The history of Google’s moonshot project to provide global Internet access using high-altitude balloons goes back nearly 1800 years, according to Google[x] chief technical architect Rich Devaul. That’s when, according to legend, a Chinese general sent a floating lantern into the sky to call for more troops.
Devaul first began to consider how to blanket the world with balloon-based wireless Internet closer to a decade ago, when he was at the MIT Media Lab. "I got as far as a napkin sketch," Devaul says. "No balloons were inflated at that time."
The fundamental insight that would fuel the now (literally) full-blown project at Google[x] was of the "fast, cheap and out of control" variety. "Many people had proposed using either tethered balloons or stratospheric airships," Devaul says. "I decided that maybe that’s not the right problem to solve." Instead, he wondered, what if the problem was figuring out how to keep the globe covered with a network of constantly moving untethered balloons?
The fruit of that idea is "Project Loon," which launched its first public trial last month in New Zealand, using 30 balloons to try to bring wireless signals to 50 testers including sheep-farming entrepreneur Charles Nimmo.
"Many of our flights did really exactly what we thought they would do," says Devaul. But some, notably, did not, and their unexpected travels reveal the challenges ahead.
Take Ibis-74. Instead of staying in a narrow band of latitude, this balloon drifted into an enormous storm, until 12 days later, while low on power, it was terminated somewhere over an ice shelf. "It was in Antarctica, technically speaking," says Devaul.
The flight of Ibis-74 points to some issues minor and fixable—solar panel systems that couldn’t deal with the extreme cold and extreme angle of the sun—but also some more fundamental, like an inability to perfectly predict the weather.
"The accuracy of the predictions is about as good as any weather predictions," says Devaul. "As you get further and further out, [the predictions] get worse." That’s not just a problem when there are storms. The balloons rely on predicting wind direction for steering, since they can only move up and down.
As the number of balloons increase, this problem will multiply, but perhaps also become more soluble, as the fundamental idea of balloons replacing other balloons will be put into practice.
"We’ve shown we can launch a few tens of balloons for a short period of time," says Devaul. "Now we need to go and launch potentially hundreds."