Astro Teller, head of X, is not shy about it: Lots of failure happens at the self-described "moon shot" factory, which is responsible for some of the most ambitious projects at Alphabet, Google's parent company.
Teller, a ponytailed scientist with a salt and pepper goatee, famously preaches a gospel of "failing fast," so you can find out something doesn't work quickly and move on to other projects.
But that failure comes with a balance.
"One of the rules of thumb we have is when you propose to try to build something or accomplish something, roughly about half the time you should be able to make it," Teller said Tuesday during The New York Times New Work Summit in Half Moon Bay, California. "That's the level of audacity I'm looking for you to deliver to me."
The reason? If someone on his team is extremely audacious and only accomplishes a goal one out of 10 times, it's hard to hold that person accountable. On the other hand, if someone accomplishes 90 percent of their goals, they're "not really shooting for the moon," he said.
X, formerly called Google X, houses some of Alphabet's most innovative projects, including Wi-Fi beaming balloons and delivery drones. The lab also developed self-driving cars, smart contact lenses and Google Glass, before those efforts were spun out of X and into separate projects.
Teller's comments give us a peek into the process of innovation at Alphabet and show how the company handles its ambitious moon shots. Over the past few months, the company has stepped up scrutiny of its moon shots. Some of those projects -- within X and at other divisions -- have been scaled back or altered. Those include Fiber, Alphabet's high speed internet service, and efforts reportedly in areas such as drone operations.
On Tuesday, Teller also said he grapples with when to kill projects he doesn't think are working out, as opposed to letting a manager kill it himself.
"I'm torn because I'm not spending Alphabet's money wisely in the short run if I let him keep going," he said. "But the empowerment and the training for him and for everyone at Alphabet that comes if I wait the extra couple of months it usually takes for him to come around to it, I think is really good for Alphabet over a somewhat longer period of time -- two or three years."
"It's a tough choice to make," he said.
CNET Magazine: Check out a sampling of the stories you'll find in CNET's newsstand edition.
Batteries Not Included: The CNET team shares experiences that remind us why tech stuff is cool.