Learn it → Apply it → Save as Skill
Not every place to push on a system is equally powerful. Donella Meadows ranked twelve broad categories of intervention — from tweaking a number to transcending the very paradigm a system was built on — and found that real leverage concentrates exactly where people instinctively don't look.
The Dr. Wallace's Favorite Mental Models is designed to turn useful ways of thinking into reusable tools. Each model gives you a fast visual overview, deeper interactive learning, practical application exercises and an AI-ready Markdown skill you can save for future use.
Instead of merely reading about mental models, you can gradually build your own AI skills library: a collection of portable reasoning skills that can help ChatGPT, Claude, Grok and other AI assistants apply the mental models you value.
In any system — a company, a market, an ecosystem, a body, a piece of software — some intervention points produce far more change than others for the same effort. The weakest points are numbers you can dial up or down. The strongest point is your own willingness to drop the paradigm you're standing on the moment the evidence contradicts it.
Important: leverage and ease of access run in opposite directions. The most powerful points on this list are also the ones a system resists changing hardest.
The model is most useful when repeated attempts to solve a problem by adjusting a number or a rule keep failing to produce lasting change.
Meadows called this list and its exact order "slithery," with exceptions to every item — it's a thinking tool for where to look, not a formula to calculate. Click each rung to explore it, starting from the weakest.
Subsidies, taxes, standards, targets — the numbers attached to a system. Example: raising a fee, tightening a quota, adjusting a budget line. Tweaking a number rarely changes how the system actually behaves, because the structure producing that number is untouched.
Cash reserves, inventory, slack capacity relative to the flows moving through them. Example: a warehouse's buffer stock, a team's spare capacity. These are usually physical or structural, which makes them powerful in principle but expensive and slow to resize quickly.
Transport networks, org charts, pipeline architecture — the physical or organizational skeleton goods and decisions move through. Example: redesigning a supply chain or a reporting structure. Powerful once changed, but expensive and slow to redesign once built.
The lag between an action and the feedback that tells you its effect. Example: the time between a decision and seeing its results. Delays that are too long relative to the rate of change cause overshoot and oscillation; shortening them is high leverage but often structurally hard.
Self-correcting mechanisms that push a system back toward a target. Example: a thermostat, a market price mechanism, a code-review gate, a QA process. Strengthening accurate, fast negative feedback is genuinely high leverage.
Loops that amplify whatever direction they're already moving in. Example: compounding interest, viral referral loops, runaway technical debt, network effects. Slowing a runaway loop buys time for correction before it overwhelms the system.
Who has access to what information, and when. Example: Meadows cites real-time, visible electricity meters cutting household consumption by roughly 30% with no hardware change to the system itself — pure information-structure leverage, cheap relative to rebuilding infrastructure.
Incentives, punishments, constraints — the literal rules of the game. Example: how bonuses are calculated, what gets promoted, contract terms. Meadows put it plainly: "power over the rules is real power," which is why lobbyists fight so hard over legislation.
The system's own capacity to add, change, evolve, or grow new feedback loops, rules, or structure in response to conditions. Example: giving teams autonomy to reorganize themselves, a modular platform or API that lets others build on top. This is a system's resilience capacity.
The purpose a system is actually organized to pursue — not the stated mission, but the revealed one. Example: what a company's incentives and roadmap actually optimize for. Changing the goal changes every rule, structure and flow built to serve it.
The shared, usually unexamined assumptions that make the goals, rules and structures seem self-evident in the first place. Example: "growth is always good," or "employees are a cost to minimize" versus "employees are an investment." Paradigms are intellectually easy to change in an instant but socially slow, because of entrenched resistance.
Holding every paradigm — including your own favorite ones — as one useful lens among many rather than as the truth. Example: the flexibility to drop a cherished model the moment the evidence contradicts it. This is the strongest point on the list, and the one a system (or a person) resists the hardest.
Open an example to see how low- and high-leverage interventions show up in different kinds of work.
A product team sees churn rising and spends a quarter testing price points (leverage point 12, a parameter). The actual driver turns out to be an incentive structure (point 5, rules) that rewards sales for new logos over renewals, so reps keep over-promising to close deals that later churn. No price point fixes a rule problem.
An engineering org sees its bug count climb and launches a push to improve the bug-tracking metric itself (closer to point 10, stock structure) rather than strengthening the code-review gate that should have caught the bugs before merge (point 8, a negative feedback loop). The metric improves briefly; the bug rate doesn't.
