| name | system-behavior-diagnosis-and-leverage |
| description | >- Use when this capability is needed. |
System Behavior Diagnosis and Leverage
When to use
Use this when a problem keeps recreating itself, growth runs into limits, feedback arrives late or distorted, or a rule, metric, or fix produces side effects instead of lasting change.
Diagnose from behavior to structure
- Observe behavior over time — What is changing, what is stable, what is oscillating, and over what time scale?
- Identify the stock — What quantity accumulates, drains, or buffers the system?
- Trace inflows and outflows — What moves the stock up or down?
- Map feedback loops — Which loops are balancing, which are reinforcing, and which dominates now?
- Separate delays — Where are the information, decision, response, and physical delays?
- Set the boundary by purpose — What outside stocks or processes matter for the time horizon you care about?
- Find the current limit — What factor is binding now, and what next limit will growth create?
- Name the level of explanation — Is the issue an event, a behavior pattern, or the underlying structure?
Pattern cues
- Stable, recurring, or goal-seeking behavior usually points to a balancing loop.
- Accelerating growth or decline usually points to a reinforcing loop.
- Overshoot, oscillation, or repeated target-miss usually means a delayed or over-strong balancing loop.
- A problem that persists despite repeated fixes usually means the structure recreates the problem.
- Sustained growth usually means a reinforcing growth loop is meeting a balancing constraint.
- If growth continues, keep asking: What limit does this growth create next?
Quick diagnostic prompts
- What is the stock?
- What are the inflows and outflows?
- What information reaches the decision maker, and when?
- Which loop is dominating right now?
- What delay keeps the system from correcting immediately?
- What hidden stock, leak, or outside process have I left in the cloud?
- Is the real bottleneck still the bottleneck, or has it shifted?
Choose leverage
Work from lower leverage to higher leverage only as needed:
- Numbers / parameters — change constants, taxes, subsidies, or standards only when a threshold can trigger deeper change.
- Buffers — enlarge buffers when the system is too vulnerable; shrink them when it is too sluggish.
- Stock-and-flow structure — redesign bottlenecks, congestion, pollution, and capacity limits when the layout itself creates the problem.
- Delays — shorten delays if possible, but sometimes slowing the system is safer than speeding it up.
- Balancing feedback — strengthen weak corrective loops and sensing where correction is failing.
- Reinforcing feedback — weaken runaway compounding loops and the advantage that feeds on itself.
- Information flows — put timely, accurate, hard-to-distort feedback where the decision actually happens.
- Rules — change incentives, constraints, permissions, and enforcement when the system optimizes the wrong thing.
- Self-organization — preserve the ability to adapt, diversify, and create new structure.
- Goals — change what the system is trying to achieve.
- Paradigms — change the assumptions that generate goals, structures, and rules.
- Transcending paradigms — hold models lightly; stay open to surprise and to being wrong.
Common system traps
- Policy resistance — multiple actors undo each other’s interventions. Search for a shared goal or a redesigned feedback structure.
- Tragedy of the commons — restore direct feedback between use and consequence.
- Drift to low performance — keep standards absolute instead of lowering them after weak results.
- Escalation — break the competitive loop or agree on limits.
- Success to the successful — level the playing field or create exits for losers.
- Shifting the burden — strengthen the system’s own repair capacity before withdrawing symptom relief.
- Rule beating — align the rule with its purpose, not just compliance.
- Wrong goal — test whether the metric actually reflects system welfare.
Working habits
- Make assumptions explicit.
- Use small steps and monitor effects.
- Optimize the whole, not the parts.
- Design responsibility so decision makers feel consequences directly.
- Use language that names feedback, resilience, sustainability, and carrying capacity.
- Expect surprises; treat errors as data.
- Do not define the problem as "lack of my preferred solution."
Quick model sketch
When the behavior is unclear, sketch a stock-and-flow model:
stock(t) = stock(t - dt) + (inflow - outflow) x dt- mark delays and information paths
- label balancing and reinforcing loops
- test how behavior changes when you vary rates, delays, buffers, or goals
Source note
Extracted from Thinking in Systems: A Primer by Donella H. Meadows.
Source: benschrauwen/wisdom-collector — distributed by TomeVault.
