Looking back, it seems Pandora may have been the tipping point, the proof of concept that swayed people and subliminally convinced them that curation could be done by robots and algorithms just as well as it could be by humans. The service did surprisingly well at picking tracks similar to what you suggested it start with, without being too predictable. Credited with taking more and more jobs, it seemed inevitable that machines would also pick our music, TV, movies, and reading content on their own. A clever algorithm coded to be fluid and ever evolving software should be the best recommendation solution, right? The current trends, however, would suggest a constant human element present is needed in making media discoverable across all types of consumer content.
The biggest music store on the planet, iTunes, uses both automated means and human power to suggest new content to users. In the form of “Genius,” the store provides suggestions based on your past listening habits in addition to other factors. While the results are fine, usually recommending things people do, or would like, it doesn’t provide the same wow factor as a human emotionally involved can. The same feeling you get from a person saying “I’ve enjoyed this music, I think you will, too.” Possibly the reason Genius results are halfway down the page and often overlooked. The highlighted and featured albums in the genre sections–which are handpicked by Apple employees–do often give that wow factor, even breaking notable artist The Boxer Rebellion based on the iTunes feature.
Also gaining more attention as a music discovery destination, NPR handpicks its features and selections shown and heard across a variety of blogs and radio shows. Amy Schriefer, who’s in charge of “First Listen” and live events, was quoted by the Wall Street Journal saying “We want to create the feeling you used to get in record stores when somebody would hand you something and say, ‘You have to hear this.'” Similarly, the streaming music service Rdio heavily uses social to introduce users to new music through the same “record store feeling.” When you glance over to the side bar on Rdio.com and see one of your friends listening to something you haven’t heard of before, it appeals to your curiosity more than if you knew it was a computer-generated recommendation. In this way, music is following written content, which has come to thrive more and more on social sharing and distribution and less on robot aggregators and portals.
Communities built around high-quality writing content like Medium heavily rely on the human factor to sift through mounds of articles and essays. The site tackles the problem of suggesting what visitors should read in several different ways. Readers don’t just “Like” articles–they recommend them. Those recommendations get tallied in the writer’s personal analytics page, which also tracks whether a piece of writing was fully read or just skimmed. The process in this case may be automated, but the emphasis is placed on people actually spending time engaging with the whole piece of writing. While clever software helps make the process easier, ultimately Medium has decided high-quality curated writing, commissioned even, was important enough to hire a director of content and is hiring more in-house editors.
Curating the news seems like a job for a computer rather than a human as an effort to cut costs in a market that has been devoured by new media. In the case of Circa, a fairly new iOS app, the company actually uses paid editors to pick out news, summarize it, and deliver it through the app. The company thinks this way of human-selected and summarized content makes sense for a mobile generation who read in small chunks of time in-between other tasks. Rather than having too many stories to choose between before actually getting to reading. Considering the growth the app has seen lately and the recent hire of Reuters social media editor Anthony De Rosa to head Circa as its editor-in-chief, it would appear human involvement is key to ventures with a heavy curation aspect.
Bucking the trend, Netflix has famously relied on using its computers to spit out suggestions for those holding the remote, desperately looking for the next thing to watch. Not many other companies, after all, are in a position to offer up a million dollars in prize money for writing the code that the service uses to provide movie suggestions. It’s debatable whether a company and operation at this scale, though, might not have any other option other than to employ computers to do the recommending over humans. Fanhattan thinks otherwise and is trying to sort through Netflix–as well as other video content services–and provide curated choices, getting rid of the “what to watch” problem.
The answer, then, is a careful hybrid. Even a few years ago, it might have seemed like we were headed towards a world where the most sophisticated recommendation engines would be picking the majority of what people watched, read, and listened to. Today, we aren’t without the clever and sophisticated software–we just won’t rely on it the way we thought we would. It seems the world still needs editors after all.
Tyler Hayes contributes to Hypebot.com and does interviews for NoiseTrade’s blog. He often writes about music and the impact tech is having on that industry, which can often be found on his personal blog, Liisten.com. Tyler also runs the site Next Big Thing, which ranks user-submitted links for an interesting hub of music-related content.
[Image: Flickr user D. Sinclair Terrasidius]