I wonder if anyone out there (perhaps anukul?) remembers the Borg Project that Joshua and I did independent study semesters for at CMU. The project was (is?) an attempt to construct a network entirely out of identical peers, which would find each other and adjust to changing node conditions quickly. The difficulty was to make this work without having a central registrar to act as the organizer. IIRC, the project was bogged down with difficulties stemming from this basic problem.
It seems that biology has provided the answer. A German university has constructed a model of a self-assembling network which is based on a technique similar to the way ants leave pheromone trails.
This could also potentially be seen as a working example of nonhierarchical and decentralized, yet successful and constructive social organization. For some reason, this pleases me.