Agile InsightsBriefing
A plain hardback with a yellow ribbon bookmark and a pair of reading glasses, an old enamel washbasin with a tap running in and water draining out, and a wooden lever on an iron fulcrum, on a dark workbench, labelled Making You Think 4, Stocks and Flows, and Leverage. A yellow bar along the bottom reads Making You Think, number 4.

Making You Think 4 · Primer and discussion · About 15 minutes in two parts

Thinking in Systems, by Donella Meadows

Meadows argues that a system's behaviour comes from its structure, so a problem that keeps coming back is rarely fixed by pushing harder on the obvious number. A primer on stocks, flows, loops, traps and leverage points, then a discussion of how well the idea holds up.

#leadership

Part 1: Primer · About 6 minutes

Download the MP3 · Linda and Leonard, AI voices, read a script written from the sources below.

Part 2: Discussion · About 8 minutes

Download the MP3 · Linda and Leonard, AI voices, read a script written from the sources below.

Cover of Thinking in Systems: A Primer

This episode is about Thinking in Systems: A Primer, by Donella H. Meadows. Find a copy on Amazon Australia, or ask your local bookshop or library.

In this episode

01Who Donella Meadows wasA chemist turned biophysicist who joined Jay Forrester's team at MIT, was lead author of The Limits to Growth in 1972, and taught at Dartmouth. The book was published in 2008, seven years after she died.
02Stocks, flows, loops and delaysA stock builds up and flows fill or drain it, so you can raise a stock by slowing the outflow. Balancing loops seek a target, reinforcing loops feed on themselves, and delays make both wobble.
03Traps and the twelve leverage pointsThe same structures produce the same trouble, such as drift to low performance or seeking the wrong goal. Of her twelve places to intervene, the numbers are weakest; goals, rules and information are stronger.
04Where the critics push backThe Limits to Growth was attacked hard, though later comparisons found the data tracking its standard run. Her leverage ranking has never been tested, and she called it tentative herself.
05A mirror, not a mapWhen a fix keeps failing, look at the structure, not the people. Check the outflow and the information before you reach for a new target, and use the ladder to see where you have been pushing.
A stock, its flows, and two loops
A stock with its inflow, outflow and two feedback loops A box labelled people in the team is filled by hiring and drained by people leaving. A balancing loop compares the team with its target and sets hiring, with a delay because hiring takes months. A reinforcing loop runs from fewer people to more overtime to more people leaving. PEOPLE IN THE TEAM STOCK Hiring inflow People leaving outflow DELAY: hiring takes months B Balancing loop the gap to the target headcount sets how hard we hire R Reinforcing loop fewer people, more overtime, more people leave Raise a stock by speeding the inflow or by slowing the outflow: the one we forget.

Our illustration, using the staffing example from the episode, not a diagram from the book. The pipes are flows, the box is the stock, B is a balancing loop with a delay on it, and R is a reinforcing loop.

The twelve leverage points, weakest at the bottom
Meadows' twelve leverage points, weakest to strongest A ladder of twelve rungs. From the bottom: numbers; buffers; stock-and-flow structure; delays; balancing loops; reinforcing loops; information flows; rules; self-organisation; goals; mindset or paradigm; transcending paradigms. The bottom three are grouped as parameters, then feedbacks, then design, and the top three as intent. Most of our attention goes to the bottom rung. STRONGEST, AND HARDEST TO CHANGE 1The power to transcend paradigms 2The mindset the system grows out of 3The goals of the system 4The power to change its own structure 5The rules: incentives, punishments, limits 6Information flows: who can see what 7How fast reinforcing loops feed on themselves 8How strong the balancing loops are 9How long the delays are 10The physical plumbing of stocks and flows 11Buffers: how big a stock is against its flows 12Numbers: targets, rates, budgets Intentwhy the system exists Designrules and information Feedbacksloops and delays Parametersthe things we count Her guess: 99 percent of our attention goes to the bottom rung. Order from Meadows' 1999 essay; she called it tentative. Grouping from Abson and others, 2017.

