Skip to main content

The smarter news apps are, the dumber they get

The smarter news apps are, the dumber they get


A curiously hot category with little to show for it

Share this story

In August 2011, CNN bought the newsreading app Zite for $20 million. At the time, the cable-news giant said that Zite’s algorithms for personalizing news would transform its digital presence. "The Zite technology can also help CNN’s websites and apps serve more personalized content, making our current digital services even better," said KC Estenson, general manager of CNN Digital, at the time.

Yet yesterday, CNN unloaded Zite onto Flipboard in an all-stock deal that gives it a small equity stake in the company. Allowed to operate mostly independently for two and a half years, the app’s growth had slowed, ad revenue was negligible, and that vaunted Zite technology never materialized on CNN’s digital properties. Naturally, everyone involved hailed the deal as a rousing success. "We didn’t wither and die, we actually thrived," Zite’s CEO, Mark Johnson, told The Verge of Zite's time at CNN. "It didn’t work out exactly as planned, but it worked out for everybody involved."

An app that keeps failing upward

But it’s time to take a more skeptical look at newsreading apps, which continue to attract millions of dollars in investment from venture capitalists who believe they will prove to be enormous businesses. Consider the recent history: Yahoo launched and killed Livestand; AOL launched and killed Editions; Google launched Currents and then folded it into its Newsstand app. Pulse, a popular early iPad app, sold itself off to LinkedIn. The fate of Facebook’s take on the genre, called Paper, remains to be seen — but it has been languishing in the App Store.

The premise of apps like these is that they will find interesting articles for you that you never would have seen otherwise. From their perspective, the more obscure your interests, the better. Johnson filled his own Zite with stories about volcanoes, linguistics, and astronomy. On Prismatic, a similar app that has raised $15 million, I’ve followed San Francisco, Spotify, and messaging apps, among other subjects.

And yet for all the millions spent and the machine-learning algorithms that have been built, none of them have improved on the big portals — CNN, The New York Times — or social networks like Twitter or Facebook, which bring you both the news and the conversation happening around it. Newsreading-app developers hire PhDs to do the easiest thing imaginable — find you something interesting on the internet that you haven’t already seen — and then beat you over the head with it, bludgeoning you with an endless barrage of links that never feel half as personalized as they’re made out to be.

Hiring PhDs to do the easiest thing in the world

The truth is that the news remains stubbornly impersonal. The crisis in Ukraine or the scene at the Oscars may not affect your life directly, but you’re likely have some interest in it anyway. That’s why big portals and social networks are so effective at delivering the news: they cater to broad audiences, and to the stories that move us to share them. Most of us don’t have enough niche interests to spur us to open a specialized niche-news app every day, no matter the quality of its machine learning. Apps like these are only ever going to get so far helping people find hidden news about volcanoes.

Notably, Flipboard has succeeded while mostly avoiding the content-discovery game. The app grew to 100 million users by offering a sleek, elegant user interface for browsing the news and social networks simultaneously. It brought a relaxed, magazine-like reading experience to the busy, often ugly world of online news consumption and has thrived as a result. Adding an expensive content-discovery layer to Flipboard is unlikely to hurt the company much — but it seems unlikely to help much, either.

For its part, CNN says its vision of a personalized news site will someday be realized. For one thing, the company kept a copy of the Zite code as part of the sale, enabling it to deploy it through CNN’s products however it likes. "We’re still in that game," Estenson told me. "I personally believe that technology can help us out a lot."

"We're still in that game."

And what does Flipboard get, exactly? The deal includes new CNN content inside Flipboard in the form of new "social magazines," and a revenue-sharing deal for advertising on those magazines. Beyond that, it’s hard to say; Flipboard did not respond to an interview request. In a blog post, CEO Mike McCue echoed Estenson from when Zite sold the first time around: "Adding Zite’s expertise in personalization and recommendations to Flipboard’s product experience and powerful curator community will create an unparalleled personal magazine for our millions of readers." That would be more than it ever did for CNN.

Zite had a small but vocal fan base, and its engineers’ talents are apparent. And Flipboard, which has a world-class team of product builders, may well prove more adept at integrating a team of software engineers than a cable news network. But twice now Zite has sold a dream of personalization that keeps failing to materialize. Algorithms have proven brilliant at many things, but newsgathering isn't one of them.