Being Less Random

I sometimes struggle with picking music. I used to be a radio DJ, so you’d think I’d be ok at it, and I am good at it. Definitely good live, but it does take almost the whole of my comprehension and attention. I like programming music with good segues and good sound, but the problem is I am inherently lazy. I believe in the power of random. Of course, the problem is random doesn’t work. Pure probability makes an eclectic but a very confused playlist. So what’s a music lover to do?

There are a few choices, and I’ll discuss three: itunes genius, last.fm, and pandora.


Actually, before all these choices is the radio. I do like listening in the car for new music, and I get that in spades mostly from KSCU and NPR. KSCU radio does a decent job of picking whole piles of music I’ve never heard of, but some DJs talk at length about the last party they went to, so it’s not a perfect solution. NPR’s All Songs Considered podcasts might as well be called “the music I will buy in the next month” sometimes; I highly respect their choices also. At home, though, I turn to my own collection of songs to find the gems.

I use itunes because I have an iphone. It’s as simple as that. I might use something else if I didn’t go with the iphone as I hear you can get away with different players with regular old ipods, but Apple’s pulled the propietary interface card for the iphone and like it or not I’m tethered to it. It’s not bad, mostly. A feature that’s come into its own is the Genius playlists. I like how it works — you pick some songs, tell itunes to make a playlist similar to those songs and it gets a big hunk of random that’s supposed to sound similar. I think this is a brilliant idea. The major flaw, I think, is that it seems to use the star rating levels to pick songs. I’ve never figured out a system to rate my songs, so the playlists tend to be very similar. I think the fallback choosing method is some combination of other people’s ratings and itunes sales… it’s all very non-transparent how the choices work. All I know is the music choices tend to be skewed in a radio release vein, as music that is very independent doesn’t get purchased nor rated in itunes as often as the big hit singles. I’ve all but given up on using it at home, so I turn to the internet for more choices.

I first heard about last.fm when I saw the word “scrobbling” dropped in a forum post somewhere. Normally I detest web 2.0 speak, but it sounded like an interesting service; you seed the playlist with an artist and based on recommendations from others it will generate a playlist for you. Sounds like a good idea, right? In the few days I played with last.fm I found it to be very hits heavy. I attribute this to the way the playlist is created from the listening public’s choices. Well, it turns out I’m a big snob. I don’t generally like other people’s music tastes. I think it comes from working at a college radio station, but again, I find myself annoyed with an oversaturation of “the hits” sometimes. I understand there’s a time and place for that, so I don’t abhor them completely, but I generally want my random music runs to be a mix of stuff I like and stuff that I haven’t listened to very much that I also like. I think basing the system on the aggregation of recommendations of other people develops the lowest common denominator rather than the equivalent of an evolutionary leap.

Enter pandora. Like last.fm and itunes genius, it starts with a seed of songs or artists. Then you vote up or down on the artist to influence the playlist. The major difference (and I think cool thing) is it bases the choices on similarities of the song’s structure, tonality, vocals vs. instrumentation levels, rhythm, and a whole host of other things that takes a more programmatic approach rather than a human recommendation model. Now that seems a bit impersonal in that it’s reducing the choices, but it definitely appeals to my music geek because it’s analyzing (with human help, I understand) the features of a song that I don’t think about, but I know that they influence how I like the song. I think it has more potential to deliver songs I like. The major reason is that because I do tend to have a focus on indie music, those songs will be delivered faster via pandora because the qualities that make the songs likeable by me are apparent immediately whereas with itunes or last.fm there has to be a groundswell of recommendations and ratings first.

While I probably won’t find my newest ukulele outings on any service, I will be sticking with pandora for a while and slowly making playlists for myself. If you have cool music finding methods or are a power user of genius playlists, last.fm, or pandora, I’d love to hear how you bend it to your will.

Leave a Reply