Project Abstract There are many situations in which a new social trend `diffuses` throughout society, including buying a certain brand of computer, using new planting and irrigation techniques, and participation in a political interest group. In these situations, there are two crucial elements. One is the element of individual choice: each person's decision is influenced by the decisions of others (if all my friends buy Macintosh computers, I would rather buy a Macintosh also), but it still comes down to my personal decision. The other is social structure: my choice is influenced by who my friends are. Although both elements are crucial, currently in the social sciences, there are very few explanations which include both. This project develops mathematical models of collective action which incorporate both individual rationality (people try to further their goals) and social structure (the patterns in which people know and interact with each other). Hence the project uses concepts from both game theory and social network theory. The key assumption is that explicit communication is the most basic mechanism for coordination, and hence people are modeled as members of a communication network, in which each person talks to her friends. The main preliminary result is that cliques, subgroups in which each member knows every other member, are sufficient and in some sense necessary for collective action. Cliques, a basic concept of social network theory, are crucial for collective action because they generate common knowledge, a basic concept of game theory. This result has implications for understanding the importance of ``strong'' versus ``weak'' links, centrality, and dispersion, and suggests an empirically verifiable distinction between ``strategic locality'' and ``information locality.'' Rather than seeing structural and strategic perspectives as opposed, by combining them this project can find insights not obtainable from either perspective alone.
