New Anonymous-Sharing App Wants to Apply the Netflix Algorithm to Content

The herd informs Heard's suggestions

With so much content being produced online, it's hard to stand out as a digital publisher to your target crowd. Heard, a new anonymous social sharing app, believes it has solved the discovery problem. The platform applies the Netflix algorithm for films and TV shows to online content, ideally predicting preferences based on previous choices—and it's already convinced Funny or Die and Huffington Post to come on board.

"You have so many different social media outlets; there's really no solution (for discovery)," Heard vp of product and strategy Benjamin Goodman explained. "Now, you can dump all your content you have on Heard, and our algorithm takes over."

Heard will be competing alongside nine other tech companies at the SXSW Interactive Release It competition on March 13 in Austin, Texas. The platform was built under the expert eyes of Heard CTO Ruben Kleiman, who was one of the architects of Netflix's recommendation algorithm, and Heard CEO Dave Vronay, who worked in social computing at Microsoft. Vronay said the company came up with the idea after realizing that people often searched for movies starring the same actor, and that model could be applied to online media.

The platform works like Netflix, in the sense that it sections platform users into like-minded groups of up to 300 users. For example, people who like reading stories about U.S. politics will see the same recommendations, while those who like My Little Pony will see another bucket of choices—but both won't see the same set of content. Those who fall into both categories will be sectioned off into another group that will see a mix of materials from the two topics. Vronay said you don't need to rate or click on that many items: By making a few selections, the model can automatically select which cohort you fall into. However, the more active you are on the platform, the more specific your groups become, and the content becomes better tailored.

Users remain anonymous because the algorithm is based on group behavior, so there is no need to track a single person. There is an option for users to get a badge on their profile that verifies their employment at a company, if they create a login with their work email. This badge shows that the user may have institutional knowledge about a certain topic. These users are still grouped with others under a topic umbrella, however, so there is no way to specifically identify their behavior.

Heard may also be able to help older content stay popular. The app continuously refreshes because so much content is produced daily. As more users upvote certain pieces or click on links, they appear more often to their target groups. And, as people find the pieces less relevant, they head to the bottom of the pile. This method allows topical older content to stay on top way past its original post date, as long as people are still interested.

"It makes use of your evergreen content and captures the longtail on it," Goodman said.

Admittedly, if you head to the desktop version of Heard right now and you aren't a registered user, the content it recommends tends to be very meme- and image-heavy. However, with new partners Funny or Die and Huffington Post, Heard is bolstering up its editorial content. Goodman said it will announce more partners soon after SXSW.

"This is an opportunity for people out there who can't build up an audience and have interesting things to say," Goodman said.