> Routers are pretty cheap, and nice routers can cover half of a city block.
Proper ones aren't really cheap. And to cover a reasonably large area with a big number of users (and I suspect everyone and their dog has some WiFi-capable device nowadays, even in poorer countries) one needs a reasonably large quantities of routers.
Oh, and if that was the idea, I would really warn against trying to provide public connectivity this with commodity out-of-shelf consumer-grade routers. Prototypes work, but it doesn't scale well. My company (a small ISP) had tried and I'd say, in my personal opinion, we mostly failed (i.e., I mean, commodity routers are not fit for such job), even though our setup was much simpler than mesh as we had every router connected to a backbone.
Short summary of issues others should be aware of, when dealing with consumer-grade devices:
- First of all, QoS is a must - and it's a bit tricky to implement. Without QoS just 3-5 torrenting users will completely take the AP down. Getting QoS to work (and perform well) on cheap routers is hard, especially considering it's mostly done in software and their CPUs are slow.
- Another problem is, routers seem practically limited in number of active peers they can handle, and putting too much of them too nearby had its own issues due to signal interference. Not that this is unavoidable, but it may add some headache.
- And another issue is hardware supply. Imagine you polish a firmware so it works acceptably fine and consumers are satisfied. Next year, those routers go out of stock and you can't really get them anymore - only newer revisions or even models are available. A hardware change even between same model revisions could be drastic, so the development struggle never ends.