Learning from Neighbors in a Changing World

This paper analyzes the in‡uence of various network architectures on the ability of a group of people to adapt to a dynamic state of the world. In highly connected societies, interaction with neighbors will clog the infor- mation pipeline as individuals overvalue the information generated by the (often incorrect) actions of others, making it impossible to distinguish the signal that contains important information on the true state. As a result, people are unable to change their action over time in a complete network. When people have an unequal number of connections, those with the least number of connections can start the adaptation process and set an example for their neighbors.