Technology

Apple Music Has Got the Beat But Not the Algorithm

By Paul Sweeting

Apple is thinking different about music streaming.

As Apple prepares to take on Spotify, Pandora and their many rivals in the streaming media sweepstakes with the launch of Apple Music later this month, it’s looking to differentiate its service through an innovative use of data-driven recommendations: It isn’t using any, at least not openly.

In unveiling the service Monday at its World Wide Developers Conference, Apple executives focused heavily on the human element.

“You’re in a special moment, exercising… and your heart’s pumping. You’re about to turn up the reps. And the next song comes on and… BZZZZT Buzzkill,” Apple Music chief Jimmy Iovine said from the stage, describing the “typical” user experience with programmed playlists on other streaming services. a’Why did that happen? It was probably programmed by an algorithm alone, that’s why. Algorithms alone can’t do that emotional task.” Iovine co-founded headphone maker Beats, which was acquired by Apple last year for $3 billion, along with Beat’s fledgling music streaming service.

Instead of algorithms, Apple Music will rely on human curators to try to improve the user experience. The on-demand streaming tool will allow users to select their own music, from the own collections as well as Apples vast online library of tens of millions of tracks. But it will also let you enable programmed playlists and let Apple’s in-house curators select the music.

“No machines, just recommendations made by real people who love music,” Apple’s senior VP of Internet software and services Eddy Cue said.

Apple’s new online radio station, Beats One, also relies on human DJs rather than algorithms to program sets.

“Trent Reznor called me up. He says ‘I got it. Let’s build the first ever worldwide live radio station, broadcast from three cities that plays music not based on research or genre or drumbeats,” Iovine said, recalling the genesis of Beats One. “Only music that is great and feels great. A station that has only one master — music itself.”

The emphasis on human curation puts Apple at least superficially at odds with the current trend among online music services toward bulking up their data collection and analytics assets and relying more on machine learning to refine the user experience. Last month, for instance, Pandora acquired music analytics firm Next Big Sound to help make sense of its the Internet radio service’s massive data trove. That move came on the heels of Spotify’s acquisition last year of M.I.T. spin-out Echo Nest to help it analyze music consumption habits. Spotify also recently launched its first programmed vertical, Spotify Running, which leverages advanced analytics to generate personalized, contextually relevant playlists.

Whether the difference is more than superficial, however, is unclear at this point. It’s hard to imagine that Apple will not carefully analyze what works and doesn’t work on Beats One radio or how long its programmed playlists keep users engaged with the streaming service. It’s also hard to imagine those findings will not, at some point, become part of the curation process.

Also unclear at this point is how much data on Apple Music users and their proclivities Apple will share with artists and record companies. Included in Monday’s announcement is a new social media component to Apple Music called Music Connect. According to Cue, Connect will let artists create their own social media pages, where they can upload their music and other content and engage directly with their fans.

But Apple has not said how much data on those fans artists will get back at this point. Both Pandora and Spotify have recently bolstered the data and analytics tools they make available to artists for touring and marketing purposes. Given how much data Apple already has on its users, from app purchases, iTunes downloads, phone location detection and other sources, it could potentially provide artists and labels with a much richer set of data tools than any of its competitors, with the possible exception of Google.

At some point, that capability could become more critical to differentiating Apple Music from competing services than human curators.