Five lessons in, you can now do the thing this whole course was for. You know what a leverage point is — a place where a small, well-aimed push produces a large, lasting change in how a system behaves. You’ve felt, on the interactive ladder, that location beats effort. You’ve toured every rung, from the shallow end (numbers, buffers, structure) up through the powerful middle (loops, information, rules) to the summit (goals and paradigms). And you’ve learned the counterintuition at the heart of it: the levers that matter most are the ones our instincts don’t reach for, because they’re hard, resisted, and a little frightening.
What’s left is to fuse all of that into something you can actually do on a Tuesday, standing in front of a system that won’t behave. Knowing the rungs is like knowing the notes; this lesson is about playing a tune. We’ll turn the course into a three-step procedure, walk it up the ladder on one stubborn real system watching the same problem yield harder and harder, and then — because a model that won’t show you its own failure modes is a model you don’t understand — name the specific ways this one lies to you up high. Let’s climb.
Before you read — take a guess
A regional aquifer is draining fast — farmers pump more each year than the rain refills. The water agency triples the price of the electricity used to pump. Two years later, the water table is still dropping. From a leverage-point view, what's the best next move?
Step 1 — Find the loops
You can’t push a system intelligently until you can see it, and “seeing” a system means one specific thing: mapping the stocks it accumulates and the feedback loops that drive its behaviour. This is the skill the prerequisite feedback-loops course handed you, and it’s step one because everything above it depends on it. Skip it and you’re not climbing a ladder — you’re flailing at whatever intervention happened to come up in the meeting.
Concretely, for the misbehaviour you want to change, ask: What is the stock at the centre of it — the thing filling or draining (the water table, the traffic on the road, the hours lost to meetings, the trust in a wiki)? What reinforcing loops amplify it (the ones that turn a small trend into a runaway)? What balancing loops are supposed to hold it in check (and are they even present)? Where are the delays that make the whole thing overshoot and oscillate?
A leverage point is never “the price” or “the rule” in the abstract — it’s a specific spot in a specific loop. If you can’t draw the loop, you can’t find the point. So the first move is always to draw it.
The one habit that does most of the work
Before you propose a single fix, sketch the loop. Name the stock, trace the reinforcing loop that’s running away, and look hard for the balancing loop that should be pulling it back — because the most common disease in a misbehaving system is a missing balancing loop, and you can only spot a hole by drawing the shape it’s missing from.
Step 2 — Locate the misbehaviour’s source
Once the loops are on paper, you diagnose. A system doesn’t misbehave in infinite ways — it misbehaves in a small handful, and each one points to a different rung of the ladder. This is the step that turns “something’s wrong” into “this specific thing is wrong,” which is the difference between a prescription and a guess.
| Symptom you see | Likely cause in the loops | Where the leverage lives |
|---|---|---|
| A trend runs away, faster and faster | A reinforcing loop with no brake | Weaken the reinforcing loop, or add/strengthen a balancing one |
| The system won’t settle — swings above and below target | A delay in a balancing loop makes it overshoot | Shorten the delay so the system sees results sooner |
| Actors keep making the “wrong” choice | Feedback is simply missing — they can’t see the consequence | Expose the information; close the missing loop |
| It stabilises, reliably, at a bad place | A balancing loop aimed at the wrong target | Change the goal the loop steers toward |
Notice that the same visible complaint (“the aquifer keeps draining”) can have more than one of these underneath it — a runaway pumping loop and missing feedback and a goal aimed at the wrong thing all at once. That’s normal. The diagnosis tells you which rungs are even in play; step three is about choosing how high to push among them.
Match each misbehaviour to the loop-level diagnosis it points to.
Place each item in the right group.
- A widen-the-road → more driving → wider-road spiral that never ends the jam
- Every farmer pumps blind and can’t see the basin’s decline until it’s a crisis
- Support staff never learn which tickets they resolved badly — no one tells them
- A team ships more and more features, each one adding maintenance that slows the next
- The calendar reliably fills to 100% utilisation — which is exactly what the org rewards
- Hiring lurches from frantic over-hiring to sudden freezes and back
Step 3 — Climb the ladder (don’t grab the nearest number)
Here’s the whole discipline in one instruction: don’t push where your instinct first lands. Your instinct lands on a number, every time, because a number is the thing that’s easy to argue about and safe to change. Instead, having found the loops and located the misbehaviour, walk up the rungs and ask, in order of rising leverage:
- Can I fix the loop itself — weaken a runaway, strengthen a stabiliser, shorten a delay?
- Can I expose the missing information — close the feedback loop the actors can’t see?
- Can I change the rule or incentive — re-aim what the whole system rewards?
- Can I change the goal — the target the entire machine reorganises itself to chase?
