ARCHIVES Problems with causal-loop diagrams

Founded in the mid1950s, the field of system dynamics has intellectual roots reaching much further into the past. The Archives section of the Review seeks to publish material from that past which can contribute to current theory and practice. The section welcomes previously unpublished but deserving system dynamics work, classics from past system dynamics literature that should receive renewed attention, and previously published articles from other disciplines of particular significance to current system dynamicists. Contributions emphasizing the philosophy and theory of model building, validation, implementation, education, and generic structures are particularly encouraged. Submissions may range in length from notes to main articles but may be edited. Send suggested material, together with a brief introduction placing it in historical and current contexts, to John D. Sterman, System Dynamics Group, MIT E40294, Cambridge, MA 02139, USA The first system dynamics work did not include the use of causal-loop diagrams. Feedback structure was portrayed by equations or stock-and-flow diagrams. Such representations were natural for engineers. In an attempt to make system dynamics accessible to a wider range of people, causal-loop diagrams have become increasingly popular. In many texts and courses they are the first tool described. Indeed, recently several analysts have proposed that system dynamics studies can be carried out without the development of formal models at all (Morecroft 1985; Wolstenholme and Coyle 1983; Wolstenholme 1985). Causal-loop diagrams often figure prominently in such analyses. Yet even those who advocate the use of qualitative system dynamics are careful to point out that in all the successful applications of such qualitative methods the analysts have had extensive experience with formal model building. Nevertheless, it seems inevitable that people at all experience levels will continue to rely on causal-loop diagrams. In the following paper dating from 1976, George Richardson describes a variety of problems which often arise in causal-loop diagramming, both in the development of the diagrams and the explication of behavior from them. The main difficulties arise because causal-loop diagrams obscure the stock and flow structure of systems. We sometimes emphasize so heavily the role of feedback structure in generating behavior that the crucial role of accumulation processes is lost. Even experienced modelers are easily misled by causal-loop diagrams. I suggest the following experiment: take the causal-loop diagram for the family feud described in Richardson's paper and ask a random sample of system dynamics modelers or students how it will behave. In my experience, one will not only receive a wide range of answers but most of these will be incorrect. Then repeat the experiment using the stock-and-flow diagram (with a different group of people, obviously). While answers will still vary, the number of correct responses should rise. In recognition of these difficulties, there has been a revival of stock-and-flow diagrams as a means of communicating structure (Morecroft 1982). Richardson's paper should not be taken as an argument to abandon causal-loop diagrams or qualitative system dynamics, however. But it serves as a caution to the facile use of an easily abused technique. Despite their problems, causal-loop diagrams are likely to remain important tools for the communication of feedback structure.