System dynamicists build models and report on the results of their studies. Model builders seem to assume implicitly that passing on knowledge, which was acquired through scientific research, is sufficient to convince people to change organizational policies. Studies on the impact of computer models on policy making have convincingly revealed two things. First, in most cases the impact is conceptual (i.e., people learn from it), rather than instrumental. Second, most of the learning takes place in the process of building the model, rather than after the model is finished. It is thus not surprising that over the last decades the emphasis in the system dynamics literature has shifted more and more towards modeling as learning and modeling for organizational learning. Model building is now increasingly seen as a method to structure debate and to create a learning environment in which assumptions and strategies can be surfaced and tested. As a result, over the last decades, various system dynamicists have more or less successfully experimented with approaches that involve client groups more systematically and more deeply in the process of building a system dynamics model. Descriptions in the literature have ranged from single case studies, through fine-grained descriptions of practices, to standard procedures, such as for instance Randers' Reference Group approach and Barry Richmond's "Strategic Forum" (see, for instance, Randers 1977;1980;Richmond 1987;Morecroft and Sterman 1994).
Although much experience has been gained in the area of modeling as learning, it seems that building models with client groups is still more art than science. Although knowledge about how to construct a system dynamics model has been well codified, knowledge about how to elicit knowledge systematically from a group and how to deal effectively with group dynamics is an issue that has received far less attention in the system dynamics literature. Richardson et al. (1989) have made a first attempt to shed some more light on the issue of knowledge elicitation in group model building and Morecroft and Sterman (1994) have edited a number of papers on modeling for learning organizations and the latter include detailed descriptions of group model-building projects and issues. Richardson and Andersen (1995) have analyzed and described the various roles involved in facilitating a model building group, while, in his book on group model building, Vennix (1996) has connected the efforts of system dynamics model builders to the realms of behavioral decision making, small group dynamics and group facilitation. These are all steps forward on the way to clarifying important issues in group model building and to grounding group model building more firmly on a scientific base.