To answer your most salient question(about whether it provides internet connectivity)
>We provide internet through peering at an internet exchange point. If you have line-of-sight to this “supernode” in downtown Manhattan you can have a fast, reliable connection with suitable hardware, and replace your ISP.
Isn't the point of mesh networking that you only need line of sight to someone else on the network who has line of sight to someone else (etc) who has line of sight to the supernode?
If we take a standard network for example then all your routers can be meshed for each router the primary backbone switches can also be in another mesh but the edge switches that say connect individual users are not meshed.
This would still count as a mesh network despite having 3 tiers with different fault tolerance / handling characteristics for each tier.
Ideally, all I want is a mesh node that I can buy, plug into a power supply, then have an app (accessed via the legacy internet/mobile network) that displays a map telling me how detached I am from the bulk of other mesh users. If I have limited/no connection, I would gladly knock on commercial/residential doors between myself and the bulk network and offer to buy them a node, to "build a bridge" and improve my connection.
The two challenges seem to be reliability and expansion to a global grid without relying on legacy backbones. For the former, resilience will of course improve with increased node density. For the latter I'm sure it will take a well funded, coordinated effort. I would gladly pay an annual tax to a new gov't agency - perhaps an extension of the National Park Service (1), to build and maintain a grid of mesh nodes accross the US, with connections to the legacy GIG backbones.
1.) The National Park Service preserves unimpaired the natural and cultural resources and values of the National Park System for the enjoyment, education, and inspiration of this and future generations.
IMO, the internet is an information landscape, very much tied to our geographical landscape, that should be free and protected just like any river or mountain.
As a knowledgeable internet consumer, I come to it with the general question, "What would such a mesh network do for me?" to which the only answer seems to be, "It gives me the fascinating task of buying a router and flashing its firmware and writing its config file, all based on sketchy jargon-filled docs."
Supposing I do that, and in the highly unlikely outcome that I don't screw it up, I will then have -- what, exactly? Besides a nonstandard router?
Supposing there existed a libremesh network in my city. Not impossible. How would I know? Could I use it to access internet content?
Another crucial question not answered is, how is a local mesh supposed in any way to be a circumvention of the traffic-shaping that we all fear the telecoms will do, now they aren't common carriers? Yeah, to the extent that a whole neighborhood is meshed up, you could have free and unfettered exchange of email and FTP among all your neighbors.
Yippee.
The content that an internet consumer really wants, and which occupies 99.9% of the bits she consumes, comes from the major sites that are out there on the backbone of the net. Ninety-nine percent of what the edge users of a mesh want to pull from their mesh connections originates far outside the neighborhood. And it has to enter the local mesh from an internet access point that is sourced by one of those big bad telecoms.
A neighborhood mesh looks like a fine way for a neighborhood to organize block parties and garage sales, but it really is just a way to share access to some ports to the same un-neutral internet everyone else uses. Except you've added a random number of local wireless hops to every ping.