Twitter recently launched a social music discovery site, music.twitter.com. It’s pretty interesting. There have been a variety of music discovery music sites launched recently, including sites like Spotify, Shuffler.fm, ex.fm, Turntable.fm, and a host of others. Only a few of them have an explicit social angle. Spotify and Turntable are probably the most social of the lot. I’ve long since abandoned Turntable like everybody else (talk about yesterday’s darling!), whereas Spotify is one of my primary music consumption environments – although, the social piece isn’t really why I use it.

Twitter music, unsurprisingly, draws its inspiration from your twitter stream, and the larger activity happening on twitter. As a result, it’s naturally personalized, but also leverages all the activity happening in the twitter-verse. And draws on a broader crowd than Spotify (because it uses Twitter, where I have a large social graph, vs Facebook, which is more tightly controlled).

I think they’ve done something simple and elegant, and I’ve gravitated to it faster than many of the other music discovery services.

Basically, they present a vaguely pinterest-style view into songs, and the songs are selected according to one of a few primary criteria, and then ordered by recency. In particular, they show songs that are:

• Popular (new music trending on Twitter – essentially, a time-bounded popularity metric)
• Emerging (hidden talent found in the tweets – I am not 100% sure how they are doing this, but I expect they are looking for songs with a high-velocity recent tweet stream)
• Suggested (Artists you may like – presumably artists that are similar to artists I follow, see below)
• Now Playing (music recently tweeted by people you follow)
• Me (music from artists I follow on twitter)

All of these options could readily apply to other forms of content discovery – e.g., a books or reading discovery engine. But whereas Twitter music naturally generates a “stream” of music to listen to, all the books recommendations engines produce more of a static “if you like x you’ll like y” recommendation. For example, see Bookish, recently launched with much fanfare, but with a lukewarm response from the books community. It’s a lot of work to get recommendations out of Bookish – I have to come up with a book I like and get recommendations against – vs. a much stronger “push” user experience.

I found the “Suggested” music of Twitter Music interesting, but had a hard time figuring out why the particular songs were recommended. I expect it’s because they were from artists similar to artists I liked, but it wasn’t clear. Twitter Music could use a “why was this recommended” button. The “Me” music, music from artists I followed, I found really fun. But, it seems quite static. In the 3 days I’ve been playing with it, the music hasn’t changed. Not sure why that is.

A downside of Twitter Music is it designed to use with a companion music player, in particular Spotify or Rdio. With Spotify, you need a Premium account to stream the songs – I found twitter music sufficiently interesting I sprung for a Premium upgrade to Spotify – an extra $5/month over my Unlimited membership. Interesting. I wonder if Twitter gets a spiff on that?

Here’s a couple of screenshots:

Overall, I found Twitter Music to be a surprisingly strong offering – simple, elegant, and “right” surprisingly often. I hope that they don’t build Twitter Books, because I want to!