What does it mean to cooperate? Who has the power of decision? Shall one be the spokesperson who speaks for others? Or, is power everywhere and nowhere, because it is diffuselydistributed over what neural network scientists call the links in the system? Every large system confronts designers and users with this problem. Shall obligatory passage points be set up (rules, standards, centralized control mechanisms) for managing the information flow, or should distributed models be used to diversely activate the information and dynamically connect people, strategies, and things? There is no one clear answer; rather, the alternative itself becomes a point of research that opens up a wide range of options for creating learning algorithms for working together.