Leverage Points: Where to Push a System

Four lessons in, you can now read a system. You can spot the stocks that fill and drain like bathtubs, the reinforcing loops that turn whispers into screeches, the balancing loops that hold a room at 21°C, and the delays that make a stabiliser overshoot and ring. That’s diagnosis. This lesson is the prescription: given that you can see the loops, where do you push to actually change what the system does?

The bad news, and also the entire point of this lesson, is that human instinct picks the wrong place almost every time. We crowd around the one dial that’s easy to argue about — the tax rate, the budget number, the speed limit — and we shove it back and forth, and the system barely twitches. Meanwhile the spot that would actually move it sits untouched, because touching it is hard and scary and changes everything. This is the systems-thinker’s last and most useful skill: knowing that the obvious lever is the weak one, and where the strong ones hide.

The canonical map here comes from Donella Meadows, a systems scientist who wrote a famous essay called “Leverage Points: Places to Intervene in a System.” She listed twelve of them, ranked. We’re going to compress her twelve into a handful of rungs you can actually carry around, because a ladder you remember beats a list you don’t.

What a leverage point is

A leverage point is a place in a system where a small, well-aimed push produces a large, lasting change in the system’s behaviour. Not a one-time bump — a change in how the whole thing runs from then on. The name is literal: it’s a lever. A child on the long end of a seesaw can lift an adult; the same child pushing directly on the adult’s chest accomplishes nothing. Same force, wildly different result, entirely because of where it’s applied.

The other useful image is the acupuncture needle: a thin pin placed at exactly the right point is supposed to ripple through the whole body, while the same pin jabbed anywhere else is just a pinprick. Whether you buy the medicine or not, the geometry is the systems-thinking truth — leverage is about location, not effort. You don’t change a system by pushing harder. You change it by pushing in the right place.

Here’s Meadows’ genuinely unsettling insight, the one that makes this more than a tidy list. She noticed that people are actually very good at finding leverage points — managers, politicians, and engineers reliably zero in on exactly the spots that matter. And then they push them in the wrong direction. They find the loop that’s driving the runaway and feed it more. They find the goal and double down on the wrong goal. The instinct for where is sound; the instinct for which way is backwards. So this lesson is half “find the strong levers” and half “and for heaven’s sake, push them the right way.”

When to use it

Reach for the leverage-point lens whenever you’ve finished diagnosing a system (you can name its stocks and loops) and you’re about to intervene. The question shifts from “what is this system doing?” to “where do I push?” — and the leverage ladder is the map of where pushes are cheap-and-weak versus hard-and-transformative.

The ladder of leverage

Meadows’ twelve points, boiled down, sort into a handful of rungs. The rule that runs through all of them: the higher the rung, the more leverage — and the more it costs, in effort and in resistance, to move it. Numbers are easy to change and barely matter. Paradigms are nearly impossible to change and matter most. Here’s the compressed ladder, weakest at the bottom:

The order isn’t perfectly strict — a clever information fix can out-perform a clumsy rule change — but the direction is reliable: as you climb, leverage rises and so does difficulty. Now let’s make at least the load-bearing rungs concrete.

Rung 1 — Numbers. A city sets the thermostat in every public building to 20°C instead of 21°C. Real, measurable, easy to legislate. Also tiny: it doesn’t change how the heating system works, only the dial it’s aiming at. Numbers are the parameters of a system — and parameters rarely change behaviour, they just retune it slightly. This is the rung everyone fights over precisely because it’s the one that’s safe to fight over.

Rung 3 — Loop structure. A factory keeps over- and under-producing because it has no fast read on actual demand. Adding a balancing loop — a weekly signal that adjusts production to real orders — changes the system’s behaviour, not just its settings. This is exactly lessons 2 through 4 cashed out: you strengthen a balancing loop to add stability, interrupt a reinforcing loop to stop a runaway, or shorten a delay to stop oscillation. Rewiring a loop is in a different league from nudging a number, because the loop is what generates the numbers.

Rung 4 — Information flows. The most famous result in this whole field: give households a meter that shows their electricity use as it happens — and consumption drops, often by 5 to 10%, with no new tax, no new rule, no new technology. Nothing changed except who could see what, and when. Information flows are absurdly high-leverage relative to their cost, because most systems misbehave not from bad intentions but from missing feedback — the actor literally cannot see the consequence of their action, so the loop that should self-correct never closes. Show people the consequence and the loop wakes up.

Rung 5 — Rules & incentives. This is your old friend from second-order thinking. Change what the system rewards and you change everything downstream, because people (and firms, and algorithms) follow incentives the way water follows gravity. Tie a sales bonus to “new accounts opened” and you get fake accounts (Goodhart’s Law, the Wells Fargo special); tie it to “verified satisfied customers” and you get something else entirely. The rules are the loop structure written in law.

Rung 6 — Goals & paradigms. The very top. Change the goal the system is optimising for, and the whole machine reorganises itself to chase the new target. A road network optimised for “vehicle throughput” will widen lanes forever and stay jammed (induced demand — a reinforcing loop). The same network optimised for “people’s access to where they need to go” suddenly finds buses, bikes, and zoning on the table. Same city, same budget, radically different behaviour — because the number the system is steering toward changed. A paradigm is the rung above even that: the shared, usually-unspoken belief that makes a goal seem obvious (“more driving = more freedom”). Shift the paradigm and the goals shift under it without a fight.

When to use it

Use the ladder as a checklist that climbs. When you want to change a system, don’t stop at the first intervention you think of (it’ll almost always be a number). Walk up the rungs: “Could I fix the loop instead? Could I just expose the missing information? Could I change the incentive? Could I change the goal?” The interesting answer is usually two or three rungs above where your instinct stopped.

