Tier 2

systems_analysis

Analyze complex systems using causal loop diagrams, stock and flow models, feedback loop identification, and system archetypes

Usage in Claude Code: /systems_analysis your question here

Systems Analysis

Overview

Analyze complex systems using causal loop diagrams, stock and flow models, feedback loop identification, and system archetypes

Steps

Step 1: Define system boundary

Establish what is inside and outside the system:

  1. IDENTIFY THE PROBLEM/BEHAVIOR:

    • What behavior are we trying to understand?
    • What is the problem symptom?
    • What pattern do we observe over time?
      • Growth (exponential, linear, S-curve)
      • Decline (gradual, rapid, oscillating)
      • Oscillation (boom-bust, cycles)
      • Equilibrium (stable, stuck)
  2. DEFINE SYSTEM BOUNDARY:

    • What elements are “in” the system?
    • What is treated as external (given)?
    • Time horizon: How far back and forward to consider?

    Include in boundary:

    • Elements that directly cause the behavior
    • Elements affected by the behavior
    • Elements involved in feedback loops

    Exclude from boundary:

    • Elements that are truly external (weather, laws)
    • Elements too distant to matter
    • Elements we cannot influence at all
  3. IDENTIFY KEY VARIABLES: List the important quantities that change:

    • Stocks: Things that accumulate (inventory, reputation, skills)
    • Rates: Things that flow (sales per day, hires per month)
    • Auxiliary: Factors that influence but don’t accumulate
  4. ESTABLISH TIME SCALE:

    • What is the dominant time constant?
    • Days, weeks, months, years?
    • This affects what feedback matters

Step 2: Map causal relationships

Create a causal loop diagram showing cause-effect relationships:

NOTATION:

  • Arrow: A → B means “A affects B”
    • sign: Same direction (A increases, B increases)
    • sign: Opposite direction (A increases, B decreases)

PROCESS:

  1. Start with the problem variable
  2. Ask: “What directly affects this?”
  3. Draw arrows from causes to effects
  4. For each arrow, determine polarity (+/-)
  5. Ask: “Does this effect feed back to earlier causes?”
  6. Continue until major relationships are mapped

CAUSAL LINK TYPES:

Positive link (+): Same direction

  • “Higher prices → Higher revenue” (+) [if demand holds]
  • “More training → Higher skill” (+)
  • “More customers → More word-of-mouth → More customers” (+)

Negative link (-): Opposite direction

  • “Higher prices → Lower demand” (-)
  • “More inventory → Less urgency to order” (-)
  • “More workload → Lower quality” (-)

DIAGRAM ELEMENTS:

  • Boxes or labels for variables
  • Arrows showing direction of causation
  • +/- signs on arrows showing polarity
  • R/B labels for feedback loops (added in next step)

QUALITY CHECKS:

  • Each link has clear causal mechanism
  • Polarity is correct (test: if A doubles, what happens to B?)
  • No missing intermediate variables
  • Important delays are noted

Step 3: Identify feedback loops

Find and classify feedback loops in the causal structure:

LOOP IDENTIFICATION:

  1. Start at any variable
  2. Follow arrows until you return to the starting variable
  3. This is a feedback loop
  4. Classify the loop type

LOOP TYPES:

REINFORCING LOOP (R): Self-reinforcing, exponential

  • Count the negative signs in the loop
  • EVEN number of negatives (or zero) = Reinforcing
  • Creates exponential growth OR exponential decline
  • “Vicious cycle” or “Virtuous cycle”

Example: Compound Interest Savings → (+) Interest Earned → (+) Savings This is reinforcing: growth begets more growth

Example: Bank Run Fear → (+) Withdrawals → (+) Bank Instability → (+) Fear Reinforcing: fear creates more fear

BALANCING LOOP (B): Goal-seeking, stabilizing

  • ODD number of negative signs = Balancing
  • Seeks equilibrium, resists change
  • Creates oscillation if delays exist
  • “Checks and balances”

Example: Thermostat Temperature Gap → (+) Heating → (+) Temperature → (-) Temperature Gap This is balancing: seeks the goal temperature

Example: Market Price Price → (-) Demand → (-) Inventory → (-) Price Balancing: price adjusts to clear inventory

FOR EACH LOOP:

  1. Name the loop (descriptive)
  2. Classify as R (reinforcing) or B (balancing)
  3. Describe the behavior it drives
  4. Note any significant delays in the loop
  5. Assess loop strength (dominant or secondary?)

