I sent this talk on the Whirlwind Computer to a colleague whose father worked on WW II radar development and was therefore in Forrester’s circle. His response:
“This is the best video I’ve seen in years.”
Forrester, who could have ridden the wave of Moore’s Law chose instead to pursue the very “World Model That Might Have Been” that I now am, like Forrester, intent on creating. Forrester went to his grave being ignored by the very industry he had a major hand in creating and the government that he served. (Watch the video.)
I asked GPT “Whatever became of Forrester’s “National Model”?”.
Its response contained the scare quoted phrase “a strong feeling” which would be grist for the Ayn Rand folks were it not for the fact that they’re just Randroids and are unable to be remotely as self critical as she was – otherwise they’d be all over Hume’s Guillotine as the logical extension of Forrester’s life’s work.
GPT 5.5 Pro responds
It seems to have become an unfinished but influential research program, rather than a maintained national policy instrument.
Forrester’s National Model began as a serious MIT effort to build a computer simulation of U.S. social and economic change for public-policy analysis. The 1976 description says it was intended to handle inflation, unemployment, recession, balance of payments, energy, and environment, with a deep policy structure running from government fiscal/monetary policy down to corporate accounting, pricing, and ordering decisions. At that point, preliminary sector models existed and the plan was to reformulate, refine, document, assemble, test, validate, and expose the model to outside academic and professional review. (ResearchGate)
There was also some official-policy visibility. A 1976 Joint Economic Committee volume describes the MIT System Dynamics Group as developing a comprehensive U.S. social-economic model with production, financial, household, demographic, labor, government, and foreign-trade sectors, interconnected by flows of information, people, money, goods, services, and orders. It says most sectors had been formulated and that assembly of a first full version was underway, expected to take about two years, with model assumptions and behavior to be made available for criticism from academic, private-sector, and government reviewers.
But it apparently did not become a federal forecasting/planning model. In the same Joint Economic Committee material, an OMB-related review of federal long-term growth models says there was “a strong feeling” that a central economic forecasting model would be counterproductive and too binding, partly because agency purposes were too diverse and interfaces among separate models were technically difficult. That is not a direct rejection of Forrester’s model, but it helps explain why a single Forrester-style national model was unlikely to become the government’s official modeling platform.
Inside system dynamics, the project lived on through papers, theses, and internal memoranda. Khalid Saeed later summarized the National Modeling Project as having left distinctive causal explanations for business cycles, Kuznets cycles, and Kondratieff/long-wave cycles, but he also says the details were scattered across Ph.D. theses, occasional papers, and internal MIT System Dynamics Group records, “not published in a succinct form” and not widely known to newer system dynamics practitioners. (System Dynamics Proceedings)
Forrester himself did not regard the effort as simply abandoned. By 2003 he described the model as something he had continued to work on for decades. He said it had once been larger, but had been simplified to “somewhat over 200 levels and 1400 auxiliary equations,” and that he had spent recent years simplifying its structure, sharpening its focus, and making it more robust. (System Dynamics Proceedings)
More importantly, by 2003 he had reframed it. He said the project was initially called the System Dynamics National Model because it was seen as representing the United States, but he had come to think that modeling a specific country was the wrong framing. The model, he said, now represented “a general theory of economic behavior” and could be parameterized for many industrial countries or even the developed world. (System Dynamics Proceedings)
The planned definitive publication seems never to have appeared. In 2003 Forrester said he was starting a book that would include a disk with the complete model and simulations of policy alternatives. (System Dynamics Proceedings) In 2013, Saeed wrote that he had asked Forrester about publication of the National Modeling Project and that Forrester said he was working on a volume titled A General Theory of Economic Behavior; Saeed added that “the book is not yet out.” (System Dynamics Proceedings) Forrester died in 2016, and I find no evidence that the book-plus-model-disk was ever published. (MIT News)
So the best summary is:
The National Model survived as a body of theory, papers, dissertations, internal memos, and archival material, but not as a widely available, maintained, executable public model.
MIT’s Forrester papers now include an archival item titled “System dynamics, study of (national project), 1960s–1980s” with a large scanned PDF, which is probably one of the main public traces of the project. (Dome) But that is an archive, not a living “national model” in the way an IAM, DSGE model, or federal forecasting system might be maintained.
My read is that the National Model had three afterlives:
-
As Forrester’s unfinished general economic theory.
By late life, he no longer saw it mainly as a U.S. national model, but as a generic theory of industrial economic behavior.
-
As an intellectual ancestor of later system-dynamics work.
Its ideas fed work on economic long waves, supply-chain dynamics, management flight simulators, and Sterman-style system dynamics pedagogy.
-
As a cautionary case.
It exposed the tension between Forrester’s ambition for comprehensive policy laboratories and the practical difficulty of publishing, validating, maintaining, and socially legitimating a huge causal simulation model.
That last point is probably why it matters for your “many analysts, one dataset” question. The National Model had the ambition of a shared policy laboratory, but the infrastructure for transparent, adversarial, reproducible model comparison did not yet exist. The model became a school of thought rather than a civic modeling institution.
