The Birthday Paradox: 23 People, Even Odds
How many people do you need in a room before it's more likely than not that two of them share a birthday? Intuition mumbles something like 180. The correct answer is 23. With 50 people the probability hits 97%; with 70 it's 99.9%. This result is so reliably offensive to human intuition that it's been a classroom staple for a century — and the reason it offends is the most instructive part.
Why intuition fails
When you hear the question, your brain quietly substitutes an easier one: "what's the chance someone shares my birthday?" That is indeed small — about 6% in a room of 23. But the paradox asks about any pair matching, and pairs multiply brutally: 23 people form 253 distinct pairs. Each pair is a fresh lottery ticket for a match. You're not playing 23 tickets; you're playing 253.
The actual math, gently
Compute the probability of no match and subtract from 1. Person two must dodge 1 birthday (364/365), person three must dodge 2 (363/365), and so on. Multiply the chain out to person 23 and the "everyone dodges everyone" probability falls to about 0.493 — meaning a 50.7% chance of at least one match. No trick, just multiplication grinding intuition into dust.
Try it at your next event
Any gathering of 25+ people is a live demonstration waiting to happen. Classrooms are perfect: a teacher with 30 students has about a 70% chance of a birthday match — ask everyone to shout their birthday month by month and watch the collision land. Wedding tables, office all-hands, five-a-side leagues: the paradox performs almost anywhere people gather in double digits.
Run a birthday-paradox lottery
Draw random numbers 1-365 and see how quickly you get a repeat — it rarely takes more than 25 draws.
Try it with the generator →Where it gets serious: breaking codes
Cryptographers know the paradox as the birthday attack. A digital signature scheme is safe only if nobody can find two different documents with the same fingerprint (hash). Naively, a 64-bit fingerprint offers 2⁶⁴ possibilities — but the birthday math means collisions appear after only about 2³² random attempts, the square root. The paradox halves the effective strength of every hash function, which is why modern systems use fingerprints so long that even the square root is astronomical.
The general lesson
Coincidences are cheap. The birthday paradox is the flagship example of a broad truth: with enough pairs, matches are inevitable — same lottery numbers twice in a year somewhere on Earth, two strangers with the same PIN in one office, déjà vu-grade flukes of every kind. The right question about a coincidence is never "what were the odds of this?" but "how many chances did it have to happen?" The second number is always, always bigger than it feels.