for folks interested in how stuff works...
Pandora.com uses the music genome project's data to identify traits in songs. These traits are then matched up, so that when you say you like Manowar, they can play other songs with similar traits.
Here's the deal, in data, trait matching is a 2 way street. If you look for all items that have the same trait(s) as A and get B, then when you do the same thing for B, you should get A in the result set.
there are some exceptions to that, but as a general observational rule of how data and querying works....
So I made a Manowar channel to hear other bands like Manowar. they give me Iron Maiden, Dio, and Sabaton. Sabaton is new to my ears, so I made a new channel for Sabaton.
With Sabaton, I don't hear Iron Maiden, Dio, or Manowar, and instead hear other new bands like Edguy and Hammerfall.
the result set for either channel is such that I could list all the bands that appear (meaning it's is rather short and finite, rather than broad).
I find it odd, because I would expect that if Sabaton is "like" Manowar, then I should see a Manowar song crop up in the Sabaton list because at least one song from either band has like traits to the other.
Let's just say I've done extensive testing on this...
What I'm finding is you'll get about 8-20 bands that show up in a channel, and after a few days of listening, you generally won't hear any new songs, just repeats.
Obviously, I could add more bands to the channel to broaden it's mix, but given the commonality 2 bands have, elements of both bands should cause more variety.
There's something restrictive in Pandora's filtering that is not obvious, and thus limiting result sets.
Pandora.com uses the music genome project's data to identify traits in songs. These traits are then matched up, so that when you say you like Manowar, they can play other songs with similar traits.
Here's the deal, in data, trait matching is a 2 way street. If you look for all items that have the same trait(s) as A and get B, then when you do the same thing for B, you should get A in the result set.
there are some exceptions to that, but as a general observational rule of how data and querying works....
So I made a Manowar channel to hear other bands like Manowar. they give me Iron Maiden, Dio, and Sabaton. Sabaton is new to my ears, so I made a new channel for Sabaton.
With Sabaton, I don't hear Iron Maiden, Dio, or Manowar, and instead hear other new bands like Edguy and Hammerfall.
the result set for either channel is such that I could list all the bands that appear (meaning it's is rather short and finite, rather than broad).
I find it odd, because I would expect that if Sabaton is "like" Manowar, then I should see a Manowar song crop up in the Sabaton list because at least one song from either band has like traits to the other.
Let's just say I've done extensive testing on this...
What I'm finding is you'll get about 8-20 bands that show up in a channel, and after a few days of listening, you generally won't hear any new songs, just repeats.
Obviously, I could add more bands to the channel to broaden it's mix, but given the commonality 2 bands have, elements of both bands should cause more variety.
There's something restrictive in Pandora's filtering that is not obvious, and thus limiting result sets.