Scan barcode
A review by mburnamfink
Thinking in Systems: A Primer by Donella H. Meadows
4.0
Thinking in systems is a fascinating, if somewhat dated book. Meadows was a pioneering contributor to systems thinking, a member of the Limits to Growth membership, who's life and career was cut tragically short by cerebral meningitis. Thinking in Systems is assembled from notes and lecturers, and represents a consistent whole anchored by the material in the essay Leverage Points: Places to Intervene in a System
Systems are modelled by stocks, flows, and sinks. Stocks are quantities that we care about, hopefully measurable, often not. Flows increase or decrease stocks, and can be fixed or vary based on natural, human, controlled, or uncontrolled processes. Sinks are much like stocks, but are assumed to be set to an infinity, an approximation that makes calculation easier. The basic dynamics of systems are feedback loops, delays, and oscillations. Negative feedback loops make the level of a stock trend towards a finite value. Positive feedback does the opposite, increasing the amount of a stock higher and higher until some countervailing feedback loop appears. And finally, due to innate delays, systems tend to oscillate rather than settling at a single configuration.
Where this gets tricky is first in mathematical modelling, because assigning limits to systems and making sure that the models match the real world is more of an art than a science ("All models are wrong, but some are useful" --George Box). And second, in convincing others of the somewhat counter-intuitive results of systems thinking. The one that jumped out to me was a simple inventory management model which entered into wildly uncontrolled oscillations as the decision to re-order became more responsive to inventory levels. Desired behave was achieved by basing re-order decisions on bi-weekly moving averages.
Meadows usefully provides the math at the end of the book. I feel like a contemporary update would include computer resources for experimentation and examples updated from the immediate post-Cold War where this book was written, but this is a useful book.
Systems are modelled by stocks, flows, and sinks. Stocks are quantities that we care about, hopefully measurable, often not. Flows increase or decrease stocks, and can be fixed or vary based on natural, human, controlled, or uncontrolled processes. Sinks are much like stocks, but are assumed to be set to an infinity, an approximation that makes calculation easier. The basic dynamics of systems are feedback loops, delays, and oscillations. Negative feedback loops make the level of a stock trend towards a finite value. Positive feedback does the opposite, increasing the amount of a stock higher and higher until some countervailing feedback loop appears. And finally, due to innate delays, systems tend to oscillate rather than settling at a single configuration.
Where this gets tricky is first in mathematical modelling, because assigning limits to systems and making sure that the models match the real world is more of an art than a science ("All models are wrong, but some are useful" --George Box). And second, in convincing others of the somewhat counter-intuitive results of systems thinking. The one that jumped out to me was a simple inventory management model which entered into wildly uncontrolled oscillations as the decision to re-order became more responsive to inventory levels. Desired behave was achieved by basing re-order decisions on bi-weekly moving averages.
Meadows usefully provides the math at the end of the book. I feel like a contemporary update would include computer resources for experimentation and examples updated from the immediate post-Cold War where this book was written, but this is a useful book.