The structures of social interaction affect individual behavior and economic performance in important ways. This leads us to ask: does the architecture of social interaction exhibit particular patterns and are these patterns stable over time? We examine interaction among economists by looking at the evolution of co-authorship relations over a thirty year period. We find that in the 1970's this world was quite fragmented with the largest interconnected group { the giant component { covering only 15% of the population. However, by the 1990's economics was much more integrated, with the giant component covering over 40% of the population. The average distance between individuals was small and declined over the period, leading us to conclude that economics is an emerging small world. A crucial stable feature of the network over this period is the existence of several stars (economists with many co-authors each of whom have few collaborators and rarely work among themselves). The world of economics is thus a collection of inter-linked stars. We also find that a growth in the average number of co-authors is the main reason behind the growth in the giant component and the fall in average distances within it. The second part of the paper develops a simple theoretical model of collaboration in economics. We find that an unequal distribution of collaborations and inter- linked stars arise naturally in this environment. Falling costs of communication and increasing credit for joint research lead to greater co-authorship and this supports a larger giant component.