In recent years, with signal processing, it has become possible to communicate, at very low speeds over HF with downright trivial equipment. 100 milliwatts or so, from a simple one-transistor sort of transmitter, running at say 0.1 or 0.01 baud, can go across the continent with a wire hanging out the window. With the right coding, software-defined radio can pick that kind of signal out from beneath the noise floor.
Similarly, with higher frequencies like UHF, the cost has come down to the point that individuals, or small groups, could construct their own large-area mesh networks. A dozen LoRa radios with microcontrollers, placed strategically, could cover a large urban area. $1000? Something like that.
It will also soon be possible for a small organization, or even an individual, to construct a satellite-based data network. Cubesats can be launched for just thousands of dollars each. It's possible to communicate with such satellites using readily-available and relatively cheap equipment. Someone will eventually realize they can launch some dozens of such cheap satellites, and have their own global low-bandwidth network. A new startup, Swarm, has already launched their first set of smartphone-sized satellites, which collect beacon/telemetry data.
I suppose my overall point is that low-bitrate radio at distance is getting easier and cheaper over time. And being able to send even 1000 bits per second between many points in an urban area between mobile devices, like LoRa-based mesh networks can do, would be a very valuable asset for an individual or organization to have, if normal communications were disrupted. Unlike a giant 10 kilowatt HF transmitter in the old days, it's kind of hard to pin down where the equipment even is, or what it's for, or who is using it. Once the cost is cheap enough, nodes in such a network are effectively disposable. Place and forget. This will be an upset for those who assume that if Internet access is controlled, most or all potential communication channels are controlled.