Transport networks and computer networks can be described with the same mathematical framework: the infrastructure can be described as a graph and on top of it a set of agents compete for its capacity, each aiming to selfishly optimize its own objective. Such agents are passengers in transport networks and TCP origin-destination pairs in computer networks.
In this talk I will review classic models that help to understand what happens in such networks from a macroscopic point of view. I will show that by acting selfishly, agents are unwittingly also collectively solving a centralized optimization problem. This can be interpreted as a collective behavior resulting from the interaction of multiple selfish agents.