For dense areas, mesh applications like BitChat (Jack Dorsey) could bypass the need for a network with p2p bluetooth mesh networks. And works with existing devices, vs something like meshtastic which needs an installed base (afaik).
For OP's situation I think runners and a store and forward system like Scuttlebutt/Briar/etc might work better. But I'd love to see a couple of thorough case studies on that kind of system, they've been around for many years targeting related scenarios.
I think the issue is more that getting a message many kilometres out using a couple of nodes sitting in non prime locations is just unreliable. The noise is too high. There’s also a limit on how many messages can be sent while using flood routing.
I don't think it's the noise that's too high, my original understanding of the meshtastic 2.0 protocol (from this video: https://www.youtube.com/watch?v=7v6UbC5blJU) was that an edge device would receive priority. So the message might go the wrong direction, say. Or, worse, a meshtastic radio in a basement might get priority.
Because they didn't understand (or care about) network topologies from the start and reliable message transmission, not to mention the versions of devices that are still on their network, I'm guessing it's gonna be a couple more years before they finally get their act together.
In Melbourne, Australia I was able to pick up 450 nodes on Meshtastic while I got zero on MeshCore after weeks of searching.
It's one of these things where for the early adopters it worked great, but then the late adopters got pissed off because it didn't work.
As for meshtastic: I'm one god damned fucking hop away from my friend through a repeater. And it doesn't fucking work!