Leadership keeps exhorting people to "work harder" on quality, with no effect. Once they make the actual bonus targets (point 5, rules) visible and realign them to reward quality instead of raw output, behavior shifts within a quarter — no exhortation required, because the rule itself was the lever.
A startup builds a real-time usage dashboard visible to every team (point 6, information flow), echoing Meadows' electricity-meter example. With no process change and no new rule, teams start noticing and fixing their own usage drop-offs simply because the information is now visible when it wasn't before.
A leadership team changes its real operating goal from "maximize quarterly revenue" to "maximize customer lifetime trust" (point 3, goals). Sales comp, the product roadmap and how support is staffed all shift to serve the new goal, without each rule being individually redesigned.
A leader built the company on a strategy paradigm that fit an earlier market. When the market shifts, they're willing to drop that paradigm (point 1, transcending paradigms) and rebuild around a new one. A competitor who can't let go of what made them successful keeps optimizing a paradigm the market has already left behind.
This is an illustrative walkthrough designed to teach the method, built around the leadership scenario introduced above.
Expand for where the framework came from and how precisely it should be treated.
Donella H. Meadows was a systems scientist, co-author of the influential 1972 report "Limits to Growth," longtime environmental studies faculty at Dartmouth, and founder of the Sustainability Institute. Leverage Points grew out of decades of work modeling complex systems.
The exact publication history is best treated as approximate rather than a single precisely dated event. An earlier, shorter treatment titled "Places to Intervene in a System" appeared in Whole Earth Review in 1997. The fuller, most widely cited version, "Leverage Points: Places to Intervene in a System," was published by the Sustainability Institute in 1999. Both venues are commonly cited; the framework is best described as first developed in the late 1990s.
No. Meadows herself called the list and its exact order "slithery," with "exceptions to every item." It is a thinking tool meant to broaden where someone looks for leverage, not a calculation that outputs a single correct answer.
Ignoring that resistance increases sharply with leverage sets up disappointment: the points worth the most are also the hardest to move.
Tweaking a number is the easiest point to touch, which is exactly why it gets reached for first — and why people are then surprised when nothing really changes.
Attempting point-2 or point-1 change without the social groundwork tends to produce backlash rather than adoption.
Mistaking a rule (how a bonus is calculated) for a mere parameter (the bonus amount) sends effort to the wrong lever entirely.
A system can quietly absorb or bypass a high-leverage change rather than genuinely adopting it, especially without follow-up.
Meadows' own description was "slithery," with exceptions to every item — it is an example-driven heuristic, not an engineering calculation.
the "system" is actually simple and linear, with no real feedback loops — ordinary analysis is faster and clearer than a systems framework.
you lack the standing or time horizon to act at the higher leverage points. An honest, well-executed lower-leverage fix beats a half-hearted attempt at paradigm change.
the situation genuinely calls for speed, and a parameter tweak is the right-sized, proportionate response — not an evasion of harder work.
This browser tool structures your thinking. It does not send your information anywhere or perform AI analysis.
Save as Skill converts the model into portable Markdown instructions that teach an AI when and how to apply the Leverage Points framework, what questions to ask, what evidence to seek, what mistakes to avoid and how to structure its analysis.
Save the skills you find useful and organize them into your own reusable library. The Markdown is intentionally model-neutral, so it can be used wherever an AI system accepts uploaded files, project knowledge, custom instructions or reusable text context. Platform capabilities differ, so this is a portable skill file rather than a claim of one-click installation.
No. "Lower leverage" means a given unit of effort there produces less systemic change, not that the point is worthless. A well-executed parameter change can still be exactly the right call when speed matters or higher points are out of reach.
Meadows observed that systems defend their paradigms and goals the way organisms defend their identity. A rule can be changed by decree; a shared, unexamined assumption has to be surfaced and let go of by the people holding it, which takes time, trust and social groundwork a number change never requires.
You can try, but Meadows' own writing and the common-mistakes pattern above both warn that leaping to paradigm-level change without groundwork tends to trigger resistance rather than adoption. The ladder is better used to locate where you currently are and where realistic movement is possible than as a race to the top.
Leverage Points is one specific framework within the broader field of systems thinking, focused narrowly on ranking intervention points by effectiveness. Related ideas like feedback loops, stocks and flows, and delays are building blocks the ranking draws on, not interchangeable with it.