Really depends on inter-village proximity. High clustering is preferred. Low cost narrow beam antennas can only grant so much range especially in highly forested regions. Luckily most third world regions are near the equator. Solar panels would provide all the power needed to run mesh relays and the nodes themselves at a relatively low acquisition cost.
CJDNS and B.A.T.M.A.N are good mesh protocols but the last time I checked, they were still under heavy development. If you really want to bring this technology to these areas, I would start by mapping out the regions then examining population density and distribution. Move on to climate and the level of cooperation among the population.
Start with one small region in great detail then move on to the next. It’s nice to fantasize, but it’s even better to get involved. If you need a brain to bounce ideas off of, let me know. I’ll try the best I can to help. Honestly, all that is really needed is commitment and leg work. Others will jump in if there is a clearly defined plan and a practical approach.
Come to think of it, these places are going to be saturated by tech giants soon enough. Add SAFE and boom! With their new found interest and technical knowledge, the inhabitants can transition to a meshnet with greater ease. It shouldn’t take more than 5 years IMO for the 60 - 70% of the 3rd would to join the www internet party. In the mean time we can create and refine strategies.