Aside from blogging, I, in fact, also do research and produce academic works. Typically I try not to blog excessively about my own work, but it seems a pity not to try and bring the two together from time to time. So, I wanted to draw attention to a two-part series appearing in the latest Science in Context that I wrote with my friend and former grad school colleague Lambert Williams. Both papers have both our names on them, and were formulated largely in tandem, but (as the writing styles will evidence) we do have our respective halves of the series.
My half is a new consideration of Jay Forrester’s system dynamics simulation methodology, which he originated at MIT circa 1960. It’s best known through its role in the 1970s “Limits to Growth” affair, but rather than recapitulate the tit-for-tat of the various proponents and critics of his simulation project, I wanted to try and elucidate what made this project so appealing to him that he has remained with it through the present. To a large extent, system dynamics has meshed into the larger background of computer simulation surrounding it, even when it first came out, but Forrester has always understood it as peculiar and opposed to the work of others.
The main key to understanding Forrester’s simulations, I argue, is to understand its origins as a pedagogical device that he originally called “industrial dynamics”. As a pedagogical device, the simulation is meant to increase subjective understanding of why systems exhibit the behaviors that they do, rather than to produce statements of objective authority (as is often imagined in the study of science, especially at postwar MIT). The first section of the paper is a straightforward institutional prehistory of Forrester’s involvement in what is now called the Sloan School of Management at MIT. Forrester’s involvement with the Sloan School has always been narrated based on his recollections, but I uncovered further details while working on the history of operations research at the MIT archives. Of particular interest to me was an application for foundation funding for the school drafted by Forrester before he was invited to join. I interviewed Forrester, and he did not remember this, but it clearly exhibits notional traces of what would become industrial dynamics, but is also useful because of the broader perspective on management it shows before Forrester had a specific project to promote and defend.
The second section is a straightforward textual analysis of Forrester’s seminal 1961 work Industrial Dynamics. Hints of Forrester’s epistemological ideas have surfaced in other accounts of his work, but I have never really seen a fully satisfactory exposition. At any rate, since the history of science doesn’t operate in the “commentary on the canon” mode as intellectual and philosophical history do, and since I don’t expect many of my readers to be familiar with the Forrester secondary lit, it seemed like a useful thing to detail.
The two-part series, as the title “The Epistemologies of Non-Forecasting Simulation” might hint, is, in true history of science fashion oriented around the epistemic imperative, though one could combine it with other sources on the history of computing to help sketch out the various traditions. Lambert, who is more formally trained in the philosophy and sociology of science, is an avowed adherent to addressing an epistemological problematic (thus his central concern for the “plasticity of error”, though I hasten to point out his long line from the postwar Princeton meteorology project through Lorenz at MIT to UC Santa Cruz in the ’70s and ’80s). Whereas I have found occasion to complain about its control on history of science writing repeatedly. It’s because of these views that I’m most proud of the third section of the paper, which chronologically links and differentiates industrial dynamics from just prior trends in business-cycle economics and operations research.
Either industrial dynamics has been linked to economics and OR as all part of some vast technocratic moment in American history, or authors have noted Forrester’s dissatisfaction with those fields. In preparing his foundation proposal, though, Forrester had familiarized himself with their methods and concerns, and only then decided to do something he regarded as different. Yet, Forrester certainly did not supersede them, nor were his reasons for dismissing them really very fair. Being careful not to dismiss either side of the argument (such as it was), I wanted to understand Forrester’s dissatisfaction, but also understand their own sources of epistemological and practical power.
Finally, I just want to point out that this paper has generally avoided broad contextualization. There’s no “Cold War and…” or “Postwar Industrial Society and…” to be found here. Naturally Forrester’s work was not independent of these things, but they don’t have much explanatory force. I’d particularly draw attention to the epistemological links drawn in the series between Forrester and the countercultural UC Santa Cruz Dynamical Systems Collective in Lambert’s half. I persuaded Lambert from his initial thought that the series was going to be about the more obvious differences between his history and mine (industrial control versus investigations of chaos), and I think the series is more powerful for our coming to an agreement about that.
I’ll leave it at that, but I’d like to get Lambert involved here soon to talk up his half, and maybe to discuss further the collaborative process. I think post hoc discussion of papers might be a useful function of blogs, and would like to give it a further try.