The United States is rich with political and social institutions, which create networks over which politicians and citizens communicate, coordinate, and cooperate. Traditional positive political theory, with its emphasis on two-player games as models for strategic interaction, often ignores the complexity of networked coordination and cooperation. This dissertation argues that network structure influences strategic outcomes in complex ways. Specifically, more network connections do not necessarily help groups to solve collective problems, as is often claimed in the deliberation, social capital, and social networks literatures. Chapter 1, "Bad Connection," shows that when individuals attempt to solve a distributed coordination problem (in which connected dyads must adopt compatible actions), adding connections can actually inhibit coordination. This chapter identifies the theoretical conditions under which additional connectivity is likely to degrade group performance in a common coordination problem, and presents experimental evidence to support the theory. The need to avoid the negative externalities of an over-connected network has bearing on a range of real-world political problems, such as the design of American executive agencies and the allocation of policy portfolios in parliaments. Chapter 2, "Segregation and Compromise," shows that players of a networked, 16-player battle of the sexes are much more likely to reach consensus when the actors with conflicting preferences are integrated in the network than when these actors are segregated. I argue that this suggests that social sorting may be a driving force behind political polarization in the US; as social sorting segregates liberal and conservative Americans, compromise and consensus become increasingly difficult. In Chapter 3, "Does Social Capital Habituate Cooperation," I draw a distinction between two popular theories of how denser networks generate cooperation; the social habits theory (dominant in political science) assumes that participation in social organizations acculturates members to cooperative norms, while the social incentives theory states that the shadow of the future makes cooperation rational. I show that social connectedness does not predict subjects' anonymous choices to cooperate in the lab, contradicting the popular social habits theory, and calling into question the argument that a more connected society is a more cooperative society