I thought I’d share an interesting term I picked up last week from John Ashenden of Grooveshark: The Coldplay Effect

As an Internet radio listener, your like/dislike history and recently played music are used to determine which songs to play next (it’s obviously much more complicated than that, but I’m speaking in generalities here, so back off!). Your most recently played songs are weighted the heaviest. That makes sense. It is of little consequence that you like The Beach Boys when you’ve been listening to Deadmau5 and the Aphex Twins for the last two hours.

This data is compared to songs in the provider’s catalog and (again, using a complex algorithm) songs are picked based on their level of “likeness” to what you currently deem play-worthy. I’m not sure exactly how song likeness is determined, but it’s probably a combination of genre, beats per minute, and many other characteristics that can be expressed mathematically.

The Coldplay Effect is an unfortunate phenomenon that arises when certain artists’ music (like Coldplay’s) ranks high for likeness across diverse genres. It doesn’t matter that you are on an electronic trance binge, because some Coldplay song is just enough like the third track on the Fight Club soundtrack to bring it into your queue.

Once the Coldplay track plays, it’s all over. Your previously specialized stream has been normalized and pretty much anything can play next.

This seems like a Hard Problem™, and I don’t think anybody has solved it just yet. Perhaps this — along with other problems faced by programmatic music recommendation engines — plays into the recent praise of TurnTable.fm, which uses humans to pick which songs play next.