This is going to work really well... not
Think about it. Compare this mythical 50-mi radius super WiFi to an existing hotspot. Or cell tower, for that matter...
1 - Contention. how many clients will be in that coverage footprint, competing for the bandwidth. Radio is a shared medium - only one source can be using it at a time (disregarding exotic and expensive tricks). So you split it up into channels - there goes your bandwidth. And you MIMO the area into sectors - bummer if you live on a sector boundary and bounce between them. No matter what you do, you have to divide a limited resource among a whole lot of users. Suddenly, small local cells look a lot better.
2 - Power. Sure, your local TV station gets great coverage (or since digital, not so much). They've got a 50-Gazillion-Watt transmitter, and it's one-way. How much power will your laptop/tablet/phone/etc. need to talk reliably to a base station 50 *miles* away? At a decent data rate, with the interference of everybody *else* trying to get the attention of that base? It's hard enough to do on analog *voice* systems. If you thought hidden-node problems were bad with WiFi, you ain't seen nothing yet! Oh, and how big are the antennas going to have to be for these lower frequencies (compared to 2.4Ghz)? The next iPad will have a band around *it* for the antenna....
3 - Infrastructure. How many of these mega-APs will get to be in a given area? Does everybody get one (hey - no license)? It's not going to be easy or cheap to backhaul all of those clients from your huge central site. It's simple to serve a small area at a time, and the cell companies certainly have the hand-off issues worked out (well, mostly). But the only long-range two-way systems out there are fairly low-bandwidth and server relatively few nodes.
You can have bandwidth, coverage, or population - pick 2.