In public, companies try to trot out any hint of a success story—real or imagined. At SXSW this week, Google took a different tack: It talked all about its failures.
Google has plenty to crow about: It has controlled the web search and search advertising business for years, it racked up $66 billion in revenue last year, and it gets an average of 5,000 job applications a day. It has upset countless industries, from mobile phones to TV.
But before a packed auditorium, Astro Teller, the Google scientist known as the “Captain of Moonshots,” talked about some of the biggest flops from Google X, the part of the company that chases highly experimental and futuristic products and services.
The team behind Project Loon, Google X’s internet ballon project, misjudged how difficult it would be to make the actual balloons themselves. The massive balloons continued to spring little leaks at high altitudes. The team proposed countless hypotheses as to why the balloons kept springing leaks, designed potential solutions and then ran test after test. The Loon team even ran a study on the fluffiness of the technicians’ socks who built the balloons. The lesson: “Sometimes the most interesting failures are the ones that you don’t expect. Particularly, when they are something that you think will be the easiest part of the project and it turns out to be the hardest part of the project instead,” said Teller. The balloons are now able stay aloft for about six months.
Google Glass, was predicted by even outside analysts to be a category defining product that would sell millions of units in just a few years. Just look at the boundless optimism of the projections below.
Turns out the offputting eyewear that inspired the term “glassholes” became known as one of Google’s biggest turkeys. “We made one great decision and one not so great decision about Google Glass. The great decision was that we did the glass explorer program. That was absolutely the right thing to do. The more bumps and scrapes you get, the better it was for you to get out into the world,” Teller stressed. The failure was not making clear to everyone else that what was out was really just a protype of the smart glassware, and too much bad publicity was really what killed Google Glass, he said.
With Project Wing, Teller and team wanted to improve the delivery of physical products by creating an airborne delivery system. The team decided to create a vehicle known as a “tail-sitter.” They quickly realized that the tail-sitter had some major failings, particularly with control. Sergei Brin gave the team a five-month deadline to make deliveries–not enough time to try a new vehicle, so they used the tail-sitter and failed. Brin’s deadline forced them to fail faster and now the team is working on a new delivery method, which Teller says is much more promising. Lesson: Deadlines that seem impossible to meet can actually be constructive.
With Project McConney, Google X wanted to create an energy kite. Basically, an airborne wind turbine that harvests the power of the wind at a lower cost than tradition wind turbines. Instead of starting with a full sized model of the turbine, the team built smaller scale versions of the turbines, of which Larry Page stressed that they needed to crash “at least five” to speed up the learning curve.
So the team picked one of the windiest places in North American, Pigeon Point in California, to test the turbines. But the turbines still didn’t fail. They didn’t crash a single one. “It says something about Google X that .. we didn’t really want to see if crash, but we all feel like we failed somehow,” Teller said with a laugh. “There is magic in everyone believing that we might have failed because we didn’t fail.”
In 2012, Google X’s self driving cars had already reached such a level of success that the company let non-Google X people drive the cars. It was during these tests that the team discovered an unforeseen problem: The team’s assumption that humans were a reliable backup for the technology was wrong, because once they trusted the technology they themselves couldn’t be trusted. “Our success was itself a failure,” he said. We realized very quickly that what we had to do was make it hyper-clear to ourselves that that was not going to work, and the only way to do that was to make a car that has no steering wheel, that has no brake pedal, that has no acceleration and that drives itself all the time from point A to point B at the push of a button.”