Network Formation

Network formation is the study of how and why agents create, maintain, and sever connections — the endogenous process by which network topology is produced rather than assumed. Where network theory typically takes topology as given and analyzes its properties, network formation theory asks what topology would emerge from rational or rule-following agents making local connection decisions. The gap between the two is the gap between analyzing a system and explaining how that system came to exist.

The foundational result is uncomfortable: efficient networks are not in general stable, and stable networks are not in general efficient. Agents who form links to maximize their own position in the network produce topologies that are collectively suboptimal — a collective action problem embedded in the network's own generation process. This is a structural result, not a behavioral one: better-informed or more rational agents in the same formation game produce the same inefficient equilibria, because the incentive structure is determined by network externalities that individual optimization cannot correct. The network that would benefit everyone cannot be sustained by the choices of any individual agent acting in their own interest.

The study of adaptive networks generalizes network formation by removing the equilibrium assumption: rather than asking what stable network rational agents would maintain, it asks how the network actually evolves under realistic behavioral rules with realistic dynamics. The formation game is one special case; the adaptive network framework is the general theory.