SXSW: GoogleX Exec Explains Why Early Glass Project Failed
The head of Google's so-called Moonshot Factory revealed why GoogleX has to fail to succeed during a keynote presentation at SXSW.
Google made headlines in January when it announced that its much-touted Google Glass Explorer program was ending.
Engineers are already hard at work on a second version of the connected headsets, which were met with mixed reviews from techies, fashionistas and influencers alike. But the failure of the first provided a teachable moment for the Mountain View, Calif., tech giant, said GoogleX's Astro Teller during a keynote speech at South By Southwest.
"The thing that we did not do well that was closer to a failure is that we allowed and sometimes even encouraged too much attention for the Explorer program," he told a packed house. "What we wanted was to say to the world, 'This is an early prototype of something that we think is really exciting.' But we also did things that encouraged people to think of this as a finished product. ... We could have done a better job of communicating that."
The Glass story fit into the theme of Teller's talk: It's better to fail early than fail late. "I'm going to tell you some stories today about things that we learned, how we learned them and how that's helping us evolve at GoogleX," he said at the start of the hourlong presentation on Tuesday afternoon in Austin.
In the case of Glass, he explained, GoogleX needed to get the gadget in the hands of technophiles and other influencers who could provide important feedback in the early phases of the product, but that generated more attention to the device than perhaps it was ready to receive. "The feedback we got from it was absolutely critical for the future of Google Glass and wearables in general," he said, noting that the company would do things differently to market early products in the future.
He also shared stories on the self-driving car that GoogleX is currently developing. "If you could make a car that really was safer than a person at driving in all of the situations that a person drives, there are more than one million lives a year that could be saved," he explained. But the challenge with creating a completely safe self-driving car, he said, is factoring in the 10,000 different things that could happen.
In fact, Teller said one of GoogleX's early failures with the vehicle was not considering human error. He explained that GoogleX sent home prototypes with of the vehicle last year for people to test on highways, but he found that his human testers weren't paying much attention when they were no longer behind the wheel.
"People do really stupid stuff when they're driving," he said to laughs from the audience. "They text when they're supposed to actually be driving. The assumption that humans can be a reliable backup for the system is a total fallacy," he said to laughs from the audience."
To remedy the situation, GoogleX removed steering wheels and break pedals from the cars to make them self-driving all of the time. Now, people are testing the cars "thousands of miles a day" waiting, hoping for the next failure to learn from.
He left the audience with the message that they should set themselves up for "creative, productive failure" so that they learn from their mistakes before it gets too expensive.