For many web users, Pandora’s streaming radio was their first taste of algorithmic recommendation engines that adapted to the listener. Pandora has a lot of competition now, but it’s expanding into a new area: podcasts.
Starting today, a few Pandora beta users have access to personalized recommendations for podcasts based on the Podcast Genome Project. That’s an extension of the original Music Genome Project, the proprietary engine that drives Pandora’s recommendation engine by finding common elements in music that a single user enjoys. Podcasts are generally spoken word and much longer than individual music tracks, so they can’t be quantified in the same mathematical way that music can. But the principle of recommending content based on things that a user enjoys or doesn’t enjoy (the familiar thumbs up/thumbs down approach) is sound. Um, no pun intended.
The recommendation engine uses 1500 different data points to cater upcoming episodes for the listener, and it covers the wide range of genres in the medium, from This American Life to Questlove. That being said, Pandora’s talking about how it has “partnered with top-tier publishers,” which probably means that getting your indie Beyblade enthusiast podcast onto the actual Pandora app won’t be as easy as it is on platforms like iTunes and SoundCloud.
Pandora is accepting early access applications at this link. The press release says that the full podcast functionality will be widely available “in the coming weeks.” Podcast listening is only available on the mobile app for now.