Recommendation Engines

Finding a great film or music you never heard about used to be happenstance. The next wave of music and film engines are making it easy for you to discover the long tail. It would have been impossible to find the music I like without Pandora. Unfortunately, it disappeared behind a copyright wall a year ago, when the music industry deemed that the site could only play the favourites of people residing within the US. Luckily there’s Last FM. Type in the name of an artist you like and you’ll arrive at a whole lot of others in the same genre without having to search for them.

Films have been a far more tricky proposition. People are loathe to spend two hours on a film they have never heard about. But Netflix developed an algorithm that pushed their rentals up by over 60% on films that were complete unknowns. They have tried everything in-house on trying to tweak the engine but got only incremental benefits. So they offered a million dollar prize in 2006 to anyone who could improve the recommendation by 10%. Hundreds of teams have given it a shot but the magical 10% goal has still not produced a winner. The problem apparently is ‘eccentric films’ the ones that polarise tastes of the people who watch them. They either hate it or love it.

Rotten Tomatoes is a great site to find obscure gems of films. But sites like Jinni are trying to make it even easier. Looks like we won’t have to rely on our friends to provide us even that favour. An algorithm wants to take that place.