[nostrademons]

A lot of Loon's early engineers were drawn from a team I was on, because Loon's founder (Mike Cassidy) was our PM's boss. When our PM left to work with him, he said "It's a crazy idea. Less than 5% chance of success." He joined it anyway. The team certainly felt it was both high risk and high reward.

[dekhn]

But... why? It's just a balloon with internet. All the failures would be a lack of business model, not the technology.

[nostrademons]

A quick list of issues that early engineers had to solve with Loon:

1. Can you get wi-fi reception on the ground from a balloon? How big a transmitter, how big a power source, how long can you keep it going for, and how much weight do you need to spend on batteries for this?

2. Will your wi-fi reception cut out if a tall building happens to pass between you and the balloon? How do you mitigate this? Do you need duplicate coverage from multiple balloons? (Starlink is dealing with this now.)

3. How do you keep the balloons in one place, or at least control where they go? If one goes out of range, how do you make sure there's another covering all its receivers?

4. How much propellant/battery can you carry on the balloon for course adjustments? How long can it stay aloft before this is exhausted?

5. Where do the winds go in the stratosphere? If the balloons are not propelled, where will they end up? Can you use altitude changes to get into different jet-streams and reduce propellant use?

6. How do you respect country's airspace? How do you avoid causing international incidents? If a balloon does come down, how do you ensure it doesn't harm anyone and can be retrieved? Early in the project, Loon had an ops team of former navy SEALs whose job was to retrieve any crashed balloons without starting any wars.

And I wasn't even on the team - this is just what I got from a few lunchtime conversations. They had to invent new algorithms, first to model the wind systems in the stratosphere and then to optimize balloon placement given that wind. I know their CTO - he's got a Ph.D in stochastic optimal control (pretty appropriate for this problem). The math behind control theory, once you get past simple PID controllers, is hard.

[dekhn]

Much of what you describe are pedestrian problems that could be addressed by straightforward, low-risk engineering. The issues with wifi reception around buildings were solved by AT&T when they developed the tech for mobile phones long ago. All the balloon control stuff- people have navigated balloons around the world for a while now. Using the altitude changes to navigate is standard.

That you need a team of NAVY SEALs to clean up your messes is a sign you're doing it wrong.

After working at Google for 12 years, I am always amazed at how people talk up their projects as risky when really, there's little or no risk involved, and the majority of the problems are political or social opposition to your plans, as well as the lack of a business model.

[allyourhorses]

Parent comment is classical old guard mentality Googler who thinks if something sounds big and impressive it automatically is big and impressive. It was a marketing trick that weirdly infected employees throughout the org, glad to see here the trick has all but completely lost its effect.