This article presents a rudimentary model of collaboration with the aim to understand the conditions under which groups of scientists will endogenously form optimal collaborative groups. By analyzing the model with computer simulations, I uncover three lessons for collaborative groups. First, in reducing the cost borne by scientists from collaborating, one benefits the members of the group. Second, increasing the number of potential collaborative partners benefits all those involved in a collaborative group. Finally and counter intuitively, this model suggests that groups do better when scientists avoid experimenting with new collaborative interactions.
