After the war, Professor Forrester led the team that designed the Whirlwind 1, the first general-purpose digital computer at MIT. Robert Everett, who worked with Professor Forrester during those years, recalls him as an exceptionally versatile and resourceful innovator. When vacuum tubes proved too short-lived for reliable high-speed data storage, Professor Forrester invented random-access magnetic computer memory, a forerunner of today’s DRAM chips. Already, he had a reputation as a perfectionist, prone to accomplishing the impossible, with little patience for those who didn’t measure up to his standards. When a new receptionist was having difficulty typing labels on file folders, Professor Forrester typed them out for her during her lunch break.
“I think there was nothing anybody in the lab could do that he couldn’t do as well or better,” says Mr. Everett. “That tended to be tough on the people who worked for him, but they knew there was nothing personal about it. He was almost always right.”
Professor Forrester’s early inventions and patents earned him a place in computer history, and he could have gone on to a long and lucrative career in the new industry. Indeed, the founders of Digital Equipment Corporation, the minicomputer company credited with sparking the technology boom along Boston’s Route 128, were all his graduate students from MIT. But by 1956, he felt that the pioneering days in digital computers were over, and he craved a fresh challenge.
He found it in a faculty position in the newly formed MIT School of Industrial Management, later renamed the Sloan School of Management. A group of executives from General Electric had come to MIT for help; their household appliance plants in Kentucky oscillated between periods of peak demand, when everyone had to work overtime, and slumps that lasted long enough to force layoffs. So Professor Forrester interviewed GE’s manufacturing people and charted the impact of their hiring and inventory decisions on orders and sales. The resulting pattern looked surprisingly like the technical patterns he had seen with servo-driven cannons in the military; the first shot would overshoot its mark, the next shot would overcompensate, missing the mark in the other direction, and the whole system would miss and correct itself several times before finally connecting with the target. GE’s pattern of overcorrection was exaggerated further by delays in the ordering process and poor communication between manufacturing and distribution. (See Exhibit 1.)
Stocks and Flows
In what would be Jay Forrester’s modus operandi for years to come, he tapped one of his graduate students to write a program, later named DYNAMO, that could translate his pencil-written calculus into the ones and zeros of a computer’s machine language. In naming this new field, he used the engineering term dynamics — which commonly referred to the interplay of physical or electrical forces over time — to indicate that his models didn’t simply represent a snapshot of a situation at any given moment, but an opportunity to see situations grow and evolve. Today, system dynamics students use sophisticated modeling programs, “playing out” the impact of possible policies or strategies by entering them into the model and running it like a computer game. In the 1950s, Professor Forrester wrote all of the code himself, using the primitive programming tools of the time.
The conceptual basis of his models was the critical relationship he had originally observed in servomechanics: the way that stocks and flows governed each other. Consider, for example, the flow of water into a bathtub. A person turns the tap to fill the tub, and when the level of water in the tub (a stock) is nearly full, the person turns off the water (a flow). The chain of cause and effect is actually a feedback loop; the person and the level of water influence each other. Similarly, a company’s reserve of cash is a stock; its profits and losses are flows that affect the level of that cash reserve. But when the cash reserve gets too low, the company’s managers do whatever is needed to increase cash flow. No single factor dominates; they all influence and regulate one another. In Professor Forrester’s world, what goes around does inevitably come around.