If I surmise those projects correctly, there was a fairly important degree of central control and some kind of authority to make decisions, and the mesh networks were of limited scope. A mesh network of the size that the article presumes would be massively more complicated and would have to be able to react dynamically to outages.
The current Internet has a combination of a limited number of players for backbone and a fairly slow routing protocol that is fairly limited who gets to participate. A routing protocol that could handle something like this hypothetical mesh network would need to function a lot more like an interior routing protocol, but those only work because a single organization controls an autonomous system. To my knowledge there is no organization-to-organization routing protocol that can react with near instant speed to the changes in the network that will inevitably occur as nodes connect and disconnect from the mesh.