Chaos vs Randomness
Here are two jittery signals. One follows a hidden rule (it's chaotic but completely determined). The other is pure chance. Watch them for a few seconds — can you tell which is which? Then make your guess and see the trick that gives it away.
Sprinkle a teeny bit of random fuzz on top of the rule. Watch the return-map curve blur toward a cloud — that's what real measured data looks like.
Plot each value against the next
In time, chaos and randomness both look like noise — that's why you can't tell them apart up top. The reveal plots every value against the one that follows it.
The chaotic signal collapses onto a clean curve — proof of a hidden rule (here, next = 4·x·(1−x)). The random signal just fills a shapeless cloud: knowing one value tells you nothing about the next.
Turn on Expert level and the curve thickens into a fuzzy band — real data is always a rule plus noise. Crank it up enough and even the rule disappears into the cloud. Finding the hidden curve inside noisy data is exactly what chaos scientists do for a living.
Determined ≠ random ≠ predictable
Chaos is deterministic: a rule decides every step, so it has structure you can uncover. Randomness has no rule and no structure.
Yet both are unpredictable far ahead — chaos because tiny errors explode (and the computer only has so many digits), randomness because there's nothing to predict at all. Same surprise, totally different cause.
For teachers & grown-ups
The chaotic signal is the logistic map at r = 4, xₙ₊₁ = 4·xₙ·(1−xₙ); the random signal is a uniform pseudo-random stream. The reveal is a first-return map (xₙ on the horizontal axis, xₙ₊₁ on the vertical): a deterministic 1-D map traces out its defining curve exactly (the parabola), while independent random draws scatter uniformly over the square. This is a real technique — reconstructing such maps from data (Takens' embedding) is how scientists detect low-dimensional determinism in messy signals like heartbeats, dripping faucets, and plasma turbulence. Two good follow-ups: (1) the long-run histograms also differ — the logistic map's values pile up near 0 and 1 (an arcsine density) while the random ones are flat; (2) chaos is short-term predictable (the curve lets you forecast one step) whereas randomness is not predictable even one step ahead. Determinism, randomness, and predictability are three separate ideas, and this tool pulls them apart.