Ludic Fallacy (Confusing Map for Territory in Probability)

Nassim Taleb's term for the error of applying neat, game-like probability models to the messy, fat-tailed real world. In a casino (ludus = game), probabilities are known, distributions are well-behaved, and the rules are fixed. In real life, probabilities are unknown, distributions have fat tails (extreme events are far more common than bell curves predict), and the rules change without notice. The ludic fallacy occurs when financial modelers use Gaussian distributions for markets, when strategic planners use scenario analysis with fixed probability weights, or when anyone treats real-world uncertainty as if it were a coin flip with known odds.

When to use it

When probability models are being used to justify precise risk assessments in uncertain domains; when statistical analysis assumes known distributions in environments with fat tails; when decision-makers confuse model precision with real-world accuracy; when anyone says 'the probability of this is X%' about a complex, open system.

How it can help

When using probabilistic models: always ask whether you're in a 'casino' domain (known rules, known distributions, bounded outcomes) or a 'real world' domain (unknown rules, fat tails, unbounded outcomes). In casino domains: trust the math. In real-world domains: treat probability models as rough guides, not precise predictions. Specifically: never trust Gaussian risk models for markets, geopolitics, or any domain where extreme events cluster and rules change. Build systems that are robust to outcomes your model says are 'impossible'—those are precisely the outcomes that matter most.