Felix Baumgartner, wearing a pressure suit that more closely resembled a space-suit than the garb of a skydiver, stood before an open door on a platform more than 23 miles above Roswell, N.M. His balloon-powered space capsule’s ascent to what many had called “the edge of space” had lasted two and a half hours. It was time to jump, and — if two years of preparations were to prove worthwhile — to survive the longest-distance free-fall and maximum vertical velocity that a man ever had attempted.
“I know the whole world is watching now,” Baumgartner said a moment before stepping into the stratosphere. His hyperbole was closer to reality than he might have known, for as Baumgartner embarked on the most dangerous leg of the October 14, 2012, event — dubbed Red Bull Stratos by its sponsor, the eponymous energy drink — the live YouTube webcast was supporting more than eight million concurrent streams. That, according to Shiva Rajaraman, YouTube’s director of project management, accounted for 8% of all Internet traffic at that moment.
The volume of viewership constituted a record in its own right, and a landmark achievement for YouTube, whose role has expanded from depository of viral cat videos to a preferred tool for viewers of longer-form video on desktop computers, tablets, mobile phones and high-definition televisions.
“We know if we can do things that are phenomenal, then we can get these masses there.”
“We know if we can do things that are phenomenal, then we can get these masses there,” Rajaraman said during his keynote address at the recent BizTech@Wharton conference. Rajaraman’s presentation, sprinkled with multimedia, ranged in subject from the worldwide phenomenon of South Korean pop star Psy’s “Gangnam Style” — an animated circle graph revealed the music video’s rapid growth from novelty to sensation — to Content ID, with which YouTube scans and fingerprints copyrighted content to give owners a chance to have violations removed or monetized.
Rajaraman also delved into TrueView video ads, which people can skip after five seconds — advertisers pay only when at least 30 seconds play — and also allow viewers to choose among several different ads, and whether they want to watch the spot before viewing a video, or have commercial breaks interspersed throughout.
“While at the cursory level it doesn’t seem like such a huge engineering feat, it’s a very powerful thing,” Rajaraman noted. “Look at the traditional model of advertising. The goal is to say, ‘Hey, you may or may not like my ad, but I’m going to find a way to show it to you a hundred times….’ There’s no quality feedback in that.”
But by making the ad skippable, he said, “all of a sudden you have viewers’ feedback. If they don’t skip the ad, you know they were engaged. You have to compete to get their attention early in the video. So we have a whole quality system that says we’re going to put the best ads out there, and effectively, the better the ad is, the cheaper it is in the auction.”
The user experience, meanwhile, has continued to evolve. According to Rajaraman, much of the audience for a creator comes from countries other than the one in which he or she resides — 80% of YouTube views, for example, occur outside the United States. The implementation of an automatic translator, though its accuracy varies, eased that crossover, Rajaraman said.
“We’re also testing things like the crowd-sourcing of captions,” Rajaraman noted. “There is no way a computer can get it right, but we know that by having that reach and having a starting point, it makes it easier for other people to come on and translate. Particularly when it comes to technology, don’t forget the human element.”
A persistent question since YouTube’s rise, Rajaraman said, has been the platform on which it is consumed — specifically, should YouTube be on TV?
“Look at the traditional model of advertising. The goal is to say, ‘Hey, you may or may not like my ad, but I’m going to find a way to show it to you a hundred times….’ There’s no quality feedback in that.”
“It turns out that a good percentage, if not the majority, of watch-time now is driven by long-form video,” he noted. “And we have few tactics to get into the living room. One is to say, ‘Let’s do deals with all the game consoles, all the connected TVs,’ and we’ve done wonderful work there…. The other side of it is that we know only so many people in the world have connected TVs, but we know everyone has a phone in their pocket. That motivated us to think about this a little bit differently. What if you could take that experience from the phone and just fling it to the TV?”
That notion started coming to life for Rajaraman and his colleagues about two and a half years ago. They noticed that mobile phones were supporting more than one billion views a day. “For us, that was an eye-opener, because in different countries, your connectivity is not that great,” he said. “And then we get to some countries like Korea. These are kind of our leading-indicator countries. We look at places like Korea because that’s typically where the world is going to be going, eventually. And in Korea, over a year ago, looking at all the YouTube traffic in that country, over 50% of it was on mobile, which was just a shocker for us.”
During his talk, he projected on the screen a peppy one-minute commercial for the $35 Google Chromecast, a two-inch stick that, when plugged into a TV’s HDMI port (and into a USB port for external power), connects to Google services including YouTube. The corresponding Chromecast app converts an Android or iOS phone into an intuitive remote control. When a user taps the integrated Chromecast button on a phone’s YouTube app while watching a video, the paired TV will power on and continue the video. The stick likewise supports services like Netflix, Pandora and HBO Go.
With its low price point, Chromecast provided a simple way for people to watch YouTube on their HDTVs; it also gave YouTube access to a more patient, sofa-sitting audience, noted Rajaraman, adding that the demand for the gadget has been significant.
“It’s a fun engineering challenge just to pull off these events…. But, [it is also] very scary. Most of the great things are going to be scary.”
“That really unlocked a whole new paradigm for us,” he noted. “The lesson here is that sometimes, technology can help you get beyond frictions that otherwise are impossible for us.” Indeed, Time named Chromecast its gadget of 2013, noting “the impulse-purchase” price. “Lots of companies have built devices to do this,” Time continued. “Chromecast is the first one that gets it right.”
A central ethos at technology companies, Rajaraman said, is to buck any expression of impossibility. “Part of that is by going really big and beyond, which is saying, ‘Yeah, it might break everything, but let’s go ahead and give it a shot,’” he noted. “Don’t be conservative.”
At the liftoff of the Red Bull Stratos balloon, YouTube reported 500,000 concurrent live streams. That, according to AllThingsD, matched the previous record for a single live Web video stream, set by Google during the 2012 Summer Olympics. As the capsule climbed beyond 35,000 feet, YouTube was supporting more than 2.2 million live streams. Throughout the world, Rajaraman recalled, people were nudging their friends to have a look. By the time Baumgartner surpassed 100,000 feet, the streaming audience had grown to five million. And as he stepped onto the platform at 128,100 feet, YouTube was supporting more than eight million concurrent live streams of his record leap.
“We need to start thinking about how the Internet works to support” such traffic, Rajaraman said. “It’s a fun engineering challenge just to pull off these events — a fantastic thing for us, and we enjoyed it a lot. But, [it is also] very scary. Most of the great things are going to be scary.”