Step 4: Model stocks and flows

Identify accumulations (stocks) and rates of change (flows):

STOCKS: Things that accumulate over time

  • Physical: Inventory, cash, equipment
  • Intangible: Reputation, skills, relationships
  • Informational: Knowledge, data, backlog

Properties of stocks:

  • Change gradually (can’t jump instantly)
  • Create memory in the system
  • Create delays between cause and effect
  • Can only change through flows

FLOWS: Rates that change stocks

  • Inflows: Add to the stock
  • Outflows: Remove from the stock

Stock change = Inflows - Outflows

STOCK-FLOW STRUCTURE:

[Inflow] → [STOCK] → [Outflow]

Example: Workforce [Hiring Rate] → [EMPLOYEES] → [Attrition Rate]

Example: Technical Debt [Debt Creation Rate] → [TECHNICAL DEBT] → [Debt Paydown Rate]

FOR EACH STOCK:

  1. Name the stock (noun - what accumulates)
  2. Identify inflows (what adds to it)
  3. Identify outflows (what depletes it)
  4. Estimate typical time to fill/drain
  5. Note current level (high, medium, low)

LINKING TO CAUSAL LOOPS:

  • Stocks often appear in feedback loops
  • The stock level affects the flows
  • This creates the feedback structure

Step 5: Recognize system archetypes

Identify common system behavior patterns:

ARCHETYPE 1: LIMITS TO GROWTH Structure:

  • Reinforcing loop drives growth
  • Balancing loop creates limiting constraint
  • Initially growth dominates
  • Eventually constraint dominates

Behavior: S-curve growth, then plateau or decline

Example: New product adoption

  • R: Success → Word of mouth → More customers → More success
  • B: Market saturation → Fewer remaining prospects → Slower growth

Leverage: Focus on the constraint before it bites

ARCHETYPE 2: SHIFTING THE BURDEN Structure:

  • Symptom prompts quick fix
  • Quick fix reduces symptom
  • But undermines fundamental solution
  • Dependency on quick fix grows

Behavior: Short-term improvement, long-term degradation

Example: Technical debt

  • Quick fix: Skip tests to ship faster
  • Fundamental solution: Build quality processes
  • Side effect: More bugs → more quick fixes needed

Leverage: Invest in fundamental solution; avoid quick fixes

ARCHETYPE 3: FIXES THAT FAIL Structure:

  • Fix addresses symptom
  • Unintended consequence makes problem worse
  • After delay, problem returns stronger

Behavior: Oscillation, worsening over time

Example: Firefighting mode

  • Fix: Heroic effort to solve crisis
  • Consequence: No time for prevention
  • Delay: Next crisis is worse

Leverage: Break the link to unintended consequences

ARCHETYPE 4: SUCCESS TO THE SUCCESSFUL Structure:

  • Two activities compete for resources
  • Success in one attracts more resources
  • Less resources for the other
  • Winner takes all

Behavior: One grows, other shrinks; lock-in

Example: Internal project competition

  • Successful project gets more budget
  • Other project starves
  • Eventually only “winner” survives

Leverage: Create separate resource pools; explicit portfolio balance

ARCHETYPE 5: TRAGEDY OF THE COMMONS Structure:

  • Shared resource benefits individual users
  • Each user gains by using more
  • Total use depletes the resource
  • Everyone loses

Behavior: Resource depletion, collapse

Example: Technical infrastructure

  • Each team benefits from using shared service
  • No one invests in maintenance
  • Service degrades, everyone suffers

Leverage: Regulate the commons; assign ownership; align incentives

ARCHETYPE 6: ESCALATION Structure:

  • One party acts to gain advantage
  • Other party responds to restore balance
  • First party escalates further
  • Arms race ensues

Behavior: Mutual escalation, exhaustion

Example: Price war

  • Company A cuts price
  • Company B matches
  • Company A cuts more
  • Both margins destroyed

Leverage: Refuse to play; find non-zero-sum alternatives

ARCHETYPE 7: GROWTH AND UNDERINVESTMENT Structure:

