Google X head talks of failing fast, Loon and autonomous cars

Google's (NASDAQ: GOOG) Captain of Moonshots Astro Teller gave I/O conference attendees a glimpse of how the company's Google X division has been able to make so much progress in the "crazy-sounding science project" known as Project Loon: by failing fast.

Teller (Source: YouTube)

The goal of Project Loon is to beam Internet service from balloons floating in the stratosphere down to the 4 billion people who don't currently have a good connection. The project calls for "flying cell towers" that talk to each other in the stratosphere and ultimately provide uninterrupted Internet service. The company has deals with Vodafone, Telefonica and Telstra and is talking to others.

The thing is, few if any people have knowledge or experience with objects in the stratosphere, let alone maneuvering them. "We knew we had a lot to learn," Teller told attendees of a session titled "Helping moonshots survive contact with the real world." But, "we misestimated how much we had to learn."

In the stratosphere, the winds are going in different speeds and directions quite close to each other, and that's good because the Loon team wants the balloons to go up and down. "We have taught the balloons to sail on those winds," but what they didn't anticipate is how much the balloons would be whipping around.

Google's models show it needs the helium-filled balloons to last at least 100 days in the air to make it a viable business. But when they found slow leaks were occurring and they didn't know why, it was time to create the "leak squad." Its goal was to create, detect and fix the leaks.

Toward that end, the engineers studied a lot of other things in the world where it's important that objects not leak, including Doritos bags and condoms. "Imagine something the size of a house, rubbing soap over the edge and filling it so full of helium that somewhere the leak would begin and then looking around it for little bubbles," he said. Or having a wand act as a helium detector to "smell" the helium. If they found the leak, they could redesign how the balloon is constructed.

"Most of the stuff did not work, but we did find several really important things." In the early manufacturing process, the balloons are so big, "we have to stand on them in order to make them," he said. Someone brought up the question: What if walking on the balloons was causing some of the leaks? "We did a test," whereby a control group wore their normal socks and another group wore fluffy socks when they walked on the balloons. Turns out, the fluffy socks made a difference--they actually created a line dance to determine how the balloons with fluffy socks had fewer leaks.

Project Loon's balloons now stay up for more than six months at a time, and engineers can routinely sail them around the world to within 500 yards of where they want them to go.

It hasn't always been that way. During another I/O session, Google engineers on a panel about engineering for the stratosphere talked about how in the early days, the balloons were built in house by a group of scrappy engineers. The balloon prototypes usually wouldn't make it too far from California, but they tracked one over Kentucky, where they started to see YouTube alerts about UFO sightings. The local news showed an image of one of the balloons as part of the "UFO" coverage.  

According to Teller, everybody needs both moonshots and failures, and getting certain things out into the real world as fast as possible helps both. Probably more than anything, self-driving cars have to get into the real world in order to work. One of the driving forces behind Google X is its projects need to be huge--like saving lives when 1.2 million people die every year in car accidents--and involve some kind of "Aha" or breakthrough science or engineering. They also need to create a "ton" of value for the world and result in a fair and equitable return for Google.

Of course, getting the autonomous vehicles to work properly is the utmost importance if they want to save lives. "Pretty good, most of the time, is not good enough when you get into a self-driving car," Teller said.

As part of its testing process, Google gave Lexus vehicles to some non-X Googlers to use on their commutes and instructed them to pay attention while in the car. The cars performed flawlessly. The people did not. "People don't even pay attention to driving when they're driving," he said. "So imagine what they do when they think the car's mostly got it covered and once in a blue moon, I'm going to need to take over. It was not pretty… We stopped doing it. Our success was a failure when you factored in human nature."

They determined that if there was going to be a self-driving car, it would need to go from Point A to Point B by itself with no help from a person. So the team ended up building a smaller car, which goes no more than 25 mph, and they will be tested on city streets this summer. They will have a temporary steering wheel, foam front ends and flexible windshields. The plan calls for removing the steering wheel at some point to make it completely self-driving. 

For a lot of Google X projects, nobody immediately knows the right way to build the solutions. "We force ourselves to seek out this contact and sometimes this turns out to be us dragging our balloons up to South Dakota to expose the balloons to arctic winds," he said. "Sometimes it's asking a really specific, tiny question like how long will this glucose sensor the size of a piece of glitter actually be able to sense glucose while sitting in this tear fluid."

The question is how and how fast you can discover that what you're working on is the wrong thing to be working on, he said, which can be discouraging to hear. "We all avoid going out into the world, throwing ourselves at the world to discover these things. But no matter how discouraging it is now, if you put more time into doing it, you will unconsciously avoid even more doing it tomorrow or a week from now or a month from now, and that's why doing it as fast as you can is actually the easiest way and the most efficient time to discover that you're on the wrong path." That's why it's central to how Google X works on solving these problems.

The postal system and other delivery methods, as well as planes and trains, each significantly changed how things were transported when they were introduced. Google X still believes a significant amount of friction exists in how physical things get moved around, and self-flying vehicles could eliminate some of the friction.

Google's Captain of Moonshots Astro Teller addresses the I/O crowd. (Source: YouTube)

Project Wing had some very bumpy months in late 2013 and early 2014, he said. The thing that sounded exciting at the beginning was the idea of delivering defibrillators to people. The thought was that if somebody is having a heart attack, it would be useful to have the mechanism available when and where they need it. "What if defibrillators could come find you--at least get to the front of the building?" maybe even all the way to the person who needed it. It's small and time matters, so the vision was to shave minutes off the process and save lives.

Google's engineers started building a version of a delivery vehicle and a real-world team went out to see if there was something wrong with the plan. The team talked with emergency responders and emergency 911 dispatchers. "We quickly found out the whole plan wasn't going to work," he said. The defibrillators are actually quite hard to use for a lot of people and 911 is just not set up to respond to where the call came from--"you would be surprised how much they don't even understand where the call comes from, it's a somewhat antiquated system. It just wasn't a good plan."

The team took a big step back and went to Australia, where engineers discovered the types of things people wanted to have delivered that they had not even imagined. Cattle ranchers, for example, don't have the storage facilities to keep vaccines on hand but if they could get them delivered right there out in the field when they need it, that would be useful. "We would never have figured that out sitting in a conference room with a white board, I promise you," Teller said.

Teller also noted that Google Glass has graduated out of Project X and is now part of Tony Fadell's part of Google; Fadell joined Google with the acquisition of Nest Labs last year. "We will be hearing more in the future about that," Teller said during a Q&A.

Overall with Google Glass, the company did something right and something wrong. "The thing that we did right was actually getting it out into the real world," he said. The Explorer program was "exactly the right thing to do." Meanwhile, "we were trying to learn also about the social issues and how people would think about Glass, but we ended up sending signals that it wasn't a prototype, that it was a finished project."

Things like putting it on a runway left people with the feeling that it was a finished product, when Google wasn't productizing it yet. "We were just trying to learn, and I think we left people with some confusing messages there, and that I wish we had done differently."

For more:
- see this Computerworld article
- see this Silicon Valley Business Journal article

Related Articles:
Google's Loon refines processes, gets close to rolling out 'thousands' of balloons
Google hopes to have Project Loon customers by 2016
Google executive tells mobile industry Loon is not a threat as team reveals lessons learned