Listening to Pandora’s Box

PandoraWhy do we like the music that we like? That is the question explored by Rob Walker in a great piece in this week’s New York Times Magazine about the Internet radio station Pandora. Pandora attempts to deconstruct the music you like and find similar songs to match your taste, without any nonmusical filters whatsoever.

I’ve already sung the praises of listening to the radio, but that has some major drawbacks: annoying DJs, repetitive set lists, lots and lots of commercials. Pandora, however, presents itself as something of a solution to these problems (they still have commercials, but only like 15 seconds worth every four songs, not the five minutes worth every three on the radio). 

Here’s how it works: You go to the website (or you get their app, which is more common, but my cell phone is from late 2008, so I can’t do that yet) and enter a song* or artist you like. From that, Pandora will construct a radio station around similar songs. You can approve or disapprove of every song, and with each judgment, the radio station refines its idea of your taste.

*Another interesting tidbit from the article is that Pandora will not play the actual song that you enter because of how much the licensing fees go up for being an “on demand” music playing website. 

What’s so intriguing about Pandora is the way that it tries to balance subjective and objective judgments. Unlike a lot of programs like this, Pandora does not rely solely on digital analysis of melodies and harmonies. They have musicians rate songs for things like how gravelly the voice is and even how “emotionally intense” a guitar solo is. In fact, the Music Genome Project, which is how the songs are described and analyzed, includes nearly 400 attributes, ranging from “avant-garde leanings” to “violent lyrics.” 

If you put in “Disturbia,” for example, Pandora will tell you to listen to “Hot N Cold” by Katy Perry because it has “basic rock song structures, repetitive melodic phrasing, major key tonality, a vocal-centric aesthetic and vocal harmonies.” This is a very interesting blend of judgments. It seems straightforward enough to pick a song with “basic rock song structures” (I mean, pretty much any pop song is going to have that, right?), but how exactly is “a vocal-centric aesthetic” judged? I don’t disagree with the assessment, but it doesn’t seem like something that can be quantified.

It’s good that Pandora doesn’t ignore these judgments: Although I’d never put it that way before, describing “Disturbia” as having “a vocal-centric aesthetic” makes perfect sense. Sometimes these difficult-to-measure qualities of music are the most fundamental aspects of a song. Just because they are difficult to measure or explain doesn’t mean they are impossible to gauge.

Pandora seems to employ some kind of intersubjective method of calibrating these things. They have musicologists listen to the songs and rate them from objective to subjective qualities. While these people are making some subjective judgments, such as how “risky” the percussion is, they are not making such judgments about the overall quality of a song; in other words, they are not judging it according to their own tastes. Their goal is not to tell you if a song is good or bad, but to tell you what a song sounds like.

This is an important distinction, since Pandora aims to be a radio station without any taste-masters or gatekeepers. There are no “experts” telling you what the “right” or “cool” music to like is; the connections are based entirely on music. If you put in a hip band and get back a lame one, it’s because the songs sound alike, in at least some way. It’s possible that you won’t like a suggested song, or that you’ll disagree with a description (I don’t know, for example, if I would say that “Come Together” has “great lyrics”), but they are at least ostensibly connected in some sonic way.

Of course, many of the reasons people listen to music are nonmusical, and that’s not necessarily a bad thing. If a friend suggests an album I may like, or if I listen to a song because it brings back good memories, then those aren’t strictly “musical” reasons for listening to something, but they are still good. Listening to music because it’s one of the five songs currently being played on the radio, or because it’s the most famous one by an artist, or because it got a positive review by a “hip” critic, or because you want people to know you like “cool” music, etc. are all pretty bad reasons to choose music. Selecting music based on musical qualities, whether it be the “repetitive melodic phrasing” of “Hot N Cold,” or the “clear focus on recording studio production” of “A Day in the Life,” is bound to result in a better listening experience.

One response to this post.

  1. Posted by James Schneider on October 22, 2009 at 5:25 PM

    I’m probably misunderstanding this, and its kind of weird since its music, but i think this falls under the argument we had about quantifiability.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: