Google’s “moonshot” to deliver Internet to remote parts of the world using high-flying balloons has survived a brutal development phase, and will enter testing with carriers in Indonesia and elsewhere this year.
But Project Loon almost didn’t make it.
Google struggled to find a balloon design that could be inexpensive and durable enough to not only float but navigate to predictably travel through the stratosphere.
“We busted a lot of balloons,” said Astro Teller, head of Alphabet’s X unit (formerly Google X), showing off some of the designs to the crowd at the annual TED conference, which kicked off Monday in Vancouver.
There were shiny balloons and round balloons and balloons that looked like giant pillows. But eventually the company found a design that could be made cheaply and still navigate precisely. That balloon, Teller said, last year travelled around the world 19 times over 187 days.
“So we are going to keep going,” Teller said, noting that what was once a slow connection has advanced enough to deliver about 15 megabits-per-second Internet access, which he pointed out is enough to deliver video — such as a live broadcast of his TED talk.
The next step will be seeing how it works delivering real Internet service to consumers. In addition to Indonesia, Alphabet has reached a deal with the Sri Lankan government to exchange access to needed radio frequency spectrum for a stake in the project.
Alphabet is in talks with carriers around the world, Teller said, adding that the prospect is very real and that a further five billion people will have Internet access within five to 10 years.
“It will change the world in ways we cannot possibly imagine,” he said.
On the TED stage, Teller also talked about two moonshots that Google abandoned. The first, he said, was vertical farming, which would have used one-tenth the water and one one-hundreth of the land demanded by traditional agriculture. But although Google grew some lettuce, it never managed to grow staple crops like grain or rice.
Another effort would have allowed landlocked countries to ship goods far more cheaply using a rocket-like air cargo ship that could land without a runway. The idea itself might have worked, Teller said, but just building the first unit would have cost $200 million. Even for a company with Google’s riches, that proved too much to gamble.
“If there is an Achilles’ heel in one of our projects, we want to know it right now,” Teller said.
In some cases, Teller said, Google shifts thinking in ways that make a project better. Such was the case with the self-driving car project. Initially, Google’s plan was to have the car drive in all but the most challenging circumstances, at which time a driver would need to take over. But because they weren’t driving most of the time, Google quickly found that its drivers weren’t prepared to take the wheel.
So the company went back to the drawing board and came up with the fully self-driving models it currently has driving through the streets of Mountain View, Calif., and Austin, Texas. At last year’s TED conference, self-driving car chief Chris Urmson said he wanted the cars available by 2020 — before his 12-year-old son is able to get a driver’s license.
Starting with testing out the hardest part of a project, Teller said, is part of how X decides which projects to kill and which to continue with. The company starts with the hardest pieces of a potential “moonshot,” and only funds those that can survive the toughest tests.
The unit, he said, depends on both wild dreaming and what he called “enthusiastic skepticism.”
Teller is among the first speakers at this year’s TED conference. Virtual-reality and augmented-reality projects will take center stage, with at least half a dozen talks and exhibits centered around the emerging technologies.
Stay tuned all week for dispatches from Re/code’s Ina Fried, who is covering the speeches, demos and people at the event.
This article originally appeared on Recode.net.