Words Matter


For the past few days I’ve been trying out some of the music recommendation services out there. I really like the idea that my computer can help me find new music to listen to. When I was using Last.fm I noticed that the words they use for rating music really don’t fit with my mental model at all. As you can see in the image below, Last.fm asks be whether I “Love” the song, or if they should “Ban” it. What if my opinion is somewhere in the middle!? This rating scheme is just too extreme. Or maybe it’s just a wording issue.

LastFM02

On the other hand, Pandora, uses language that is simple, and matches my mental model. While listening to each song, I can either tell Pandora that “I like it” or “I don’t like it”. Of course, I’d still like to rate it somewhere in the middle, but still, the less harsh wording encourages me to give my opinion. When the words are harsh, I’m less likely to agree with the system. That means I’ve been ignoring Last.fm’s rating feature altogether!

Pandora02

That’s just my two cents on the matter. Soon enough I’ll be heading over to MyStrands to give Justin‘s employer a go at this whole music recommendation thing. ๐Ÿ˜‰

, , ,

2 responses to “Words Matter”

  1. LaunchCast (http://launch.yahoo.com) was the first of these I’ve ever tried though I tend to use Pandora now given the embedded player in Yahoo Messenger isn’t working and was somewhat unstable.

    A cool, relatively recent feature of Pandora is the “I’m tired of this song” so don’t play it for a month action. Sometimes, I like the song but I just don’t care to hear it for the next while.

  2. Hey Josh, cool post and graphics. I second your issue with “mental models” and “rating schemes”. I’d probably say that no rating scheme is perfect, but most are better than nothing. MyStrands has a rating feature that you can use on the website, but the ratings aren’t an integral part of the recommendation process.

    Rather than harvesting tenuous “like/dislike” rating data, we harvest playlist based association data. I’d be interested in hearing what you think about the strengths/weaknesses of both approaches… be brutal… well maybe not too brutal ๐Ÿ™‚