There was a period of time when I frequented this site known as Last.FM. It doesn’t really do much for you, but it has a cute feature for your account which tracks the songs you’re listening to on your computer, and creates a playlist of them which you can stick on your signature. It also updates automatically, so your signature will always reflect your updated tracks. It would be snazzy for people to stumble upon mutual songs you like, though in my case it’s not very possible. It’s still cool to see what your friends happen to be listening to at the time, and sometimes by some stroke of chance you might be listening to the same thing!
But another feature of Last.FM, which I believe I’ve also used, is the “similar artistes” line. It’s pretty accurate, and tells you which artistes are similar to the ones you like. I believe UVERworld drew me to Flow and High & Mighty Color, and while I disliked the latter, the former definitely became a favourite with me. They base these similarity ratings on tags. For instance, if UVERworld is tagged as J-rock and alternative rock, the similarities list will include artistes also tagged as J-rock and alternative rock, with artistes that’re tagged as both (or basically as similar to UVERworld as possible) listed at the highest similarity level. I wonder, though, how accurate this can be. Have there ever been times when someone told you that if you liked so-and-so, you’ll most certainly also like XXX, and when you listen to XXX you think that it’s nothing like so-and-so at all? I know for me, it’s hard to tell whether I’ll like a song or not. It’s really unpredictable with me, and has nothing to do with whether it’s the same or different from what I listened to.
And I think people are indecisive like that when it comes to preferences. If a song is too much like the kind they’re used to, they don’t see the point of listening to this new one. If it’s too radically different, it “doesn’t suit their tastes”. It’s really a matter of fate and mood, not so much the music, that determines what they like. Social Psychology also brings in another point known as the Mere Exposure Effect, which explains why songs seem to sound better after listening to them multiple times. The Mere Exposure Effect says that you tend to like something better once you have been exposed to it even just once by mere contact. What this means is that if you’re asked to choose between a song you’ve heard once before and a song you’ve never heard before, you’re apt to choose the former just because you’ve heard it before.
So, really, do people really use the similarity tags to find other songs they might like, and are they always satisfied with these recommendations? It would be interesting to conduct a study to find out.