  • Growth creates demand for capacity
  • Underinvestment in capacity
  • Performance degrades
  • Demand drops, “justifying” low investment

Behavior: Growth stalls, self-fulfilling low expectations

Example: Startup scaling

  • Growth creates load on infrastructure
  • Don’t invest in scaling infrastructure
  • Site becomes slow
  • Customers leave, growth stops

Leverage: Invest ahead of demand; watch performance standards

Step 6: Identify leverage points

Find high-leverage intervention points:

LEVERAGE POINTS (from least to most powerful):

  1. CONSTANTS AND PARAMETERS Numbers: taxes, subsidies, standards Easy to change but usually low leverage Example: Adjusting price by 5%

  2. BUFFER SIZES Stabilizing stocks relative to flows Example: Increasing inventory safety stock

  3. STRUCTURE OF STOCKS AND FLOWS Physical structure of the system Example: Adding a new warehouse location

  4. DELAYS Length of delays relative to system change rate Example: Faster feedback loops

  5. BALANCING FEEDBACK LOOPS Strength of stabilizing loops Example: Stronger quality control

  6. REINFORCING FEEDBACK LOOPS Strength of growth/decline drivers Example: Reducing viral coefficient

  7. INFORMATION FLOWS Who has access to what information Example: Making performance data visible

  8. RULES OF THE SYSTEM Incentives, constraints, agreements Example: Changing how bonuses are calculated

  9. POWER TO ADD/CHANGE RULES Self-organization, evolution Example: Enabling teams to set their own processes

  10. GOALS OF THE SYSTEM Purpose and direction Example: Shifting from growth to sustainability

  11. PARADIGM/MINDSET Deep beliefs and assumptions Example: From “resources are scarce” to “abundance”

  12. TRANSCENDING PARADIGMS Not being attached to any paradigm Highest leverage, hardest to achieve

FOR THE ANALYZED SYSTEM:

  1. List potential intervention points
  2. Classify by leverage level
  3. Assess feasibility (can we actually change this?)
  4. Assess unintended consequences
  5. Prioritize: High leverage + Feasible + Low risk

Step 7: Analyze interventions

Evaluate potential interventions and predict effects:

FOR EACH POTENTIAL INTERVENTION:

  1. DESCRIBE THE INTERVENTION:

    • What specific change would be made?
    • Which leverage point does it target?
    • What mechanism do we expect?
  2. TRACE DIRECT EFFECTS:

    • What variables change immediately?
    • What is the expected direction and magnitude?
  3. TRACE FEEDBACK EFFECTS:

    • Which feedback loops are affected?
    • Do reinforcing loops amplify the change?
    • Do balancing loops counteract it?
    • How strong are these effects?
  4. CONSIDER DELAYS:

    • How long until effects are visible?
    • Will there be oscillation during transition?
    • Could we declare failure before success appears?
  5. IDENTIFY UNINTENDED CONSEQUENCES:

    • What side effects might occur?
    • Which archetypes might we trigger?
    • Who else is affected and how might they respond?
  6. COMPARE TO DOING NOTHING:

    • What happens if we don’t intervene?
    • Is the system self-correcting?
    • Is patience an option?

INTERVENTION EVALUATION:

InterventionLeverageFeasibilityTime to EffectConfidenceUnintended Risks

RECOMMENDATION:

  • Which interventions should be pursued?
  • In what sequence?
  • What should we monitor to verify effects?

When to Use

  • When dealing with problems that resist simple solutions
  • When interventions seem to make things worse
  • When symptoms keep recurring despite treatment
  • For understanding organizational dynamics
  • When there are long delays between action and effect
  • For strategic planning in complex environments
  • When multiple stakeholders with conflicting goals interact
  • For understanding market dynamics and competition
  • When exponential growth or decline is observed
  • For policy analysis with broad systemic effects

Verification

  • System boundary is clearly defined
  • Causal links have clear mechanisms (not just correlations)
  • Feedback loops are correctly classified (R vs B)
  • Stocks and flows are properly distinguished
  • Relevant archetypes are identified
  • Leverage points are prioritized by actual leverage
  • Interventions are traced through feedback effects

Input: $ARGUMENTS

Apply this procedure to the input provided.