Today, across YouTube, rebuffering occurs far too often. Any given video will stop and start around 3 to 5 percent of the time. This differs vastly by country—videos in Japan rebuffer less often than video in India, where the Internet is slower—but the company’s goal is to improve these rates everywhere. “Imagine if you got a new television provider and your videos stopped 5 percent of the time, or imagine if it took you five seconds to change channels,” says Shishir Mehrotra, a vice president of product at YouTube. “You would get a new television provider, and you wouldn’t change channels very much.” Mehrorta’s goal is to reduce YouTube’s rebuffering rate to around 0.1 percent, and he wants videos to start playing within 200 milliseconds of your choosing them, or what he calls “instant time.”
To make this possible, last year YouTube implanted a new serving technology it calls “sliced bread.” Today, if you play a YouTube video on a PC browser (and soon on other devices), the company will break up the clip into lots of little pieces, and then it will send you each bit of video as it becomes necessary. The old streaming method was akin to throwing a full manuscript of Infinite Jest in the mail; the new way is like sending each page of Infinite Jest by a superfast bike messenger. The approach lets the company change how your video is being streamed as you’re playing it. It can send you lower-quality slices of video right at the start, so your clip begins to play instantly, and then improve the quality if it determines that your connection is solid. So far, the company says, sliced bread has reduced rebuffering by 40 percent, resulting in tens of millions more hours viewed.
The discovery problem is thornier than the speed issue. YouTube’s partner program has led to lots of new content on the site. Thousands of channel owners, many of whom have garnered millions of subscribers, now earn money from their YouTube videos. Some of these are quite popular: The comedy duo Smosh have 10 million subscribers, the gaming channel Machinima has 8 million, and a fashion and beauty channel run by Bethany Mota has 2 million. But aside from a few breakout successes, most YouTube channels aren’t of mainstream interest; by design, they appeal to niche audiences. To make itself comparable to traditional TV, YouTube needs to fashion these small videos into a larger, more comprehensible viewer experience for viewers. It needs to know what you like, and to serve you a new video that you’ll love each time the previous one ends.
To do so, YouTube has subtly changed its approach to figuring out viewers’ interests. In the past, YouTube leaned heavily on predictive algorithms that used your previous viewing behavior to guess what you might want to do next. “Our assumption has been that users are inherently lazy—they’re not going to take a lot of effort to tell us what they like,” says Noam Lovinsky, a director of product management at YouTube. “But I think that’s a mistake, and you’ll see us change that approach.” In particular, YouTube has been experimenting with ways to encourage more viewers to sign up for more channels. Once people do sign up for a channel, they tend to come back to YouTube more often, and when they do, they watch for longer stretches of time. (The number of people subscribing to channels has doubled in the last year, YouTube says, though it declined to give Slate the actual figures.)
The big question is how much like traditional live TV YouTube will become after it makes these improvements—and how much it will stand apart from TV, too. The dream is that when you flip on your YouTube, you’ll get popular long-format shows as well as niche-ier stuff that you can watch in line at the supermarket or in a doctor’s waiting room.
“If we do a good job of this, the ways people will engage with YouTube will go beyond the reasons they currently turn on their TV,” Mehrotra says. “The number of things that compete for users’ time has increased enormously in the last 30 years, yet every year the amount of time that people spend watching television has gone up.” In other words, as good as YouTube gets, the real tube isn’t going anywhere. Instead, YouTube is betting that our overall appetite for screen time will keep increasing. YouTube and live TV will exist side-by-side, and none of us will ever get anything done.