By night, Bob Coons is a jazz guitarist, playing at clubs in San Francisco. By day, he clocks in at an office, where he sits in front of a computer screen alongside other local musicians, listening to artists like Metallica and Barbra Streisand through a pair of bulky headphones and filling out a computerized scorecard detailing the musical nuances of each song.
Mr. Coons, 45 years old, is contributing to what Oakland, Calif.-based Pandora Media Inc. calls the Music Genome Project — an ambitious effort to create DNA-like blueprints for songs and then use them to make music recommendations based not on opinion, but on science. The company believes its music recommendations are so accurate that users will pay $36 a year for them.
Pandora’s approach is different from the recommendation techniques used by online retailers like Amazon.com Inc. and Apple Computer Inc.’s iTunes Music Store, which focus on buying patterns. For instance, Amazon.com suggests that fans of the latest album by crooner Norah Jones will like CDs by Diana Krall and Sarah McLachlan — but also the latest album by rock group U2, who sound little like Ms. Jones but make the list because a number of customers purchased both CDs.
“We kind of reject the idea of artists belonging to a particular social circle, and we want to break those barriers down,” says Tim Westergren, who founded Pandora six years ago.
By Vauhini Vara
More here.