Decision Theory / Pascal's Wager
The Bet That Broke Philosophy's Brain
A seventeenth-century mathematician tried to prove you should believe in God using arithmetic — and the argument still hasn't been cleanly refuted.
The Idea
Blaise Pascal never intended his Wager as a sincere proof of God's existence. He was doing something stranger and more interesting: applying the logic of expected value — the same reasoning behind insurance and gambling — to the ultimate existential question. His argument runs like this. If God exists and you believe, you gain infinite reward. If God exists and you don't believe, you suffer infinite loss. If God doesn't exist, the stakes either way are trivially finite. Multiply any finite probability of God's existence by infinite reward, and the math overwhelmingly favours belief. Bet on God. The philosophical trouble begins immediately. Which God? Pascal assumed the Christian one, but the argument applies equally to any deity with infinite stakes attached — including ones who reward skeptics and punish the faithful. This is the Many Gods Objection, and it doesn't so much refute the Wager as dissolve it: once you accept infinite expected values, every possible theology becomes equally compelling, which is to say none of them are. There's a deeper problem too, sometimes called the Mugging Problem: the same logic can be used to justify almost any absurd action if someone simply attaches a large enough potential payoff. A con artist who promises you infinite cosmic reward for handing over your lunch is, technically, making a Pascal-style argument. What the Wager really exposes is a genuine flaw at the heart of expected value reasoning itself — the machinery breaks down when infinities enter the room.
In the World
In 2011, the philosopher Nick Bostrom updated Pascal's logic for the digital age with a thought experiment that became surprisingly influential in Silicon Valley: what if a future superintelligent AI could simulate suffering or bliss at incomprehensible scale? Even a tiny probability of such an outcome, multiplied by the sheer magnitude of beings involved, should — by expected value logic — dominate every other moral consideration. Some effective altruists took this seriously enough to redirect enormous resources toward AI risk mitigation, treating it as the rational response to Pascal-style reasoning applied to civilisational stakes. Critics, including the philosopher Émile Torres, argued this was precisely the failure mode Pascal's Wager illustrates: once you allow speculative infinities (or near-infinities) into your ethical accounting, your moral compass stops pointing anywhere useful. The numbers become so large that real, present suffering — poverty, disease, ecological collapse — gets arithmetically crowded out by hypothetical future catastrophes of arbitrary scale. The episode is a live demonstration of why Pascal's Wager matters beyond theology. It's not a historical curiosity — it's a template for a kind of reasoning that keeps reappearing wherever large stakes and uncertain probabilities meet. The Wager didn't settle anything about God. What it did was reveal how spectacularly our standard tools for making decisions under uncertainty can fail us when the numbers get big enough.
Why It Matters
Most decisions you face don't involve infinite stakes, which means Pascal's Wager seems safely abstract. But the underlying pattern — letting a tiny probability of an enormous outcome hijack your reasoning — is surprisingly common in everyday life. The person who never flies because of crash statistics, the investor who bets everything on a longshot, the worrier who runs worst-case scenarios on a loop: all of them are, in a quiet way, running Pascal's arithmetic. Knowing the Wager's failure modes gives you a practical mental check. When a decision feels dominated by a scenario that is both catastrophic and highly uncertain, it's worth asking whether you're genuinely weighing evidence or whether a large number has simply captured your attention. The Wager is also a reminder that rationality isn't a single, clean method — it's a toolkit with real limits. The wisest decisions often come not from running the expected value calculation harder, but from recognising when the calculation itself is the wrong tool for the job.
A Question to Ponder
Is there a decision in your own life where the fear of an unlikely but catastrophic outcome is quietly running the show — and if so, is that fear actually tracking reality, or just the size of the number?
Get a new one of these every morning.
Start learning with Thinkable