovery since 2016, and the site launched Bandcamp Clubs in 2025. These deliver one human-selected album a month, artist interviews, and live listening parties to subscribers.”
Being centralised and corporate-owned, it’s bound to enshittify sooner or later. But there’s a lot to be learned from what they’ve done so far.
On the privacy issue… Yeah netflix had no issue pulling it off, being centralized and preaumably aggregating across all users whether they wanted it or not… But in our case it would be interesting to see what happened in an opt-in world where you could also control whose recommendations you were going to aggregate. So in other words, I’d get to pick people I feel confident have good taste. Not sure this would really work or if it only worked well because Netflix had scale and the statistics of it worked out.