If systems resist, where can anyone push? In 1999 Donella Meadows published an essay called "Leverage Points: Places to Intervene in a System." It ranked twelve kinds of change, from the least powerful to the most. Meadows offered it with humility, as a map rather than a law.

Numbers at the bottom

Near the bottom of her list are parameters: the numbers in a system, such as tax rates, subsidies, speed limits and minimum wages. These are what most political debate is about. Yet Meadows argued that numbers such as taxes, subsidies and standards are usually the weakest leverage point, even though they get the most attention. Changing a parameter rarely changes the structure that produces the behavior. Turning a tax up or down a point moves things a little; the system goes on working as before.

Structure in the middle

Higher up are the physical and feedback features covered in earlier levels: the size of buffers, the layout of stocks and flows, the length of delays, and the strength of balancing and reinforcing loops. These are more powerful but often slow or expensive to change.

Above them sits a point that is cheap and surprisingly strong: information flows. Who knows what, and when? Making information visible to the people affected can change behavior even without new limits. Meadows's favorite case was the Toxics Release Inventory, created by a 1986 United States law. It did not cap pollution. It simply required factories to report their toxic emissions publicly each year. Once local newspapers began printing lists of the worst polluters in town, reported emissions fell substantially within a few years.

Rules, goals and paradigms

Higher again are the rules of the system: incentives, punishments and who has the right to decide. Above the rules is the power to change the system's structure itself, which Meadows called self-organization, the ability to learn, diversify and evolve. Above that is the goal of the system. A corporation whose goal shifts from serving customers to maximizing its share price will change in every part, even if no rule changes.

At the top she placed the paradigm. The paradigm is the shared mindset out of which the whole system arises, and Meadows placed it near the very top of her list. Beliefs such as "nature is a stock of resources," "growth is always good," or "every person has equal worth" shape the goals, rules and flows below them. Paradigms are hard to change, but when they shift, everything else shifts with them. The abolition of slavery and the extension of the vote to women were, at bottom, paradigm changes. Above even that, Meadows placed the ability to hold any paradigm loosely, knowing that no single view captures the whole.

Pushing the wrong way

Meadows recalled a warning from Jay Forrester: people often find the right leverage point but push it in the wrong direction. Leaders who want more prosperity may push for faster growth, when the system needs slower, steadier growth; officials who want more safety may add rules that people evade. Because systems are counterintuitive, the obvious push can make things worse.

Resilience

Leverage is about change. Resilience is about survival. Resilience is a system's ability to recover from shocks, and it usually comes from redundancy and from many overlapping feedback loops. A body heals because it has many ways of repairing itself. A food system with many farms, routes and stores can lose one without famine.

There is a trade-off. Efficiency means cutting slack: smaller inventories, fewer suppliers, no idle capacity. That saves money in normal times, but optimizing for efficiency often removes the spare capacity that resilience depends on. When the COVID pandemic disrupted factories in 2020 and 2021, carmakers that kept almost no stock of computer chips had to halt production lines for months. A buffer that looked wasteful on a spreadsheet would have kept them running.

Worked example: cleaning a city's air

Imagine a city council that wants cleaner air and has a modest budget. Walk up the ladder.

At the parameter level it could raise fines for polluting vehicles. That might help a little, but owners may pay the fines or register cars elsewhere, as Mexico City discovered. At the level of delays it could install street sensors so that pollution is measured hourly rather than monthly, shortening the time before a problem is noticed. At the level of information it could publish a live map showing the worst-polluted streets beside schools, which gives parents, journalists and businesses a reason to act. At the level of rules it could give neighborhoods the right to close streets to through traffic during school hours. At the level of goals it could judge its transport department by how many people reach work cleanly and safely, not by how many cars it moves.

Notice that the cheapest steps, the map and the change of goal, sit higher on the ladder than the fines. Notice too where resilience fits: a city that keeps several ways to travel, rail, bus, bicycle and road, can absorb the closure of any one. The council's best plan combines a few high-leverage changes and protects the buffers that let the system recover when something goes wrong.