Here’s something you already know: There is a lot of stuff on the Internet. Maybe too much stuff. Users upload more than 400 hours of video to YouTube every minute. Instagrammers post more than 80 million photos daily. During one weekend of Coachella, people tweeted 3.8 million times. (Presumably they also listened to music.) Netflix accounts for more than a third of US Internet traffic, and Facebook basically is the Internet for many people. Such enormous numbers communicate one thing: You could never read and see everything online. Not in 100 lifetimes. Not even if you never ate or peed or worked again.

The other obvious truth is most of the stuff on the Internet is bad. Or, to be a little more glass-half-full about it, the limitless selection means there’s certainly something out there you’d really like to see, if you could find it. But how do you find it?

This is the question that vexes Twitter and Facebook and YouTube and Netflix and Apple and every other content-delivery company you could name. They have millions of users, all of whom want to see the best stuff and all of whom have a different definition of “best stuff.” It wasn’t enough for these platforms to put stuff on the Internet (or, rather, get you to do it for them) and then make it easy to find. No, if they want to keep everyone engaged in an era of hilariously short attention spans and nonexistent platform loyalty, they must become recommenders. They have to tell you what you want.

Today, YouTube rolls out a redesign to its iOS and Android app that totally overhauls the homepage. Rather than a bunch of videos grouped by channel or topic, you’ll see a single feed of recommended videos. They fit one or two at a time on the screen, with big images and only a little information. Aesthetically, it’s cleaner and more visual in a way that suits YouTube. Far more interesting, and important, is where those recommended videos are coming from.

Trust Me, You’ll Love It

YouTube has spent years refining how it makes recommendations, and the algorithms are now so accurate that the company feels comfortable suggesting just a video or two. “Even when we show fewer videos on the page,” says Johana Wright, VP of product management, “we can make them even more relevant to you.” In the past, YouTube always hedged: You seem to like music videos, here’s some popular music videos! Want to watch another Veritasium video like the one you saw yesterday? Recently, though, the team reached a point where it felt confident recommending an individual video, not just a topic or channel. Couple that with a desire for an image-heavy UI, and you have the redesigned app. Wright and her team don’t care if you open the app; they care if you watch the video. Better recommendations mean more people click play.

The immensity of YouTube’s platform is a huge asset and a huge problem. “For everyone in the world,” Wright says, “there’s a potential feed that represents them and what they want to watch.” YouTube’s new neural network-powered recommendation tools consider a vast library of data and content: Where are you? What have you watched before? What have you liked or disliked? What device are you on? What time is it? YouTube even has teams working on understanding the content of a video, trying to figure out what it was about a specific video you liked. The new system is designed to show you more new videos, from creators you already like or might like, and keep you in the flow of watching videos.

Before Wright joined YouTube, she worked on Google’s search and Now teams. “All of these things are about how you pull data together,” she says. “How do you take collective learnings from a lot of people, and turn it into products that people love?” All over Google, and elsewhere, others are asking these same questions. “Having all of this amazing, curated content doesn’t really matter at all unless it’s delivered accurately to the right person at the right moment,” Peter Asbill, head of streaming for Google Play Music, told me before the company launched its podcast integration. He’s right. A platform can offer everything anyone could ever possibly want, “but if you’re a country fan and I’m delivering you a metal experience, then who cares? It doesn’t matter how good a metal playlist it is, it’s the wrong thing.”

You Don’t Know What You’re Missing

All across the Internet, everyone’s trying to surface the good stuff. Netflix and Hulu are deeply invested in improving their recommendations, and Hulu in particular is embracing the fewer-but-better strategy. “My goal for you as the viewer of Hulu,” says Ben Smith, the company’s head of experience, “is to show you the most interesting and compelling thing to watch, right away.” You and everyone else, Ben.

When Twitter announced a tweak to reveal the best tweets at the top of your timeline, product manager Michelle Haq said user research revealed “people thought it was really hard to catch up on the best of Twitter.” These people got serious FOMO, which deepened as they got more into the service. “For a heavier user,” Haq said, “who might follow a thousand accounts, who knows what they’re missing?” When Instagram announced a similar feature, it said people don’t see 70 percent of the images in their feed. In both cases, these are people you explicitly want to see stuff from! Facebook long ago embraced using an algorithm to generate the news feed for exactly this reason. Now everyone else with the problem of too much content hopes to solve it the same way.

It’s almost cliche to say these companies aren’t competing with each other; they’re competing with your attention. Louis CK recently went on a rant about how there’s so much porn online that they could stop making it and you’d still never see it all. The same is obviously true on YouTube, and the pace of creation is only accelerating. (The numbers are crazy!) There’s so much good stuff to watch, read, hear, see, and play with, which means people have dwindling patience for crap. The platform people trust to know them better than they know themselves and to know what they want will be the one they keep returning to. The goal for every platform is the same: Make it so that all you ever have to do is press play.

Jump to original: 

We’re Drowning in Content. Recommendations Are What We Need