Why the obvious lever is the weak one

Here’s the paradox worth sitting with: the leverage points that are easiest to push are weak precisely because they’re easy. A number is easy to change exactly because changing it doesn’t disturb anything structural — you can move a tax rate or a setpoint with a single vote and zero rewiring. That same property is why it barely moves the system: nothing about the loops, the information, or the goal has changed, so the system absorbs your nudge and carries on. Easy-to-move and low-leverage are the same fact seen from two sides.

A goal or a loop, conversely, is hard to change for the identical reason it’s powerful: changing it disturbs everything. Rewire a loop and every flow downstream behaves differently; change the goal and every decision in the system re-points. Of course that’s hard — you’re not turning a dial, you’re moving the thing the dials are attached to. People and institutions resist it ferociously, which is itself evidence you’ve found a real leverage point. The amount of resistance is a rough thermometer for how much leverage you’ve actually found.

Watch the contrast play out on the traffic problem:

The toll is one rung; the goal is the top rung. The toll is a Tuesday-afternoon council vote; the goal change is a decade-long political brawl. That difficulty gap is not a coincidence — it’s the signature of the leverage difference.

When to use it

Whenever you catch a group (or yourself) locked in a heated argument over a single number, step back and ask: “Is this number even where the leverage is, or are we fighting here because it’s the only spot that’s easy to touch?” Often the real lever — a loop, an information gap, the goal — isn’t on the agenda at all, precisely because moving it is hard.

The practical move: find the loops, then climb

Here’s the whole course turned into a procedure. You can now do step one, which most people can’t:

1. Find the loops. Map the stocks and the feedback loops actually driving the behaviour you want to change. (Lessons 1–4. This is the part that took you four lessons to learn.)
2. Locate the misbehaviour’s source. Is a reinforcing loop running away? Is a balancing loop too weak, or aimed at the wrong target? Is a delay causing oscillation? Is feedback simply missing?
3. Climb the ladder. Don’t grab the nearest number. Ask, in order of rising leverage: Can I fix the loop? Expose the missing information? Change the rule/incentive? Change the goal? 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 whole ladder, because they’re high-leverage and cheap. Sort your candidate interventions before you commit to one:

Tying the whole course together

This lesson is where the four before it pay off. Each kind of structure you learned to recognise comes with its own high-leverage move:

Notice that the highest-leverage move on a balancing loop isn’t strengthening it — it’s changing the goal it aims at. A thermostat set to 30°C will faithfully, stably cook you; making it a better thermostat doesn’t help. The leverage is in the setpoint, which is to say the goal, which is to say the top of the ladder. Strengthening a loop pointed at the wrong target just gets you to the wrong place more reliably.

And this connects straight back to second-order thinking. A leverage point is, almost by definition, the place where a single push produces the biggest cascade of downstream effects — which is exactly the “and then what?” chain, run in reverse: instead of asking where a push ends up, you ask where to push so the cascade goes where you want. The incentives engine you learned there is just rung 5 of this ladder: rules are high leverage because they re-aim every actor at once. The two courses are the same machine described from two ends.

When to use it

When you’ve diagnosed which structure is misbehaving, jump straight to its matching move: harness or interrupt a reinforcing loop, strengthen-or-re-goal a balancing loop, shorten a delay. The diagnosis you learned in lessons 2–4 hands you the prescription directly.

The pitfall: right point, wrong direction

Remember Meadows’ unsettling observation — people find the leverage points and then push them the wrong way. This is the failure mode that separates a dangerous systems-thinker from a useful one, so let’s name its shapes.

Pushing a strong lever backwards. A system that’s already overshooting its limits — too much growth, too much extraction — is in trouble because of a runaway reinforcing loop. The instinctive fix from people who love growth is to add more growth (subsidies, stimulus, “we’ll grow our way out”). They’ve correctly identified the high-leverage loop and then fed the fire. The right direction was to weaken the reinforcing loop or strengthen a balancing one, not to crank the runaway harder.

The wrong information. “Add information flows” is high-leverage — but only if it’s the right information arriving at the right place. Flood people with the wrong metric and you’ve built a high-leverage pipe pumping noise. A dashboard of forty vanity numbers can be worse than no dashboard, because it buries the one signal that mattered. More information isn’t the goal; closing the specific missing feedback loop is.

And the meta-pitfall: high-leverage points are powerful in both directions. The same spot where a wise push transforms a system for the better is exactly where a careless push wrecks it. Low-leverage numbers are forgiving — get a tax rate slightly wrong and you can retune it next year. Get a goal or a core loop wrong and you’ve reorganised the entire system around a mistake, at enormous cost to undo. The higher you climb the ladder, the more the move rewards humility, slow testing, and reversibility. Big levers are not toys.

When to use it

Run the direction check every single time you push a high rung. The higher the leverage, the more this matters: confirm you’re weakening the runaway (not stoking it), surfacing the right feedback (not just more of it), and steering toward the right goal — and pilot it somewhere reversible first.

Recap

Where this goes next

That’s every teaching lesson in the course. You started with stocks and flows, learned to read reinforcing and balancing loops, watched delays turn stability into oscillation — and now you know where to push a system once you’ve read it, and why the obvious dial is the weak one.

What’s left is to prove it. The Final Exam is a graded, one-question-at-a-time run across everything in this course — stocks and flows, both loop types, delays and oscillation, and leverage points. 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 end. You need 70% to pass. Take your time on each question, because once you commit, you commit. Go close the loop.