Meadows' list from her 1999 essay, in our plain words. The four groups are from Abson and others (2017), who built on her list; neither tested the order.

Your turn

Three questions to ask yourself

Think of a problem at work that has been fixed more than once and keeps coming back. What was changed each time: a number, or the goal, the rules, or who could see what was happening?
Look at the last three changes your team made to fix something. Were they numbers, rules, information, or goals?
Was any of those changes pushing in the wrong direction?

Two things to try, alone, with a pen

Ten minutesDraw the stockOne problem that keeps coming back, drawn as a stock with its inflow, its outflow, and the outflow you have been ignoring.
Ten minutesA leverage auditEvery intervention in one rollout you lived through, sorted into numbers, rules, information and goals.
Open the activity sheets (print or save as PDF)

About Donella H. Meadows

Donella "Dana" Meadows (1941 to 2001) was born in Elgin, Illinois. She took a degree in chemistry at Carleton College and a doctorate in biophysics at Harvard, then joined the team around Jay Forrester at MIT that built the World3 model. She was lead author of The Limits to Growth (1972), the report to the Club of Rome. She taught at Dartmouth from 1972 until her death, wrote the weekly newspaper column The Global Citizen from the mid-1980s, was a Pew Scholar (1991) and a MacArthur Fellow (1994), and founded the Sustainability Institute in 1996, now the Academy for Systems Change. She died of bacterial meningitis in February 2001, aged 59. Thinking in Systems, drafted in 1993, was edited by Diana Wright and published in 2008.

Other writing (6)
The Limits to Growth: The 30-Year Update (2004) · Chelsea Green
With Jørgen Randers and Dennis Meadows. The thirty-year revisit of the 1972 report and its scenarios.
Beyond the Limits (1992) · Chelsea Green
With Dennis Meadows and Jørgen Randers. The twenty-year revisit of The Limits to Growth.
The Electronic Oracle: Computer Models and Social Decisions (1985) · John Wiley & Sons
With J. M. Robinson. How computer models are used, and misused, in public decisions.
Groping in the Dark: The First Decade of Global Modelling (1982) · John Wiley & Sons
With John Richardson and Gerhart Bruckmann. What a decade of world models did and did not tell us.
Leverage Points: Places to Intervene in a System (1999) · Sustainability Institute
The essay that became chapter 6 of the book: twelve places to intervene, weakest to strongest. Free to read.
The Global Citizen columns (2001) · Donella Meadows Project
Her weekly newspaper column, archived online. A collection was published by Island Press in 1991.
Sources and links (13)
Thinking in Systems: A Primer · Chelsea Green
The publisher's page for the book (Chelsea Green, 2008), edited by Diana Wright.
Leverage Points: Places to Intervene in a System · Read at the Donella Meadows Project
Meadows' 1999 essay, which became chapter 6: the flip-chart story, the twelve points, the meter in the front hall, and her caveat that the list is tentative.
Donella H. Meadows, archive record · Dartmouth Library
Dartmouth Library's record of her life and papers: birth, education, Dartmouth years and awards.
MacArthur Fellow, class of 1994 · MacArthur Foundation
The MacArthur Foundation's page on her fellowship and work.
The Limits to Growth · Club of Rome
The Club of Rome's page for the 1972 report she led.
Lethal Model 2: The Limits to Growth Revisited · Read at Brookings
William Nordhaus, Brookings Papers on Economic Activity, 1992. Critical of the model's treatment of technological change, but calls dismissing the concern reckless.
Is Global Collapse Imminent? · University of Melbourne
Graham Turner, Melbourne Sustainable Society Institute, 2014. Compares the Limits scenarios with data to date.
Update to limits to growth · Wiley
Gaya Herrington, Journal of Industrial Ecology 25(3), 2021. Compares four World3 scenarios with newer data.
Leverage points for sustainability transformation · Read at PubMed Central
Abson and others, Ambio 46(1), 2017. Builds a research agenda on her list without testing its order.
Leverage Is Fractal, Relative… And What Else? · OCAD University
Ryan Murphy, RSD11, 2022. Argues the ranking has never been validated.
Book review: Thinking in Systems · drmaciver.com
David MacIver, 2018. A reader who liked it but found it less actionable than it claims.
AI Risk Management Framework 1.0 · NIST
NIST, 2023. Not about this book: describes AI systems as socio-technical.
Specification gaming: the flip side of AI ingenuity · Google DeepMind
Krakovna and others, Google DeepMind, 2020. Not about this book: systems that meet the literal objective and miss the intent.