Push as high as you can actually reach. The bias to build is upward: information flows, rules, and goals beat numbers nearly every time, and information flows in particular are the bargain of the ladder — high-leverage and cheap. You won’t always be able to reach the top rung (changing a company’s goal is a year-long brawl, not an afternoon), but the point is to aim high and settle downward only when you must — never to stop at the first number because it was the first thing you thought of.
Fill in the procedure.
Pick the right option for each blank, then check.
To change a system: first driving the behaviour, then locate the in those loops, then climb — preferring over the nearest number, and pushing as high as you can actually .
The worked climb: one draining aquifer, six rungs
Enough abstraction. Let’s take the aquifer from the pretest and walk the same problem up every rung, watching it yield harder and harder. This is the pattern to internalise: a stubborn system doesn’t have one “solution” — it has a ladder of interventions, and the higher you push, the more the thing actually moves.
The system: a shared groundwater basin. The stock is the water table. The runaway is a reinforcing loop — as the table drops, each farmer, fearing they’ll be left with nothing, pumps harder to grab their share before it’s gone (a classic tragedy of the commons). There is no balancing loop pulling draw back toward recharge, and crucially, no feedback: every farmer sees only their own field, never the basin. The goal the whole system optimises for is “maximise acres farmed this season.”
Leverage points
Walk the aquifer up the ladder
Pick a place to intervene, then push. The effort bar is the same wherever you push — but watch how differently the system actually moves. Low rungs barely budge it however hard you shove; a light touch up high moves it a lot.
Where you push
You spent 100% effort at "Numbers" → the system moved 16%: a low-leverage point — easy to push, but the loops, information, and goal are untouched, so the system just absorbs the nudge.
Now the same climb in a table — the intervention at each rung, and honestly how much it moves the water table:
| Rung | The intervention on the aquifer | What actually happens |
|---|---|---|
| 1. Number | Triple the price of pumping electricity | Farmers grumble, pay more, keep pumping — the panic to grab your share outweighs a bill. The table keeps dropping. Barely helps. |
| 2. Buffer / structure | Build a reservoir to bank surface water in wet years | A useful cushion that buys time — but it doesn’t touch why they overdraw. The basin still drains; you’ve just padded the fall. Modest. |
| 3. Loop fix | Meter every well and cap total draw to the basin’s recharge rate — a real balancing loop | Now there’s a brake where there was none: draw is pulled back toward what the basin can refill. The runaway finally has a governor. Big change. |
| 4. Information flow | Show each farmer, live, their own draw and their neighbours’ and the basin’s recharge | The missing feedback closes. Overdraw becomes visible and social; the tragedy-of-the-commons blindness lifts. Often cuts draw sharply — for almost no money. |
| 5. Rule / incentive | Issue tradable water rights capped at sustainable yield — pump less, sell the surplus | Now the incentive rewards conservation instead of punishing it. Every actor re-aims toward using less. Very high leverage. |
| 6. Goal / paradigm | Change the system’s goal from “maximise acres farmed” to “a basin that refills every year” | The whole machine re-points. Crop choices, rights, metering, and reservoirs all now serve sustainability. Every lower rung falls into place beneath it. Highest — and brutally hard. |
Read the table top to bottom and the shape of the course jumps out. The number barely twitches the table. The loop fix installs the brake the system was missing. The information flow — the cheapest move on the whole list — closes the blindness that caused the tragedy in the first place. And the goal change reorganises everything beneath it, which is why it’s simultaneously the most powerful and the one that takes a decade of political fights to land.
The same walk works on whatever stubborn system you’ve got. Swap the aquifer for a company drowning in meetings and the rungs still line up: a number (cap meetings at 30 minutes) barely helps; a structure tweak (no-meeting Wednesdays) is modest; a loop fix (any attendee can kill a meeting that’s lost its purpose) adds a real brake; an information flow (show the true dollar cost of every meeting on the invite — twelve people times an hour is not free) closes the missing feedback; a rule (every team gets a fixed weekly meeting budget they must spend wisely) re-aims the incentive; and the goal change (from “maximise alignment and visibility” to “protect deep-focus time”) re-points the whole culture. Different system, identical climb.
Match each aquifer intervention to the rung of the leverage ladder it sits on.
Pick a term, then click its definition.
Pitfall 1 — Right point, wrong direction
Now the failure modes, because finding a leverage point is only half the job — pushing it the right way is the other half, and it’s the half that separates a dangerous systems-thinker from a useful one. This is Meadows’ most unsettling observation, worth quoting nearly verbatim: high leverage points “are not intuitive… and when they are found, they are almost always pushed in the wrong direction.” People’s instinct for where is often sound. Their instinct for which way is backwards.
