A tiny change.
An impossible-to-predict result.
(best: 0)
Chaos can't be beaten, only understood. Tiny changes, enormous outcomes.
🏅 New achievement unlocked
The term comes from meteorologist Edward Lorenz, who in 1961 re-ran a computer weather simulation using a rounded version of a number from a printout — 0.506 instead of the full 0.506127 — expecting a nearly identical result. Instead, the forecast diverged completely within a few simulated days. Lorenz had stumbled onto a system where two starting points close enough to look identical produce wildly different long-term outcomes, a property now called sensitive dependence on initial conditions. The vivid "butterfly" image — a butterfly's wingbeat in Brazil setting off a tornado in Texas — comes from the title of a 1972 talk Lorenz gave, and by most accounts that specific title was suggested by the session's chairman rather than coined by Lorenz himself; the metaphor is a simplification for a real, precisely defined mathematical property, not a literal claim that anyone has ever traced a real tornado to a real butterfly.
This game reproduces that property directly instead of just describing it. Each scenario's "Simulate" button runs your adjusted parameter through a chaotic mathematical process (a repeated nonlinear calculation, not a literal ramp or pendulum simulation) many times in a row — and small errors amplify with every repetition, the same mechanism that makes long-range weather forecasting fundamentally limited past about two weeks, no matter how good the model or the data. Move the slider by a fraction of a percent and you can land almost on target one attempt and far off the next: the system is fully deterministic (the exact same input always produces the exact same output) but practically unpredictable, because you can never measure or set your starting parameter with infinite precision.
No — it's a metaphor for how sensitive a chaotic system is to tiny differences, not a real documented chain of cause and effect. What is real and well-documented is Lorenz's own discovery: rounding a single number in a weather model's input changed its output completely within a simulated few days, which is exactly the property this game lets you trigger by hand.
Because the simulation behind each scenario repeats the same calculation on your value dozens of times, and in this kind of calculation, differences don't stay small — they roughly double with every repetition. After enough repeats, two starting values that were 0.1% apart can point to opposite ends of the target range. That's sensitive dependence on initial conditions, not randomness: run the exact same value twice and you get the exact same result.
Yes. Later levels run more repetitions of the underlying calculation and amplify differences faster, so the same size of slider nudge that was manageable on level 1 can swing wildly off target by level 5 — mirroring how uncertainty in a real forecast grows faster the further out you try to predict.
Adjust the scenario's parameter between -5% and +5% and press Simulate.
Try to get as close as possible to the target (0m distance). You have 3 attempts per level.
Tiny changes can give completely different outcomes — that's deterministic chaos.