The idea is by sticking with LEO satellites vs geosynchronous the ping stays reasonable (~700 miles ~= 9ms vs ~22,236 miles ~= 279ms ping ping) and the bandwidth is much higher than traditional satellite internet. Further you don't need any effort aiming a dish thus vastly reducing installation issues.
https://en.wikipedia.org/wiki/Teledesic
It's interesting to me that our cities and their governance are so awful - with their limited voter participation, poor cost control, class warfare politics, crazy fringe political advocacy, and unreasonable franchise agreements - that going into space is considered easy compared to working with city governments to get fibre in the ground.
If SpaceX can prove reusability, which is looking likely, putting up a constellation like this will become significantly cheaper. They also can just reuse first stages that their own customers already paid for (their customers pay for the flight, not the stage itself).
http://www.popsci.com/samsung-wants-launch-thousands-satelli...
ED: Here is the paper. https://arxiv.org/ftp/arxiv/papers/1508/1508.02383.pdf
Now, the costs of these satellites and receiver's will make or break this idea. But, it's surprisingly viable even if they only cover the continental US to start.
PS: Iridium NEXT (https://en.wikipedia.org/wiki/Iridium_satellite_constellatio...) is aiming for some data bandwidth. Though presumably at very high costs.
They also have to get radio spectrum for the uplink everywhere they operate. Have a good look through the history of Iridium (Iridium PREVIOUS?) and all of the unexpected costs Motorola ran into building and maintaining a satellite fleet.
If you go to the middle east and many third world countries much of the Internet is still satellite. If a particular country wanted a way to insure that they still had access to most of the world's internet traffic (as economies grow and infrastructure is built) they might prefer a constellation of LEO sats capable of carrying the entire planet's traffic in a point to point fashion owned by a US based company who is very reliant on the USG for most of its other business.
But would they want this to exist enough to pay for any of it or have a hand in maintaining it? Beyond how much they already are by awarding SpaceX government and NASA launch contracts?
And all because muni government sucks.
Most people rarely, if ever, need to deal with municipal departments that regulate infrastructure, health, and safety. As soon as you do, you find out that they're pretty uniformly terrible places with terrible people dealing with terrible things.
It looks like they will beat both SpaceX and Google to the task - which is unfortunate for them since the founder of OneWeb was involved with attempting to develop the same project with both companies.
http://spacenews.com/oneweb-gets-1-2-billion-in-softbank-led...