The classic shape: a system is in trouble because of a runaway reinforcing loop — an economy overshooting its resource limits, a basin draining, a company burning cash to chase growth. Someone correctly identifies the growth loop as the high-leverage spot. And then, because their whole worldview says growth is good, they push it harder — more stimulus, more subsidies, “we’ll grow our way out.” They found the fire and fed it. The right-direction move was to weaken the reinforcing loop or strengthen a balancing one, not to crank the runaway. Correct spot, catastrophic direction.
The tell of a backwards push
Before you push a high rung, run the direction check. Are you weakening a runaway loop or feeding it? Closing a missing feedback link or burying the signal in noise? Steering toward the right goal or doubling down on the wrong one? The wrong direction on a strong lever does more damage than any weak lever ever could — a botched push on a powerful point isn’t a small mistake, it’s a big one aimed precisely.
Pitfall 2 — Mistaking activity for progress
The second trap is quieter and more seductive: furious activity on a low-leverage number feels like doing something. You call the meeting, draft the memo, argue the tax rate up a point, ship the dashboard — motion, effort, visible busyness, the comforting sense of action. And the system doesn’t move, because you were the child bouncing next to the pivot: working hard, lifting nothing.
Motion is not leverage. The number everyone is fighting over is, almost by definition, a weak one — the fight is there precisely because it’s the safe, easy dial that disturbs nothing structural. So the tell is uncomfortable: if your intervention was easy to agree on, met little resistance, and generated a lot of activity, be suspicious of it. Real leverage usually looks like a small, hard, resisted move that changes a loop or a goal — and produces far less busywork than the theatre of pushing a number back and forth.
A city is choking on traffic. The transport department launches a high-profile campaign: new signage, a public dashboard of daily car counts, three council votes to nudge the parking fee, and a press tour — months of visible, energetic work. Traffic is unchanged. What's the leverage diagnosis?
Pitfall 3 — Wrong information isn’t more information
“Add information flows” is one of the highest-leverage, cheapest moves on the ladder — but only if it’s the right signal, at the right place, closing the specific missing loop. This is where the rung gets abused. Someone hears “information is high-leverage” and builds a dashboard of forty metrics, most of them vanity numbers, and buries the one signal that actually mattered under a pile of noise. That’s not a high-leverage information flow; it’s a high-leverage pipe pumping garbage, and it can be worse than no dashboard at all, because now the real signal is hidden and everyone feels informed.
The aquifer meter works not because it shows farmers more data but because it shows them the one thing they couldn’t see and needed to: their draw against the basin’s recharge, right where their decision is made. That’s the whole art of the information rung — not volume, but closing the exact feedback loop that was open. More information is not the goal. Closing the specific missing loop is.
Pitfall 4 — Big levers cut both ways
The deepest pitfall, and the reason humility isn’t optional up high: the same spot where a wise push transforms a system is exactly where a careless push wrecks it. High leverage is powerful in both directions. Low-leverage numbers are forgiving — set a tax rate slightly wrong and you retune it next year, no harm done. Get a goal or a core loop wrong and you’ve reorganised the entire system around a mistake, at enormous cost to undo, because everything downstream has already re-pointed toward the bad target.
So the higher you climb, the more the move rewards care. Pilot high-rung changes somewhere reversible — one region, one team, one season — before you rewire the whole thing on a hunch. And hold onto a genuinely liberating fact: a clean low-leverage fix can beat a botched high-leverage one. A well-set number that everyone actually implements outperforms a grand goal change that nobody buys into and that quietly breaks three things you didn’t see coming. Aim high, but push gently, watch the loops, and be ready to back out.
No — aim high, but handle high levers with care. The ladder tells you where the most leverage is, not where it’s safe to flail. Three caveats to carry:
- A clean low-leverage fix beats a botched high-leverage one. A well-set, well-implemented number can outperform a half-baked goal change that no one commits to.
- Resistance signals leverage and warrants listening. Sometimes the system resists a high-rung change because it’s protecting something you haven’t seen yet. Push, but stay curious about the pushback.
- Because big levers move in both directions, test where you can reverse the result. Pilot the new incentive in one region; don’t re-goal the whole company on a hunch and discover the mistake after everything’s reorganised around it.
The skill is to reach as high as you can — then push gently, watch the loops, and keep a hand on the exit.
