Now it just needs to charge / run from solar and it can be dropped anywhere with signal. I wonder what sort of wattage this thing draws.
I think the biggest challenge to getting somewhere small enough that it can be "dropped anywhere" is definitively the solar panels. If you add the qualifier "where you have permission" it gets a lot easier - I don't think you need a very large panel to be able to power something suitably basic like this.
In my head I'm thinking an air tight tupperware container with a pi zero inside, setup for meshing. It would be cool if it was fairly flat and had panels on both sides, so you could throw it on a roof top and be reasonably sure it'll land with a panel facing up to run.
I mean it's an idea I've thrown around a lot in my head since college, but in all my scenarios, I never opened it up to the public, it was just going to be used for me. If that's the case, then yeah it's doable.
Deploying updates is not that hard - you'll want a flash that's large enough for at least two partitions so you can keep a known good version around. Other than that, signing updates isn't a big deal, and tamper-proofing key storage enough for casual use isn't that hard. I wouldn't trust sending senstitive data clear-text over a network like that, but I wouldn't do that over public wifi anyway.
Latency is a bit more of a challenge if you're somewhere that'd need lots of hops, but if you are then presumably the alternative is worse, especially if you have nodes with LTE connectivity as a possible fallback.
I keep thinking the Meshtastic network is heading in that direction, maybe Disaster.Radio might have some features, and generally there seem to be a ton of projects that contain some but not all of the pieces you'd want.
I'm down to build the hardware, I just haven't a clue on the software side of things.
If you can come up with a hardware package that is resilient enough, then "all" you need on the software side is a small Linux install and a bootloader setup that will try either of at least two boot partitions. You'll want a watchdog to force a reboot to a know good install if it doesn't respond after a while. Routing is the tricky bit, but as mentioned if the number of nodes is small enough a small routing daemon to periodically exchange routing information and discovery performance fluctuations is not a big deal.