Los indios tabajaras… that’s a name I’ve not heard for a long time. Now that I’ve been reminded, I’ll have to go see if I can find some.
1 Like
And yet somehow the completely meaningless label of “alternative rock” that connects them still manages to be more relevant than whatever bit of data got the other music put on that map.
I’ve always been disappointed by human attempts to make recommendations based on similar work, but it turns out the machines are so much worse at it…
3 Likes
If you want to do that sort of thing, you need boil the frog
It tries to create a playlist that goes from one artist to another in steps of similar music tracks, so that you can plausibly start listening to one and end up at the other.
1 Like
Who is Al, and why should I trust his recommendations?
2 Likes
AI (愛) is love.
The original Pandora engine was based on human extracted song features. Those features make a multi-dimensional space based on the way the music sounded and the suggestions were about finding songs that sound similar. If you were careful in tuning your channel, it was great for discovering smaller bands. The original hope was for Pandora to give greater visibility to unknown artists. It would allow them to make more of a living (middle class musicians). Then Spotify (and others) creamed Pandora in the market. Now the default on Pandora is to copy Spotify and the rest: “people who liked X also liked Y”. Why? Sadly, it makes more folks happy than helping them find new artists.
I’ve worked with the public Spotify data a good bit. There isn’t enough in there to get much more than “people who liked X also liked Y”. And that really makes tools like this less exciting than they should be. You only discover a new band if it was popular but you happened to not know about them.
Side note: Pandora does still have a Discovery mode and I think it uses the old engine. Granted even that method has a limit. I remember meeting the original Pandora CEO and he talked about how artists like Frank Zappa gave the system fits. There just weren’t many artists that sounded like Frank. He also had a funny story about a customer complaint. The customer didn’t understand why the system would keep suggesting Celine Dion. The company looked at what the customer had ‘liked’ in the channel. They said “You seem to like female singers, whose songs have features X, Y, and Z.” The customer finally said, “OMG, I do like Celine Dion!” What makes songs appealing is so tough, especially when there are non-music biases mixed in.
2 Likes
Sadly, it makes more folks happy than helping them find new artists.
Human nature in a nutshell. People like to feel safe (even if that feeling is entirely built on illusory ground.) Being told “people who liked X also liked Y” makes one feel as though one is part of a crowd and that inherently feels safer than standing out. Even when it’s something as individual as musical taste.
I can’t find a band called “Living at the end”…
There is band called The Living End and A-ha has song called Living At The End Of The World.
That tag map in the OP is like a list of concerts I’ve been to over the last couple of years: In the past couple years: Duran Duran, The Cure, Kraftwerk, Psychedelic Furs, New Order (which also kinda counts as Joy Division), Pet Shop Boys, Tears for Fears, Depeche Mode, Midge Ure (which kinda counts as Ultravox and Visage), and OMD. I’ll be seeing Echo and the Bunnymen, Peter Hook (which also counts as New Order and Joy Division), and Gary Numan next year.
I’ve been playing with this site and it’s actually pretty neat. I’ll definitely be using to it to find new music to listen to based on links to groups I’ve got on heavy rotation.
1 Like
This topic was automatically closed after 5 days. New replies are no longer allowed.
