the-aios/aios/leverage-points

Leverage Points — where to intervene in a system

Most effort goes to the lowest-leverage places (tweak a number, add a buffer) because they're obvious and safe. The highest-leverage places (the system's goal, its rules, its governing paradigm) are less obvious and more resisted — which is exactly why they're high-leverage. This skill is a decision lens for moving your intervention up the leverage ladder.

The core move

When you're about to act on a system, ask: "What's the highest-leverage place I can realistically intervene here?" — then push one rung higher than your first instinct.

Meadows' 12 leverage points, lowest → highest effect:

Full descriptions + examples: `references/meadows-leverage-points.md`.

How to apply (the workflow)

1. Name the system and the symptom. What's actually misbehaving, and what's the recurring pain?
2. Locate your default fix on the ladder. Most first instincts are #12–#10 (change a number, add a buffer). That's fine to note — it's your baseline.
3. Climb one or two rungs. Could you change an information flow (#6 — make the invisible visible), a rule (#5), or who's allowed to restructure (#4) instead of just a parameter? Often a small change at #6/#5 beats a big push at #12.
4. Check the goal (#3) and paradigm (#2). When a fix never sticks, the system's goal is usually fighting you — you're optimizing a number while the system is steered toward a different end. Surfacing the real goal is frequently the actual intervention.
5. Pick the highest rung you can realistically move, and state explicitly which leverage point you're acting on and why.

Counter-intuitions worth remembering

• High-leverage points are often pushed in the wrong direction. People find them by instinct, then turn the dial the wrong way (e.g. adding reinforcing feedback where they needed balancing). Higher leverage = higher blast radius; verify direction before force.
• Parameters rarely change behavior. #12 fixes feel productive and almost never alter the pattern. If you've tuned the same number twice, climb the ladder.
• Information flows (#6) are the cheapest high-leverage move. Giving an actor feedback they previously lacked changes behavior without changing rules or structure. Look here first when "rules" feel too heavy.

Worked example

Symptom: a recurring carry that never gets done. Default fix (#12): re-prioritize it again tomorrow (a parameter tweak — and it's failed twice). Climb: #6 — make why it's carried visible next to the carry (information flow); #3 — question whether the underlying goal is even still active. The real intervention isn't "try harder tomorrow" (#12) but "surface the reason or kill the goal" (#6/#3).

Relationship to the rest of the system

This is the systems-science backing for the antifragile principle "fix the system, not the symptom" and the "Systems Over Goals" operating principle. When /aios:drift or /aios:challenge surface a stuck pattern, use this lens to choose the intervention altitude.