An application can be designed to communicate over more than one protocol. There is more than one transport protocol used on the internet. For example, the UDP

An application can be designed to communicate over more than one protocol. There is more than one transport protocol used on the internet. For example, the UDP protocol may be used for for P2P / mesh networks. Applications that support UDP may also support TCP. TLS runs "on top of some reliable transport protocol (e.g., TCP). UDP is not a reliable transport protocol. There are also transport protocols other than TCP / UDP [1]. Anyway, if you are making your own mesh that does not require you to have access to an ISP, then maybe you make that mesh over something other than IP. If you did use something other than IP, that would not prevent any of the nodes from including IP in their networking stack. In that scenario, a human user of the mesh might believe that they are completely off the grid / internet, but surprise, all their traffic or leaked metadata is actually being aggregated over the internet because one of the nodes is connected. I mainly brought this up because it popped into my head and I thought it might make for an interesting discussion. Also, because I have thought about making a distributed mesh network that does not use IP where nodes communicate via HAM radio. [1]: https://en.wikipedia.org/wiki/Transport_layer https://en.wikipedia.org/wiki/Transport_layer