Skip to main content

What do you want to listen to? Algorithms still can’t tell like humans can

want listen algorithms still cant tell like humans can dj
Last week, the Associated Press excitedly unveiled its plan to essentially hand over its earnings reports to a swarm of robots later this month. The news, naturally, had human-type bloggers in full “they took our jobs” mode, as though this sort of announcement signaled the beginning of the end. As someone who’s written up his share of earnings reports, I feel pretty confident in saying that anyone liberated from writing them is breathing a sigh of relief. Like most ‘bots, those employed by AP will be performing a fairly thankless task — one that often feels like the journalistic equivalent of Mad Libs.

The AP claims the move will free up resources for reporters to spend more time digging into meatier pieces, and there’s certainly reason to believe this. Newsrooms — even those as padded as the Associated Press’s — are crunched, just like everyone else. The less money spent slogging through number pieces, the more that’s freed up to do more substantial reporting (of course, we’ll have to keep watching to see if the AP keeps its promise here). This does, however raise the perennial question of machines and creativity: Will humans always be necessary for higher level tasks?

Will humans always be necessary for higher level tasks?

The answer, historically at least, has always been “yes.” And despite the AP’s move, two other events last week only reinforce the idea that maybe us human beings are worth keeping around for a little while longer, after all.

The first happened last Monday. Rdio, a music-streaming service akin to Spotify, created by the cofounders of Skype, announced that it had purchased TastemakerX, a smallish San Francisco-based startup focused on music curation and discovery. It wasn’t huge news in and of itself, of course. Both parties are relatively small fish in much larger ponds, and besides, music discovery and curation has always been a major part of Rdio’s strategy, with the company placing a tremendous emphasis on social networking functionality.

What happened a mere two days later, on the other other hand, rightfully turned a lot of heads. Google announced that it would be doubling down on the music game by buying Songza. You’ve heard of it, right? Songza is yet another music-streaming and recommendation service.

What sets the service apart from a number of competitors, however, is Songza’s reliance on human curators. This stands in contrast to, say, the likes of Spotify’s artist radio feature, which relies mostly on algorithms to pick music you might like. Even this cold mechanical apparatus picks up the warmth of human touch in the forms of a thumbs up or thumbs down from the end user, which helps further fine tune your listening. At the heart of it, however, Spotify’s got machines doing the work of a million music journalists on a million typewriters.


Songza, on the other hand, uses human music experts to create playlists tailored to listeners’ moods. The approach has proven successful for the startup, scoring more than 1 million downloads within the first 10 days of its iPad app launch back in 2012. Spotify certainly noticed. In May of last year, the company went out and purchased playlist competitor Tunigo. That acquisition resulted in the launch of Spotify’s Browse feature, which now greats users with a slew of mood-based playlists every time they fire up the app.

Even Pandora, which famously pioneered music curation by algorithm, has a beating human heart in the Music Genome Project. Songs aren’t just fed into a program, they’re analyzed by human music scholars to catalog different elements that machines just can’t recognize, like “hard rock roots, mystical qualities, mild rhythmic syncopation, repetitive melodic phrasing and demanding instrumental part writing.”

Apple seems to recognize the value of a good ear, too. When Cupertino announced its purchase of Beats Electronics in May, many folks (present company included) suggested that the real, erm, apple of Apple’s eye wasn’t headphones so much as the recently launched music-streaming service, Beats. It makes sense, really. Apple’s clearly been looking to expand its digital-music empire, and streaming appears to be the next logical step.

The real secret sauce fueling the whole thing is a team of humans — songwriters, critics, radio DJs and the like.

So, what makes Beats Music so special? Well, even more so than Spotify or Rdio, the service was launched with playlists in mind. The service’s killer feature is the ability to generate playlists on the fly using data input by users. Sure, there’s some computer processing that needs to occur in order for those emotions, activities and times of day to become a playlist, but the real secret sauce fueling the whole thing is a team of humans — songwriters, critics, radio DJs and the like, who were hired by the company to provide the service with expert curation.

There’s a quote famously misattributed to Elvis Costello that compares the business of writing about music to “dancing about architecture.” But as futile an exercise as writing about music might feel at times, it certainly comes closer to capturing the spirit of music than feeding it into a machine does. And so does human music curation.

When Netflix was banging its head against the wall back in 2008, attempting to master the art of mastering recommendations, it referred to the issue as its “Napoleon Dynamite problem” (yet another inadvertent Elvis Costello nod, it would seem). That problem was so named because its algorithms had trouble predicting whether or not Netflix users would enjoy the quirky 2004 cult film based on past viewing habits.

It’s perhaps only a small consolation to armies of music critics who have found themselves washing dishes and partaking in other “real jobs” as music mags have suffered the same fate as the rest of the publishing industry. While Rolling Stone may not go on a hiring spree any time soon, the most recent moves in the tech space prove that as far as music is concerned, there’s still no way to replace the human touch.

Editors' Recommendations