Friday, May 4, 2007

Pandora: Case Studies in Frustration

Here are some of the incidents that helped me determine the 7 Reasons Pandora Sucks.



















You can kind of figure out how they get from one selection to the next and that kind of predictability can be good, but it's also repetitive and boring. From Bruce Springsteen I get John Fogerty, Bob Seeger (obviously), Elvis Costello, Steve Miller Band, DMB, The Grateful Dead (really?) and some god-awful thing called Blessed Union of Souls that I want to skip but can't because I've exhausted my 6 song limit--again. Am I listening to a classic rock station because I thought I was listening to my own personal DJ.

From the Arcade Fire, I get to the Arctic Monkeys and some other bands that all sound kind of the same and not very interesting. Eleventeen? The Don’ts? Love Battery? A lot of these bands I would write off for their names alone. In about 40 minutes, I get 3 other listenable songs: The Arcade Fire, Stellastarr* and Guided By Voices. I’m not introduced to anything new I didn’t know I liked and I’m frustrated. Disgusted, almost.

If I make a station based on Fiona Apple or Joanna Newsom, that doesn’t mean I want to hear all female artists—quite the contrary, I rarely like listening to female singers. I like these two, though, and they are each quite unique and only they can pull off what they are doing. I don’t really want to hear Ani DiFranco, Joni Mitchell or anything else with a harp in it, either, thank you.

My most consistent complaint about Pandora was the sheer volume of absolute crap it played for me—regardless of the musician I started with. A radio is supposed to be some sort of filter and this type of filter is very limited. The problem with Pandora is they evaluate plenty of technical characteristics in a kind of objective way but there is no discernible quality control measure. Because quality in art is a subjective assessment, that makes it difficult to quantify.

One way Pandora might approach this is take into account the number of thumbs down or thumbs up ratings for the songs not just by individual user but total. Popularity measures always have their drawbacks and prevent less popular songs from rising to the top, but this might be one way to get some better songs played more. Also, one way I like to investigate bands is checking out the acts they have toured with in the past—that’s generally a good indicator. Artist's influences are also a great way expand in a historical way.

You would think that their "Quick Mix" mode might fix some of the problems of one-dimensionality but really, it just increases the types of bad songs you might hear. Maybe I’m really picky, maybe it’s the fact that I’m an avid media-consumer and I feel like I don’t have the time or patience to listen to anything I don’t like. Maybe Pandora just sucks.

Does it?

No comments: