Google's Project Loon rode stratospheric winds and steered balloons by changing altitude

Google's Project Loon began from a blunt statistic: two out of every three people on Earth had no fast, affordable Internet connection, and this was not a problem that time would solve. Terrain was one enemy, jungles, archipelagos and mountains; cost was the other, with a connection in most southern-hemisphere countries priced higher than a month's income.

The obvious approaches each failed somewhere. Ground networks were uneconomic in remote terrain, and earlier high-altitude Internet schemes tried to hold a platform over one fixed area, which required station-keeping of major cost and complexity. Google X inverted the assumption: instead of keeping balloons still, it let them sail freely on the winds. The balloons fly about twice as high as commercial airliners, powered by the sun, and steer by moving up or down to catch the wind layer that is blowing in the direction they want to travel.

That inversion created a second problem: a balloon that never stays put must be scheduled. Google described managing a fleet sailing around the globe so each balloon is over the area you want at the moment you need it, solved with complex algorithms and large amounts of computing power. The system beams Internet to the ground at speeds similar to today's 3G networks or faster, and the vision was that ordinary phones on existing carriers would connect with no new ground equipment.

In June 2013 the team started a pilot in the Canterbury area of New Zealand, launching 30 balloons in one week and connecting 50 testers on the ground, the first trial at that scale. Google said the experiment was very early days, that the pilot would teach it how to improve the technology and balloon design, and that it wanted pilots in countries at the same latitude plus partners for the next phase.

Stop trying to hold position: let balloons drift and steer by altitude, choosing the wind layer that blows toward the target

Drifting removes the expensive part: no propulsion and no station-keeping hardware are needed to stay over a fixed area, because the design never tries to stay over one.

The stratosphere is layered, so a small climb or descent picks a different horizontal wind: altitude is the steering wheel and the wind is the fuel.

Solar cells power the payload and flight systems, so a balloon can stay up across day and night without refueling infrastructure on the ground.

Beaming 3G-class connectivity to ordinary phones means the hardest part of rural access, the last-mile infrastructure, is not built at all.

When station-keeping makes a system ruinously expensive, make position a free variable and steer it with the cheapest control available, then treat routing as a scheduling problem

The announcement was a first test, not a proven network: the Canterbury pilot launched 30 balloons and 50 testers were trying to connect, and Google described the results as lessons still to be learned. The Hacker News thread on June 15, 2013 drew 189 points and 50 comments, with readers split between enthusiasm for connecting the unconnected and skepticism about privacy, airspace and physics. Google said it would seek pilots at the same latitude and partners for the next phase, and the project moved forward as one of Google X's public moonshots.