Our preliminary model identified specific CSFs and associated them with specific hypothesized stages of the collaboration process. One problem with this conceptualization is that the cases do not appear to evolve in predictable stages but evolve in an iterative process of feedback, learning, and change. Moreover, by trying to identify only a specific set of CSFs, we risked ignoring other factors that may be more relevant to our area of study. Fortunately, the open-ended nature of the interview questions allowed interviewees to describe success factors that were not predicted by the literature. For example, in some cases participants emphasized the importance of a mutual “need to succeed.” In the Cadastre Quebec case, that need was based on the desire to regain the mutual and external credibility of the two participants after a public failure. In the New York GIS case, this need reflected an acknowledgment by an informal community of practice that their goal of a statewide spatial data program would not be achieved unless they cooperated informally, relentlessly, and without compensation to keep the issue in front of policy makers. In Hotjob, the sponsoring organization changed its name as well as its operation in order to disassociate itself from previous public dissatisfaction. Other CSFs included the value of networks of personal and professional relationships for working through the problems that were not addressed by formal agreements or fixed work processes, as well as “agreements to disagree” about certain issues whose resolution was not essential to the early success of the collaboration. Some interviewees pointed out how voluntary personal leadership, regardless of formal position, led to important progress toward their goals. Willingness to accept risks and manage them skillfully for the mutual benefit of all partners constituted another critical success factor in a number of cases.