>>>http://www.bbc.co.uk/news/technology-22905199
>>
>>That's an idea that's been explored for some time e.g. http://en.wikipedia.org/wiki/Stratellite , but Google's looks like the first stab at it by a company with serious financial clout.
>
>That, on the face of it, sounds like a more sophisticated approach. Powered to mainstain a stable position and with a much wider footprint.
>
>As the article states, I can't see how the Google version will even be able to reliably dictate the course. If the only option you have is changing altitude to select the best wind direction you'd still be limited to, say at most, 20-30 degrees. And although it is easy to cause a balloon to descend I'm not sure how they intend to cause an ascent - you have to re-inflate the balloon somehow which implies carrying reserves of gas. And with only a 20km reception radius you'd need an awful lot of them to provide continual coverage.

I get the impression Google is going to deploy a lot of them - enough to provide a good statistical chance of coverage. Like mobile cell coverage in reverse - instead of fixed cell stations handing off because a phone is moving, mobile balloon "stations" hand off when fixed ground stations go out of range. I believe other comms networks use a similar strategy e.g. Iridium with LEO satellites.

I think you're right, any requirement to regularly ascend/descend will ultimately limit the duration. But, there might be clever ways to make use of warming during daylight and cooling at night e.g. ascend into westerlies during the day and descend into easterlies at night. I don't know if there are any hydrogen or helium compressors that might be used rather than just venting gas for a descent.