Modeling Worlds

Given a system description, produce three outputs in order: World Model (structure), Scenario Analysis (branching possibilities), Future Projection (most likely trajectories with confidence levels).

Input: "Model the ride-sharing market in a mid-size city entering a recession"

World Model:
- Actors: drivers, riders, platform companies, city regulators, competitors (public transit, car ownership)
- Forces: fuel prices, unemployment rate, regulatory caps, platform pricing algorithms
- Relationships: driver supply ↔ rider demand ↔ price; regulation ↔ platform margins

Scenario Analysis:
- S1 (40%): Demand drops, drivers leave for other gig work → supply shortage → prices rise → further demand drop (death spiral)
- S2 (35%): Platforms subsidize fares to retain riders → margin compression → consolidation/M&A
- S3 (25%): City subsidizes transit as alternative → ride-sharing shrinks to niche high-value trips

Future Projection:
- 6mo: S2 most likely to dominate short-term (price wars)
- 18mo: Convergence toward S1 or S3 depending on unemployment trajectory
- Key leading indicator to watch: driver churn rate

Progress:

• Step 1: Define system boundary — what's in scope, what's external/exogenous
• Step 2: Identify actors/agents and their incentives, constraints, and behaviors
• Step 3: Map forces and feedback loops (reinforcing vs. balancing)
• Step 4: Construct the World Model (structural map of actors + forces + relationships)
• Step 5: Generate 3-5 divergent scenarios by varying key uncertain drivers
• Step 6: Assign rough probability/plausibility weights to each scenario
• Step 7: Produce Future Projection with time horizons and leading indicators
• Step 8: Flag assumptions and blind spots

1. World Model

Structural snapshot of the system as it stands:

• Actors: who/what participates, their goals, resources, constraints
• Forces: exogenous pressures (economic, regulatory, technological, social)
• Relationships: causal links, dependencies, feedback loops (label reinforcing 🔁 vs. balancing ⚖️)
• Current equilibrium: describe the present steady state, if one exists

2. Scenario Analysis

• Identify 2-4 genuinely uncertain "swing variables" (not fixed facts)
• Combine them into distinct, internally consistent scenarios — avoid a single "optimistic/pessimistic/base case" template; make scenarios structurally different, not just better/worse versions of one story
• Each scenario: name it, describe the causal chain, assign a plausibility (%, or High/Med/Low)
• Ensure scenarios are collectively exhaustive enough to cover the outcome space, not just the obvious ones

3. Future Projection

• Break into time horizons (short/medium/long, calibrated to the domain's pace)
• State which scenario(s) dominate each horizon and why
• Name concrete leading indicators — observable signals that would confirm/disconfirm a trajectory early
• Note tipping points where the system could jump from one scenario track to another

Example 1: Input: "World model for a SaaS startup facing a well-funded competitor entering their niche" Output:

• World Model: actors (startup, competitor, shared customer base, investors), forces (funding runway, switching costs, feature parity race), reinforcing loop: competitor's funding → aggressive pricing → startup churn → weaker startup fundraising position
• Scenario Analysis: S1 (Price war, 30%) — competitor burns cash for market share, startup must differentiate on service/niche depth; S2 (Coexistence, 45%) — market segments by customer size/use case; S3 (Acquisition, 25%) — competitor acquires startup for talent/customers
• Future Projection: 3mo — S1 likely as competitor establishes presence; 12mo — converges to S2 unless startup's churn exceeds 5%/mo (leading indicator), in which case S3 becomes dominant path

Example 2: Input: "Model a coral reef ecosystem under rising ocean temperatures" Output:

• World Model: actors (coral, algae symbionts, fish populations, human fishing/tourism), forces (temperature rise, ocean acidification, local pollution), balancing loop: fish grazing controls algae overgrowth; reinforcing loop: coral bleaching → symbiont loss → coral death → habitat loss → fish decline → algae overgrowth → further coral stress
• Scenario Analysis: S1 (40%) — gradual decline with periodic bleaching events, partial recovery between; S2 (35%) — threshold collapse if a marine heatwave coincides with weak fish populations; S3 (25%) — resilience if a thermally-tolerant coral strain outcompetes
• Future Projection: leading indicator = summer sea surface temperature anomaly; tipping point = 2 consecutive bleaching years without recovery interval