A / CONTROL
0 agentsOrigin world
Population and resources will appear here.
A CONTROLLED EXPERIMENT IN CONSEQUENCES ● LOCAL / DETERMINISTIC
THE BUTTERFLY
ENGINE.
Two worlds begin alike. Change one initial cell in the second world. Watch a tiny disturbance spread—or disappear.
Not a forecast, psychological test or empirical model of organisations. It is an inspectable toy ecosystem for thinking about feedback.
01 / THE EXPERIMENT
ONE WORLD. THEN TWO.
Starting local deterministic simulation…
Ready.
B / PERTURBED
0 agentsEcho world
Population and resources will appear here.
DIFFERENCE FIELD / B − A
Where worlds diverge
Cells with differences glow; darkness means equality.
LOW DIFFERENCE HIGH DIFFERENCE
LIVE MEASUREMENTS / GENERATED FROM STATE
Distance between outcomes
0total absolute resource + occupant cell differences
- Tick
- 0
- Population gap |B−A|
- 0
- Resource gap |B−A|
- 0
- Weather (shared)
- —
- Differing cells
- 0
Zero divergence later is possible. Sensitivity does not guarantee permanent amplification.
TIME TRAVEL / STORED STATES
Replay & timeline
Event ledger
- Initial worlds established.
CAPTURE / REVISIT
Saved observations
Snapshots are local to this page session. Selecting one restores its exact recorded world state. Export saves that snapshot as JSON for inspection.
No snapshots yet. Save a moment you want to revisit.
DISCOVERY / STUDY THE RULES
What exactly is changing?
At tick zero, both worlds share the same seeded resources, occupants and weather schedule. Only the resource amount in the centre tile of World B is altered. The shared seed is not noise injected at each step.
How do agents behave?
Each agent considers staying or moving one tile toward higher available food, with fixed direction priority. It consumes one resource if possible, loses one energy per tick, gains two energy when fed, reproduces at energy ≥ 8 if there is room, and dies at energy ≤ 0. Each cell holds at most one agent.
Where does feedback enter?
Occupants deplete food while empty or occupied cells regrow food based on weather and a 4-neighbour nutrient rule. Local food differences can change movement, survival and reproduction; agents in turn alter food.
Is this mathematically chaotic?
Not established. This is a deterministic, discrete, nonlinear feedback system that can show sensitivity and path dependence. A visible divergence alone does not demonstrate mathematical chaos or a positive Lyapunov exponent.
02 / RESEARCH WITHOUT THE INCENSE MACHINE
THE RIPPLE IS NOT
THE PROOF.
People say a butterfly can cause a storm. The phrase is memorable, but this experiment demonstrates a narrower proposition: in a rule-based world, a controlled local difference may change later trajectories. The size and direction of the effect depend on the rules, thresholds, available pathways and observation horizon.
Sensitivity concerns what happens when initial conditions change. Randomness concerns unpredictable draws; this experiment makes none after seeding, and both worlds receive the same deterministic weather. Path dependence concerns how earlier events constrain later options. Our agents cannot undo consumed food, and later populations inherit those conditions.
Change the sign
Run +1 resource for 40 ticks. Reset with the same seed and choose −1. Compare the gap and timing, not merely which world “wins.”
Change magnitude
Keep the seed fixed. Compare +1 and +6 resources. Larger input need not produce proportionally larger downstream effects.
The one-policy analogy
Imagine a tiny change to onboarding resources in a fictional organisation. Where could reinforcement, scarcity, or thresholds create indirect effects? An analogy, not evidence that real workplaces follow these rules.
“An initial condition is not a destiny. It is a starting constraint that meets a set of rules.”
03 / TAKE THE EXPERIMENT WITH YOU
A GUIDE FOR THE
CURIOUS OBSERVER.
A designed multi-page field guide with model rules, measurements, three guided investigations, organisational analogies, epistemic cautions and further reading.
↓ Download Field Guide (PDF)