Study Takes Critical Look At Music Recommendation
(UPDATED) In "Music Taste And Web 2.0" witten for the Music Management & Media program at the University of Gloucestershire, Joel Stamp did extensive research and analysis of how consumers discover new music. His work comes just as iTunes Genius, Google Instant Mix, The Echo Nest and others are bringing machine generated playlists to market. Others, like Pandora, take a hybrid approach, adding a human touch to data.
Current industry thinking holds that fans don't want to put extensive effort into music discovery. There's just too much music and too little time. But Stamp's research shows that may not be true. "A perfect music recommendation service neither exits or is truly desired by music fans," he states.
"Music Taste And Web 2.0" investigated how participants view music recommendation services in relation to other sources of music exposure. A key finding showed that, particularly for committed music fans, the joy of discovering music "lies in the labour of the task". In fact, the real appeal of using music recommendation services, suggests Stamp, may be precisely because they aren't perfect and users still have to sift through bad suggestions.
"Services are falling over themselves to claim that they are the perfect solution to consumers’ music discovery needs," concluded Stamp. "But making a perfect music recommendation service for all users can’t be done." The realization, that machine music recommendation is a tool rather than a solution, could shape both how fans perceive and use these services, as well as, how they can be improved.
Read the full study here.
Joel Stamp has finished school and can be reached via LinkedIn.
MORE: Echo Nest Compares It's Playlist Engine To iTunes Genius, Google Instant Mix
interested to take a further look at the study when time opens up, but upon reading “the joy of discovering music “lies in the labour of the task,” i can’t help but make a mention of how we’re crowdsourcing the music discovery process at http://gorankem.com
although the resource is still in it’s relative infancy, the results improve with each successive contribution as the sample from each artist’s fanbase continues to build.
sifting through all of the noise is only going to get worse, and we’re hopeful the GoRankem can truly become the ‘cliffsnotes for the catalogs’ to help us find all those wonderful songs we were always meant to hear!
“A perfect music recommendation service neither exits or is truly desired by music fans,” he states.
On the one hand, not everybody knows what they want till they have it, like Twitter.
On the other, I’m less interested in music recommendation systems than in music identification systems. Don’t know if that’s what they’re called but it’s the software that allows your mobile phone to identify music in the environment. So when that ill bit of music flashes by in some loud car, you can find out what it is. That’s what’s always eluded me and I think it would be cool to have that.
Music from loud cars is rarely interesting. 😉
At Epitonic.com, we’ve always taken an editorial approach to our music recommendations, relying on the tastes of our editors to help people discover new music. Our daily playlists provide a smattering of new tunes and then users can check out similar artists. All the thousands of tracks are free and legal to download. Users can even make their own playlists and share them with friends.