Google[x] Head Astro Teller Says Moonshots Are All About Embracing Failure

Google’s ‘Captain of Moonshots’ Astro Teller admitted to a packed crowd at SXSW that a number of Google[x] projects have experienced a series of “bumps and scrapes” over the last five years. But he also highlighted how the Google lab attempts to overcome those issues and “embrace failure” in projects that aren’t going so well.

The original moonshot proposal for the secret Google lab was to get people to the moon. It has since spun out 10 different projects that involve everything from biotechnology, robots and cancer detection nanotechnology.

But it has not been without a lot of problems. One example is Google Glass, which Teller said he regrets launching as if it were a finished product.

“We made one great decision and one not-so-great decision. The great decision was the Explorer program. The thing we did not do well is that we allowed and somewhat encouraged too much exposure to the program. We did things to encourage people to think that [Glass] was a finished product,” he said.

It turns out building autonomous vehicles is also pretty difficult. For Google’s self-driving car project, the lab gave a bunch of Lexus SUVs to non-Google employees. They drove fine on the highway but not on city streets. The mistake Google made was in assuming humans would pay attention the whole time and be a reliable backup. That didn’t work out so well all the time.

“People do really stupid things while driving,” he said. Afterward, Google switched to making the cars in-house, and that project is still underway.

We did things to encourage people to think that [Glass] was a finished product Astro Teller

He also shared Google’s “maniacal interest” in seeing projects through, even those which don’t seem to be working. Project Wing, a self-flying delivery vehicle, fell under that category. The team believed the idea to be a failure 18 months after the fact.

“Fifty percent of the team knew it was a bad project after a few months, 80 percent knew after a year-and-a-half but they didn’t kill it,” he said. The team is still at it on that one, too. “I look forward to telling you more about that later this year,” Teller said.

Google’s Project Loon, a plan to deliver Internet around the world using giant 20-meter balloons, had problems because the balloons leaked when they hit a certain altitude. The team resorted to studying things like the fluffiness of researcher’s socks to figure out what the problem was. “It turned out that actually mattered,” Teller told the crowd.

These moonshots are all about embracing failure, according to the head of Google[x].

“You make a ton of progress by making a ton of mistakes,” he said. “The longer you work on something, the more you don’t really want to know what the world is going to tell you. The longer you put off that learning you will unconsciously put off that news because it is disheartening to hear that what you have been working on is not working.”

Google’s self-driving car project now travels several thousand miles of city streets every day to find ways to get stuck. “It’s not about ensuring you don’t have these bumps and scrapes. It’s about ensuring we learn from them,” Teller said.