Is store-forward over temporary wifi between smartphone and pirateboxes, a plausible alternative to true mesh networking?

Instead of having pirateboxes talk to eachother automatically (requring complex coding), would it be possible to create a store and forward system using smartphones as a sneakernet?

So essentally the user would install an android/iOS app. When they come to the vicinity of a recognized piratebox, a temporary connection is opened. From there, a manifest/messages/files is exchanged before closing.

The messages the user receives may be forwarded to the next piratebox, the next time they walk past the location. (To keep power usage low, its probbly nicer to only transmit file list and small messages.)

With this store/forwarding system, you cannot exactly have a real time conversation. But the rather large size of smartphone memories does allow for carrying enough data, between pirateboxes to start a psudo-local UseNet network.

This has the advantage of cost savings in that you do not need to build radio antenna strong enough to reach the other box (or enough boxes inbetween to link together via mesh). Also the link via sneakernet is not line of sight, and can traverse all sorts of terrains by foot. Just have to deal with latency, and the lack of control of packets.

Sorry for resurrecting an old thread but it already fitted my topic so well

The serval-mesh project kinda works in this way. [www.servalproject.org]
So far they have an android app running and some daemon for unix-like systems with a simple cmdline interface.

I just recently started working with the code (https://github.com/servalproject/serval-dna) to evaluate it for emergency networks. So far file distribution seems to work on TP link routers and messaging also... kind of... documentation is a bit sparse sometimes but it already has quite a few interesting features besides "just filesharing".
The nice thing about serval is that it can use whatever link is available, bluetooth, wifi direct, base station mode, mesh and even serial links!

The Rhizome "file system" sounds like exactly what is needed. How easy is it to implement and use with existing services? It seems like an app would need specific integration with it.

I'd also like to know more about any evaluation of its security from a third party.

On a very light-weight scale, there was a project I had found years ago called FETHR. (http://dsandler.org/brdfdr/doc/iptps-fethr/) It was the product of a research paper trying to find a technical means to decentralize a microblogging service like Twitter. It provides distributed, secure, and authenticated relay of messages.

There was a python implementation of a microblogging system - something that perhaps could be evolved into a threaded forum.. So there's yet another approach that's not a drop-in solution.

At the moment there is a native Android app that exposes much of the rhizome functionality to the user. On unix-like systems some functionality is available on the commandline but not very userfriendly. I started developing a simple qt gui for some of the functionality powered by the commandline interface. A better way would probably be to use the rest-interface and develop some kind of pretty web frontend for it. But so far I have only tested the system with a few nodes in direct sight, I couldn't find any information on how it performes in larger networks.
So far we managed to build it for various OpenWRT builds, Mac and Ubuntu Linux, the Android APK also works, just a nice iOS client is missing (some design docs are already there)...

Another interesting way for distributed datastores is groundstation (https://github.com/richo/groundstation) unfortunatly it seems to be discontinued. It uses git objects for storage and has a PoC web frontend called airship with some twitter like functionality.