I just started using Last.fm yesterday.
“What?” you say. “What kind of self-respecting Web 2.0 kid isn’t using Last.fm?”
Well, for a while, I wasn’t. I love me my music, don’t get me wrong. I have 6600 tracks in my iTunes Library. I like to crank it with the rest of ‘em. But my music has been stored on my home eMac which is generally not connected to the web often (you need your machine to be connected to take advantage of Last.fm). Since I so often listen to podcasts to and from work, I generally only have a very few albums on my work machine at a time. So, I was never in a situation to get much out of Last.fm, so I didn’t use it.
Well, my new MacBook Pro affords me the ability to add some music to the work machine. So, I copied over about 2700 tracks in the last couple days of my very favorite albums (I know, that’s a lot of favorites). But over 425 alone are Teenage Fanclub and Yo La Tengo. And that’s not including all of them. So, it adds up quickly.
So, what does Last.fm do? When you are web-connected and listening to music, you run the Last.fm app and it records what you are playing on iTunes (I’m sure it works with other music-playing apps, but seriously… what others are there?). It then uploads that data to the Last.fm server. You are able to view charts of your recently played tracks and artists as well as tracks and artists played over time.
But like any hip Web 2.0 site, Last.fm “harnesses the collective intelligence” of the users in order to provide additional features. The first, of course, is a similar artists feature. This isn’t an arbitrary music critic choosing who he thinks is most similar. Actual listening patterns are determining the suggestions.
How is this different than Pandora? Pandora allows you to select an artist or song and it then serves up similar music—as determined by the members of the Music Genome Project. You can guide the recommendations by giving a thumbs up and thumbs down, but you are really going by what the Genome Project thinks. Last.fm builds the entire recommendation system totally based on what you listen to.
Another feature is, of course, a friends feature. In fact, if you visit any profile of a user, there is a “taste-o-meter” that shows how compatable your musical tastes are with that person. For example, here’s me rated against my pal Nick:
Every time you visit an artists page, it also shows the top listeners to that band. I find it interesting to poke around and see who people’s #1 artist is. I love that Caterina of Flickr has Mogwai as #1.
I found Last.fm pretty interesting because I’m working on a couple projects that specifically deal with software-generated recommendation systems. One of them I’ll be writing about on the CogBlog soon, and the other… perhaps I’ll be able to write about it down the line. But Last.fm has been more than just handy. It has inspired some ideas of how to handle these serendipitous recommendations.
If you use Last.fm or decide to sign up, here’s my profile.
Update: Just noticing that some of those tracks in the image in this post were listened to when I was not connected. So, it tracks your offline plays and uploads when you connect. So, I could have been using it all this time! D’oh!