Birthday Paradox shows that a room of just 23 people yields roughly a 50 % chance that two share a birthday. The calculation flips the problem: instead of enumerating matches, it multiplies the decreasing pool of available days, yielding (1‑0.4927≈0.5). This counter‑intuitive result appears in textbooks and casual math blogs alike.

When analysts in a 1930s insurance math bureau asked how often three of sixty coworkers would share a birthday, they multiplied \((1/365)^3\) by \(\binom{60}{3}\) and by \((364/365)^{57}\), arriving at a probability of about 0.0006. Austrian mathematician Richard von Mises later argued that this approach fixes the birthday in advance, vastly under‑estimating the true occupancy probability.

Von Mises treated birthdays as balls thrown into 365 boxes, counting any box that receives three or more hits. Applying his formula \(E(x_3)=365\binom{60}{3}(1/365)^3(1‑1/365)^{57}\) yields 0.22, meaning roughly one triple‑match per four to five groups of sixty. The insight reshapes how probability textbooks treat collisions, emphasizing expected counts over fixed‑day events.

Understanding occupancy probabilities matters beyond birthday anecdotes; hash functions in databases and distributed caches exhibit similar collision behavior. Engineers can use von Mises’ expected‑value model to estimate how many keys will map to the same bucket, guiding bucket size choices and reducing performance spikes. The math thus informs practical system design.