Matching Music In The Phantom Tail
The rise of digital music and long tail economics was heralded by some as the beginning of a golden age of possibility for independent music. However, the pull of the similarity vortex means that the long tail may actually be a phantom tail…
No Crowds, No Wisdom
As the amount of user-generated content on the Internet increases, it becomes ever more important to have effective filtering mechanisms.
Unfortunately, many music recommendation systems rely on collaborative filtering – the typical ‘if you like x, then you might like y’ approach.
This can be quite effective for popular artists or products, but is quite useless for introducing new acts that have yet to establish a significant listenership.
Paul Lamere provides an excellent summary of the problems facing recommendation systems over at Music Machinery.
Stuck In The Head With You
The iTunes store recently reported that about 80% of its inventory failed to sell even one copy last year. Neilsen Soundscan reports that 80% of sales in 2007 were from only 1% of the 4 million tracks available.
So, the long tail for music is really a phantom tail – we still feel as if it’s there, even though it has effectively been cut off by a combination of current music recommendation systems and its own innnate obscurity.
As Lamere points out, CF systems need an artist to have a critical mass of listeners before they can break out of the long tail. He cites an analysis of 240,000 artists by Dr. Oscar Celma of MTG UPF:
…even though there are nearly a quarter million artists to chose from, 48% of all recommendations are drawn from a pool of the 83 most popular artists. The other 52% of recommendations are drawn from the mid-tail. No recommendations at all bring you to the long tail.
Sell Me A Song
In fact, the long tail is only really a viable model for retailers or wholesalers – selling less of more might work, but for an unknown artist this would require the creation of an enormous back catalogue.
Digital music stores can afford to offer an infinite supply of music, because it costs them nothing to stock it – and they make a profit off the handful of bands that sell one download a year. For the store, it all adds up – but for the bands, they have still only sold one track.
Of course, it is possible for independent artists to make a living from a relatively small fanbase – if you don’t have any label to pay and have control of all production and distribution, you need a much more modest audience than in years gone by.
However, even this level of success requires breaking out of the phantom tail and into the lower reaches of the midsection – so let’s hope that music recommendation systems evolve to allow newcomers a better slice of the pie. Having said that, it’s certainly no easy task – if humans have such difficulty determining quality in music, then it’s a lot to ask of a machine…