Where this sits in the wider latticework
A leverage point isn’t a lonely idea — it’s one face of a pattern that shows up all over systems thinking: in a connected system, influence is wildly unevenly distributed, and the whole game is finding the load-bearing node before you act. The same insight wears different names in different courses:
| Model (course) | The shared insight |
|---|---|
| Bottleneck (constraints) | The single tightest constraint governs the whole system’s output — relieve it and everything speeds up; work anywhere else and nothing does. That’s a leverage point wearing a hard hat. |
| Keystone species (ecosystems) | One node holds up far more than its weight; remove it and the web cascades. High leverage, biological edition. |
| Critical mass (tipping points) | A threshold where a small push tips the entire system over — leverage concentrated at a single point on a curve. |
| Second-order thinking | A leverage point is precisely where the biggest downstream cascade begins — the “and then what?” chain, run in reverse: not where does a push end up, but where do I push so the cascade goes where I want. |
Line them up and they’re the same theorem restated: systems aren’t uniform, their sensitivity is lumpy, and a handful of spots carry most of the weight. Find the load-bearing node — the bottleneck, the keystone, the tipping threshold, the leverage point — before you spend your force. Push everywhere and you’re doing the child-bouncing-by-the-pivot thing at civilisational scale.
Recap
Big picture
Leverage Points — the whole course in one picture
- Leverage Points
- What a leverage point is
- A place where a small, well-aimed push → big, lasting behaviour change (a lever)
- Location beats effort — where you push matters far more than how hard
- The ladder (weak → strong)
- 1. Numbers / parameters — easy, LOW leverage
- 2. Buffers & stocks — modest
- 3. Loop structure — rewire / strengthen / weaken loops (HIGH)
- 4. Information flows — who sees what, how fast (HIGH + cheap)
- 5. Rules & incentives — change what’s rewarded (VERY high)
- 6. Goals & paradigms — change the target / mindset (HIGHEST)
- Why obvious = weak
- A number is easy to change BECAUSE it disturbs nothing structural
- Goals & loops resist BECAUSE they change everything
- Resistance ≈ a thermometer for real leverage
- The procedure
- 1. Find the loops — map stocks & feedback
- 2. Locate the misbehaviour — runaway? weak brake? delay? missing feedback?
- 3. Climb — fix the loop → expose info → change the rule → change the goal
- Bias upward; reach as high as you can actually move
- The pitfalls
- Right point, wrong direction (feeding a runaway instead of weakening it)
- Activity mistaken for progress (motion ≠ leverage)
- Wrong information ≠ more information (close the RIGHT loop)
- Big levers cut both ways — pilot reversibly, handle with humility
- Latticework ties
- Bottleneck — the constraint governs the whole system’s output
- Keystone species — a node holding up more than its weight
- Critical mass — a threshold where a small push tips everything
- Second-order thinking — where the biggest downstream cascade begins
- What a leverage point is
Check yourself: the whole course
What is the correct first step of the leverage procedure, before you propose any intervention?
Check your answer to continue.
Key takeaways
Turn the whole course into one procedure: find the loops (map the stocks and feedback driving the behaviour), locate the misbehaviour (a runaway reinforcing loop? a delay causing oscillation? a weak or wrongly-aimed balancing loop? feedback simply missing?), then climb the ladder — fix the loop, expose the missing information, change the rule, change the goal — pushing as high as you can actually reach, because the bias to build is upward and the nearest number is nearly always the weakest spot. The worked aquifer shows the payoff: the same draining basin barely moves for a price hike, gains a real brake from a metered cap, and transforms when you close the farmers’ missing feedback or re-aim the goal from “maximise acres” to “a basin that refills.” Then respect the four ways this model lies to you up high: right point, wrong direction (feeding a runaway you should weaken); activity mistaken for progress (loud busywork on a number that moves nothing); wrong information isn’t more information (close the right loop, don’t bury the signal); and big levers cut both ways (a botched goal change reorganises everything around a mistake — so pilot reversibly, handle high rungs with humility, and remember a clean low-leverage fix can beat a botched high-leverage one). And zoom out: a leverage point is the same insight as a bottleneck, a keystone species, critical mass, and second-order thinking — in a connected system, influence is lumpy, so find the load-bearing node before you push.
Where this goes next
That’s every teaching lesson in the course. You started at the counterintuition — that location beats effort and the obvious dial is the weak one — toured the shallow end and the powerful middle and the summit, and now you can take any misbehaving system and run the whole procedure on it: find the loops, locate the misbehaviour, climb the ladder, and check your direction before you push.
What’s left is to prove it. The Final Exam is a graded, one-question-at-a-time run across everything in this course — what a leverage point is, the full ladder from numbers to paradigms, why the obvious lever is weak, the find-the-loops-then-climb procedure, and every pitfall up high. It’s a one-way door by design: each answer locks the moment you submit it — no going back, no retries, no restart — and you’ll see your score only at the very end. You need 70% to pass. Take your time on each question, because once you commit, you commit. Go find the load-bearing node.