Written from public sources and checked on 15 and 16 October 2026, and reviewed by a person who has read the book. Meadows' 1999 essay on leverage points, Nordhaus (1992), Turner (2014) and Abson and colleagues (2017) were read in full; Herrington (2021) and Murphy (2022) as abstracts; Nordhaus (1973) and Models of Doom as citations only. Sales figures for The Limits to Growth vary widely between sources, so none is given. None of the AI sources mentions Meadows: the parallels are the presenters' own. Check the book itself before quoting.

Our fact check · 21 claims: 16 confirmed, 3 reported, 1 disputed, 1 unverified

Before we record, every claim the hosts make is checked against the sources (this table was last updated 16 October 2026). Confirmed means a primary source or two independent reports; Reported means one party says so; Partly right, Disputed, Unverified and Not yet known mean what they say, and the hosts say so on air. This is the table as we left it.

The author

  1. ConfirmedBorn 13 Mar 1941, Elgin, Illinois (née Hager); died 20 Feb 2001, Hanover NH, aged 59, of bacterial meningitisDartmouth archives; The Dartmouth, 22 Feb 2001; Some obituaries say cerebral meningitis
  2. ConfirmedBA chemistry, Carleton 1963; PhD biophysics, Harvard 1968Dartmouth archives; MacArthur
  3. ConfirmedPart of the MIT team under Jay Forrester; lead author of The Limits to Growth (1972, for the Club of Rome)Club of Rome; Dartmouth; her 1999 essay; Say "MIT team", not "MIT faculty". Meadows Project eulogy page says 1971: wrong
  4. DisputedLimits sold "more than 9 million copies in 26 languages"Meadows Project; The Nation (2012) says 12m, 37 languages; Carleton says 90m (error). On air: "millions"
  5. ConfirmedTaught at Dartmouth 1972–2001; Pew Scholar 1991; MacArthur Fellow 1994Dartmouth; MacArthur
  6. Confirmed"The Global Citizen", weekly column from the mid-1980s to 2001, 20+ papers, 700+ columns, Pulitzer nomination 1991Meadows Project; The Dartmouth; MacArthur says bi-weekly; no readership figure exists
  7. ConfirmedFounded the Sustainability Institute 1996 (now the Academy for Systems Change); Cobb Hill cohousing, Hartland VT, was her visionMacArthur; Academy; Cobb Hill; Carleton says 1997
  8. ConfirmedCo-founded the Balaton Group, 1981MacArthur; Dartmouth; Academy says 1982

The book

  1. ConfirmedDrafted in 1993 and circulated informally; unfinished at her death; edited by Diana Wright; Chelsea Green, Dec 2008, 240 pp.; Earthscan, London, 2009Dartmouth archive item (Jan 1993 draft, comments by John Sterman); catalogues
  2. ConfirmedThree parts, seven chapters; chapter 6 is the leverage-points essayLibrary catalogues
  3. ReportedA system is elements, interconnections and a function or purpose; purpose is read from behaviourSecondary summaries; Check the printed page
  4. ConfirmedStocks, flows, balancing and reinforcing loops, delays1999 essay; summaries; In concept
  5. ReportedEight traps: policy resistance, tragedy of the commons, drift to low performance, escalation, success to the successful, shifting the burden to the intervenor, rule beating, seeking the wrong goalSecondary summaries (strong consensus)
  6. UnverifiedChapter 7 guidelines (15 in an earlier summary)an earlier summary; The essay "Dancing With Systems" has 14. Scripts give no count

