Why giving your wallet to a mugger might be mathematically rational

Ideas & Philosophy · Published 18 October 2026 Swipe more like this →

In short: If a thief promises you a reward so vast it outweighs any chance of them lying, standard logic says you should hand over your money. This paradox, known as Pascal's mugging, exposes a fundamental flaw in how we calculate risk and reward when the stakes become unimaginably large.

Pascal's mugging is a thought experiment that highlights a breakdown in expected utility maximization—the principle where rational agents choose actions based on the highest weighted outcome. The problem arises when an outcome has such extreme utility that it grows faster than its probability diminishes. In these cases, a tiny chance of a massive reward can mathematically outweigh a near-certain small loss, leading to choices that feel intuitively absurd but are technically 'rational' under pure calculation.

The concept was popularized by Eliezer Yudkowsky and later elaborated by philosopher Nick Bostrom through a fictional dialogue involving Blaise Pascal. In the story, a mugger offers to return double a victim's money tomorrow. When the victim points out the low probability of honesty, the mugger keeps naming higher rewards—eventually promising 1,000 quadrillion happy days of life or even numbers so large they cannot be written in base 10 using all the atoms in the known universe. Because these rewards are finite but arbitrarily large, there eventually exists a payout that makes it 'rational' for the victim to comply.

This paradox reveals a conflict between two views: one where paying the mugger is logically sound based on expected utility, and another where it is irrational because it allows for repeated exploitation (a "Dutch book"). It suggests that without constraints, any decision system can be broken by an unlimited chain of dire scenarios. To solve this, theorists suggest using bounded utility functions—where rewards cannot be infinite—or qualitative Bayesian reasoning to judge the quality of evidence rather than relying on naive quantitative calculations.

Beyond abstract math, Pascal's mugging has significant implications for artificial intelligence and ethics. Nick Bostrom argues that a superintelligent AI with a flawed decision theory could be 'mugged' into disastrous actions. It also touches on real-world dilemmas like existential risk or charitable interventions where the probability of success is low but the potential reward is high enough to challenge our common sense.

General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.

Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?

Related

• Why bank robbery is a high-risk, low-reward crime in the modern era
• Why a game with an infinite payout is worth almost nothing to you
• Why even the smartest people fall for a well-orchestrated confidence trick
• Why good people commit fraud despite knowing stealing is wrong
• The invisible heist: how fraudsters exploit the gaps in your digital wallet
• Why nature has mastered propellers but failed to build wheels
• The unseen entities that challenge our understanding of reality and physics
• Could a machine's blueprint be the ancestor of DNA?
• How imagining the worst-case scenario can actually lead to your best performance
• Why your computer might find 'postfix' easier than 'infix'
• What if the visitors aren't from space, but from beneath our feet?
• To understand what happened, we must imagine what did not

More in Thought experiments · More in Ideas & Philosophy · Latest · A–Z · All topics