|
SD and critical pedagogy
Posted by John Sterman on 10/12/2008
In Reply To:SD and critical pedagogy Posted by Karl North on 10/10/2008
I have read some of this thread with interest, and now am compelled to weigh in.
There are strong pressures for specialization within the academy. Certainly there is system dynamics (and other!) research that focuses on problems that are excessively narrow from the point of view of some of you, and perhaps from the point of view of what's best for society. Jay, among others (myself, Dana Meadows, George Richardson, Barry Richmond, Peter Senge, many others) have been quite vocal about the failures of excessive reductionism and the need for research that takes a broader perspective, crosses disciplinary boundaries, extends time horizons, and explicitly engages multiple stakeholders.
However, I would make several points.
First, there is a great deal of system dynamics work that does exactly this (and also other work in academia -- SD people do not have a monopoly on broad model boundaries, cross-disciplinary research, etc.). A comprehensive list is not possible here, but it includes Jay's original Urban Dynamics and World Dynamics work, and all the hundreds of studies that flowed from them. There is recent work on climate change, on fishery management, on energy, water, and many other issues where economics, ecology, politics, sociology, etc. come together. There is work examining why organizations get "stuck" in inferior and harmful ways of organizing and operating, why process improvement and change are so difficult, and how they can get out of the trap. There is work examining the dynamics of global supply chains, which is not merely an issue of how firms can improve profits, but also intimately affects consumer welfare, working conditions, and the environment. There has been a huge volume of work on energy and electric power markets. And on and on. It would be helpful if those involved in this discussion learned more about the literature and current work; there are various resources available that can help, including the System Dynamics Review and other journals (much of the best SD work is not published in SDR but in other journals), the conference proceedings (many available online), textbooks such as Business Dynamics, and others.
Second, the scope and architecture of a model depends to a great extent on its purpose. We don't and can't and shouldn't try to build a model of everything to ensure that all clouds are challenged and all feedbacks captured. It's impossible, and attempts to do so have always failed. Bigger and more detailed models aren't always, and perhaps not even usually, better; it's a question of building a model that is appropriate to its purpose. That purpose should be a worthy one, and the modeling process should be designed so that at least one of the following results is likely: improvement of our understanding of important processes (adding to scientific knowledge), change for the better in the real world (successful implementation of beneficial policies), reflection and dialogue about what "beneficial" means (beneficial for whom?), improvement in knowledge about and methods to accomplish the first three. I believe these all go together -- one cannot expect to design and implement successful beneficial policies without deep scientifically rigorous knowledge of the underlying processes in the relevant situation and a process that engages all relevant stakeholders.
There is a limit on how broad the boundary of a model can be and still have policy impact, given the mental models and inquiry skills of the people one needs to influence in order for implementation to take place. In other words, modeling for policy impact is inherently situated in a social and political context, and politics is the art of the possible. In many cases, a model with a very broad boundary may not result in any implementation, while one with a somewhat narrower one may. Of course this is not to be interpreted as a Machiavellian statement that modelers should pander to clients just to have influence. To the contrary, I and others are explicit about the need for modelers to consider the ethical dimensions of their work, to advocate for broader boundaries, longer time horizons, and expanded stakeholder groups, to push back on clients who would use the work inappropriately and to speak truth to power. A very good treatment of these issues is Dana Meadows' and Jenny Robinson's book, The Electronic Oracle, which has recently been reissued by the system dynamics society (http://systemdynamics.org).
Third, modeling, as Jay argued long ago, should be seen as a process, not a product (see Forrester, 1971, "The Model vs. a Modeling Process", reprinted in the SD Review, 1985, vol 1, 133-134). In a research program there may be a need for detailed models of particular processes that may appear to be excessively narrow if taken in isolation. These models can inform and build confidence in the formulations of models of broader scope in which all the detail of the submodel cannot be represented, and may be socially and politically important in building confidence among colleagues, funders, and others so that the overall research program succeeds.
I do not know the particular model of N uptake by cows described in Karl North's post (below), but I can imagine legitimate reasons why such a model could be useful and relevant. It may not be; I don't know. Personally, I agree with the critique of our agricultural system Mr. North offers. I however would not wish to critique the Cornell N-uptake model from the point of view of one purpose (reforming the US agricultural system and "the institutions of capitalism") when that may not be the purpose of or context for which the model was designed. Perhaps the discussion on this list would be advanced if the authors of the Cornell work had been asked in advance about the purpose of their work (too late for that, but not too late to bring them into the discussion now). It also seems presumptuous to make strong attributions, apparently without any data, about the beliefs of the authors of the Cornell work and "scientists at Cornell" in general, whom, we are told, "never" question the "institutions of capitalism", which, we are also told (without evidence) are the real problem. Here at MIT and the MIT Sloan School of Management there are many faculty members who frequently raise these issues and question the institutions of capitalism; I would be surprised to find it's different at Cornell.
In addition to "asking why five times" (which, by the way, is an old tool in the total quality movement, not invented by Peter Senge, though nicely presented in The Fifth Discipline), one of the most useful tools to surface our mental models is the "ladder of attribution" (see Argyris et al.). In Mr. North's comments I see many attributions, mostly negative, and in every case unsupported by data, attributions about particular individuals (the authors' of the Cornell model), attributions which are then generalized to entire groups ("scientists at Cornell" and "our academic culture"). It may turn out that Mr. North is entirely correct about the Cornell model and the beliefs of Cornell scientists, but there is no data provided at present. A fundamental principle of system dynamics is that the behavior of a system arises from its underlying structure; a corollary is that blaming particular individuals and groups for the problems we experience is usually a low leverage intervention. It doesn't work, and, often, provokes defensiveness and resistance that rewards blame with blame, eroding our ability to cooperate for the common good. I think the quality of the discourse on this list and its impact would be improved if everyone took care to ensure that posts avoid unsupported attributions about the motives, beliefs, and capabilities of other individuals and groups.
John Sterman
|