Criticism and reappraisal

  1. ConfirmedModels of Doom (Sussex SPRU, 1973) critiqued Limits, with a reply from its authorsWorldCat; Bibliographic
  2. ConfirmedNordhaus 1973, "World Dynamics: Measurement Without Data"Economic Journal 83(332); Citation; Targets Forrester's World Dynamics, not Limits
  3. ConfirmedNordhaus 1992: both Limits models "rule out ongoing technological change"; but dismissing ecological concerns "would be reckless"Brookings Papers 1992(2); Read in full
  4. ConfirmedTurner 2008 and 2014: the data track the standard run; collapse onset in the model about 2015; Limits did not predict collapse by 2000Global Environmental Change; Univ. of Melbourne MSSI; 2014 paper internally reviewed
  5. ConfirmedHerrington 2021: best-fit scenarios show a halt in growth "over the next decade or so"; only one is a collapseJournal of Industrial Ecology 25(3); Abstract; Press overstated it
  6. ConfirmedThe leverage ranking has not been tested empiricallyMeadows herself; Abson et al., Ambio 2017; Murphy 2022; As critique
  7. ReportedBetter at diagnosis than prescriptionMacIver blog, 2018; Informal; Presented as the hosts' view

Want to know more?

Sign up for our podcast at agile-insights.com.au.

Read the transcript · Part 1: Primer

Linda Thinking in Systems says that a system's behaviour comes from its structure. So a problem that keeps coming back is rarely fixed by pushing harder on the obvious number. I'm Linda, and this is the fourth episode of Making You Think. The book is Thinking in Systems, A Primer, by Donella Meadows, published in two thousand and eight. By the end of this part you will know who she was, her argument in four steps, and the handful of terms that carry it.

Leonard And I'm Leonard. First, the author. Donella Meadows, known as Dana, was born in Illinois in nineteen forty one. She trained as a chemist, then took a doctorate in biophysics at Harvard. At M I T she joined the team around Jay Forrester, who founded a field called system dynamics. That team built a computer model of the world economy. In nineteen seventy two she was lead author of the report it produced, The Limits to Growth, which sold millions of copies. She taught at Dartmouth for nearly thirty years, wrote a weekly newspaper column... and died in two thousand and one, aged fifty nine.

Linda Which explains an odd thing about this book. It was published seven years after she died. She drafted it in the early nineteen nineties, it circulated among colleagues, and it was finished by her editor, Diana Wright. Her best known essay, on leverage points, became one of its chapters.

Leonard So, step one. What is a system?

Linda Meadows says a system is three things. There are elements, there are the connections between them, and there is a purpose. The elements are the easiest to see and matter least. You can change every player in a football team and it is still the same team. The purpose matters most, and it is the hardest to see. Her test is to read the purpose from what the system does, not from what it says. Our example is a help desk whose poster says quality first. If every reward goes to closing tickets fast, the system's purpose is closing tickets fast.

Leonard I have worked at that help desk.

Linda Most of us have.

Leonard Step two is the plumbing, and this is where her words start to matter. A stock is anything that builds up and can be counted at a moment. Water in a bath, money in an account, trust in a team. A flow is what changes the stock, in through the tap and out through the drain. Her point, and it is a good one, is that you can raise a stock by slowing the outflow as well as by speeding the inflow. If you are losing staff, hiring faster is one answer. Keeping the people you have is the one we forget.

Linda Then the loops. A balancing loop pulls a stock towards a target, like a thermostat bringing a room back to twenty degrees. A reinforcing loop feeds on itself, like interest compounding in an account, or a rumour spreading through an office. Add a delay, and the system wobbles. Picture a shower with a long pipe. You turn the tap, nothing happens, you turn it further... and then you are scalded. Most of the strange behaviour of organisations, she argues, comes from loops and delays that nobody drew.

Leonard Step three is that the same structures produce the same trouble, wherever they turn up. She calls these the system traps. Drift to low performance is a goal set by recent results, so every bad quarter lowers the bar for the next. Shifting the burden is a quick fix that weakens the system's own ability to cope. Think of overtime used to cover a staffing gap, until the team cannot run without it. And seeking the wrong goal is a system that delivers exactly what it is measured on. The way out, as her readers summarise it, is not to confuse effort with result.

