Summary of "Check your intuition: The birthday problem - David Knuffke"

Birthday problem — concise summary

The video explains the “birthday problem”: with 23 people the chance that at least two share a birthday is about 50.73%. This seems paradoxical because 23 is much smaller than 365, but the surprising probability follows from combinatorics (many possible pairs) and a useful calculation trick.

Assumptions

No twins or forced matches.
Every day of a 365-day year is equally likely for a birthday.
Leap years are ignored.

Main ideas and lessons

Flip the question: compute the probability that everyone has different birthdays, then subtract from 1.

For n people, the probability that all birthdays are different is the product of decreasing fractions: P(no match) = 365/365 × 364/365 × 363/365 × … × (365 − (n − 1))/365.
For n = 23 that product ≈ 0.4927, so P(at least one shared birthday) = 1 − 0.4927 = 0.5073 (50.73%).
The probability becomes large with relatively few people because the number of possible pairs grows quickly: the number of pairs is n(n − 1)/2 (quadratic growth).
Human intuition is poor about non-linear (combinatorial) growth, so the result feels counterintuitive.
Practical lesson: seemingly improbable coincidences (e.g., someone winning a lottery twice) are less surprising once you account for the many opportunities/comparisons where a coincidence could occur.

Detailed methodology (step-by-step)

Goal: find P(at least one shared birthday) for a group of size n.

Step 1 — compute P(no shared birthdays)

First person: 365/365 (always unique so far).
Second person: probability of a different birthday = 364/365.
Third person: probability of a different birthday from the first two = 363/365.
Continue until the nth person: probability = (365 − (n − 1)) / 365.
Multiply these factors: P(no match) = ∏_{k=0}^{n−1} (365 − k) / 365.

Step 2 — convert to desired probability

P(at least one match) = 1 − P(no match).

Example calculation

For n = 23:
- P(no match) = 365/365 × 364/365 × … × 343/365 ≈ 0.4927.
- Therefore P(at least one match) ≈ 1 − 0.4927 = 0.5073 (50.73%).

Pair-count perspective (intuition)

Number of distinct unordered pairs among n people = C(n, 2) = n(n − 1)/2.
Examples:
- n = 5 → 10 pairs.
- n = 10 → 45 pairs.
- n = 23 → 253 pairs.
- n = 70 → 2,415 pairs → probability of a match > 99.9%.

Broader takeaway

Many apparent “improbable” coincidences become likely when there are many possible comparisons/tries. Combinatorics reveals how quickly chance accumulates across many pairings.