Linda Step four is the part people remember. In her essay she tells of a meeting about the new world trade rules, where she went to a flip chart and listed the places to intervene in a system. There are twelve, from weakest to strongest. At the weak end are the numbers, the rates, the targets and the budgets. Further up are delays, feedback loops, and who gets what information. Then the rules, then the system's goals. At the very top is the mindset the whole system grows out of.

Leonard And the warning that comes with the list. In her words, leverage points are not intuitive. Most of our effort goes into the numbers, because the numbers are easy to change. She quotes Forrester, who found that people often know where the leverage is. As he put it, everyone is trying very hard to push it in the wrong direction. Think of a call centre with long queues that sets a shorter call-time target. The calls get shorter, more people ring back... and the queue grows.

Linda That one stings. Let me say the terms back. A system is elements, connections and a purpose, and you read the purpose from its behaviour. A stock builds up, and flows fill it or drain it. A balancing loop seeks a target, a reinforcing loop feeds on itself, and a delay makes both wobble. A trap is a structure that reliably produces trouble. A leverage point is a place to intervene, and the strong ones are the goals, the rules and the information, not the numbers.

Leonard So here is one question to take with you. Think of a problem at work that has been fixed more than once and keeps coming back. What was changed each time? Was it a number, or was it the goal, the rules, or who could see what was happening?

Linda We will not answer that for you. Give it a minute before you press play again. In Part Two we ask how well the idea holds up. Her most famous model drew fierce criticism, and her own list of leverage points has never been properly tested. We also look at what she says to anyone deploying AI, though she wrote before it. Join us there.

Read the transcript · Part 2: Discussion

Linda Thinking in Systems says that a system's behaviour comes from its structure, so a problem that keeps coming back is rarely fixed by pushing harder on the obvious number. I'm Linda, and this is Part Two of Making You Think, episode four, on the book by Donella Meadows. A quick recap. Stocks build up and flows fill or drain them. Loops either seek a target or feed on themselves. Leverage points are places to intervene, and the numbers are the weakest of them. By the end you will have the case for the idea, the case against, and some thinking to do.

Leonard And I'm Leonard. The case at its strongest first. As we read it, the book's gift is a shared language for problems that belong to nobody. When a team keeps missing the same target, people argue about who is to blame. Meadows moves the argument to the structure. Which stock is draining, which loop is missing, where is the delay? Her essay has a story she introduces with the words, the story goes. In a row of identical houses, some had the electricity meter in the front hall, where people could see it spin. Those houses used thirty percent less power.

Linda Thirty percent.

Leonard Just from moving the meter.

Linda And notice what changed there. Not the price, not a rule, not a target. Only who could see what, and when. That is leverage point six, the structure of information flows, and it is cheap. She makes the same point about our habits. Her guess is that ninety nine percent of our attention goes to the numbers, the weakest point on her list. Once you have that list in your head, it is hard to look at a new policy without asking where on the ladder it sits.

Leonard So here is the first activity, and it takes about ten minutes on paper. Pick one problem that keeps coming back. Draw it as a stock, a box with the thing that builds up. Draw an arrow in for what fills it and an arrow out for what drains it. Then add the outflow you have been ignoring. For staff turnover, it might be people leaving after their first year. For a backlog, it might be work that is reopened. Notice which arrow every past fix has pushed on.

Linda Now the pushback, and it starts with the most famous thing she wrote. The Limits to Growth, in nineteen seventy two, was attacked hard. A group at the University of Sussex published a book of critiques called Models of Doom, and the authors replied in it. The economist William Nordhaus had already gone after Forrester's earlier world model. His nineteen seventy three paper was titled Measurement Without Data. Twenty years later, in nineteen ninety two, he was still critical. He wrote that both versions of the model rule out ongoing technological change. But he also conceded that dismissing the ecological worry altogether would be reckless.

Leonard And the scoreboard has moved since. Graham Turner, then at C S I R O, compared the model with thirty years of real data in two thousand and eight. He found that the world was tracking the standard run fairly closely. In a twenty fourteen paper from the University of Melbourne, he noted that Limits never forecast collapse by the year two thousand, despite the popular story. In twenty twenty one, Gaya Herrington compared newer data with four of the scenarios. The two that fitted best both showed growth coming to a halt over the next decade or so. Only one of them was a collapse. Those are comparisons of curves, not proof the model was right about why.

Linda Which brings us to the criticism that matters most for this book. Her ranking of twelve leverage points is a thoughtful list, not a tested finding. She said so herself. Near the end of the essay she calls it tentative, and says its order is slithery. In twenty seventeen, a group led by David Abson used the list to frame research on sustainability, but they did not test the order. A twenty twenty two paper by Ryan Murphy argues that the ranking has never been validated... and that leverage depends on where you stand.

Leonard I want to push on that, because I think it matters more than the Limits fight. People quote the ladder as if it were a law. Goals beat rules, rules beat numbers. But sometimes the number is the leverage. A change to a single interest rate can move a whole economy. One reader, David MacIver, writing on his blog in twenty eighteen, liked the book but found it less useful in practice than it claims to be. That is one review, not a verdict. Still, I suspect many people finish it able to describe their problem beautifully...

Linda I partly disagree.

Leonard ...and no closer to fixing it.

Linda The ladder is not a recipe, and she never sold it as one. Its use is as a check on your own habits. If every fix your organisation has tried sits at the bottom, at the numbers, that tells you something. It does not tell you which goal to change. But it stops you pretending that a new target is a new strategy.

Leonard That I will accept. Then we can settle it this way. The book is better at diagnosis than at prescription, and it is honest about that. Use it to see where you have been pushing, not to choose your next move for you. That is our verdict, and you are welcome to argue with it. A question for you, then. Look at the last three changes your team made to fix something. Were they numbers, rules, information, or goals? And was any of them pushing in the wrong direction?

Linda Now, AI, and carefully. Meadows died in two thousand and one and wrote nothing about the kind of AI we use now. But she did write two books about computer models and how decision makers use them, in nineteen eighty two and nineteen eighty five. The second is called The Electronic Oracle. So the worry about trusting a machine's picture of the world was hers. The parallels that follow are ours. America's standards body, N I S T, calls AI systems socio technical in its risk framework. It does not cite her, but the idea is close.

Leonard Here is the parallel that struck us. Researchers at DeepMind describe something they call specification gaming. A system meets the literal objective it was given and misses what its designers intended. That sounds a lot like her trap of seeking the wrong goal. No one we found has linked the two in print, so treat it as our observation. And it applies to the people as much as the machine. An organisation that measures AI adoption by licences issued is seeking the wrong goal of its own.

Linda We have all seen that dashboard.

Leonard We have.

Linda The second activity is a leverage audit, and it follows from that. Take one technology rollout you have lived through, AI or not. Write down every intervention that was made: the training, the targets, the rules, the dashboards. Next to each one, write whether it changed a number, a rule, the information people had, or the goal. Then notice where they cluster, and write one intervention higher up the list that nobody tried.

Leonard So, what to carry away. One, when a fix keeps failing, look at the structure, not the people. Two, check the outflow and the information before you reach for a new target. Three, read her ladder as a mirror, not a map. It shows you where you have been pushing. It does not promise where to push next.

Linda If you'd like to know more, sign up for our podcast at agile-insights.com.au.

Leonard Thanks for listening.

More on #leadership

Episode 38 · Primer and discussionBudgeted for seats, billed for workEpisode 34 · Primer and discussionWhen the test stops telling you anything: hunger beats tenureEpisode 23 · Primer and discussionWhat FDE means at OpenAI: skills, not custom code.Episode 21 · DiscussionTrust cannot be rushed: how much authority will you give AI?Making You Think 13 · Primer and discussionHow to Lead a Quest, by Jason FoxMaking You Think 12 · Primer and discussionThe Game Changer, by Jason Fox

Every episode tagged #leadership

All of Making You Think (13) · Follow Making You Think in Apple Podcasts or any podcast app

Everything from Agile Insights: briefings, Making You Think and the reference series