About

Selecting a random public figure for trivia nights, creative writing prompts, impersonation games, or educational exercises requires an unbiased selection mechanism. A naive approach using Math.random suffers from predictability and non-uniform distribution across categories. This generator uses crypto.getRandomValues for cryptographically secure index selection across a curated dataset of 200+ verified celebrities spanning 8 domains: Film, Music, Sports, Science, Literature, Television, Business, and History. Each entry is cross-referenced with birth year, nationality, and a verified achievement.

The tool enforces non-repetition within your session history. Category filtering uses set intersection to guarantee uniform sampling within the selected subset. Note: the dataset is intentionally limited to deceased or globally established figures to avoid recency bias. Regional representation skews toward English-language sources. For statistically rigorous sampling across cultural demographics, a larger corpus with stratified sampling would be required.

Formulas

The selection algorithm avoids Math.random due to its pseudo-random nature. Instead, it uses the Web Crypto API for uniform distribution:

index = crypto.getRandomValues(Uint32Array) mod n

Where n is the count of celebrities in the filtered subset. Modulo bias is negligible for n < 2¹⁶ against a 32-bit random source.

The probability of selecting any individual celebrity from a filtered set of size n is:

P(x) = 1n

With history deduplication active, the effective pool shrinks per draw. After k non-repeating draws from pool n, the probability of the next unique selection is:

P_k+1 = 1n − k

When k = n, the history resets automatically to allow full cycling.

Reference Data

Category	Count	Example Figures	Time Span	Primary Regions
Film	30	Meryl Streep, Akira Kurosawa, Audrey Hepburn	1889 - present	USA, UK, Japan, Italy
Music	30	Mozart, Freddie Mercury, Billie Holiday	1685 - present	Global
Sports	25	Muhammad Ali, Pelé, Serena Williams	1895 - present	USA, Brazil, Jamaica, Europe
Science	25	Marie Curie, Nikola Tesla, Ada Lovelace	1564 - present	Europe, USA, Asia
Literature	25	Shakespeare, Toni Morrison, Dostoevsky	1265 - present	Global
Television	20	Oprah Winfrey, David Attenborough, Lucille Ball	1911 - present	USA, UK, Australia
Business	20	Steve Jobs, Coco Chanel, Henry Ford	1839 - present	USA, Europe, Japan
History	25	Cleopatra, Nelson Mandela, Genghis Khan	69 BC - 2013	Global
Total entries: 200+ unique individuals

Frequently Asked Questions

The generator uses crypto.getRandomValues(), which provides cryptographically secure random numbers sourced from the operating system's entropy pool. This is fundamentally stronger than Math.random(), which uses a deterministic PRNG. The random 32-bit unsigned integer is reduced modulo n (filtered pool size). For pool sizes under 2¹⁶, modulo bias is less than 0.0015%, which is statistically negligible.

The tool maintains a session history of up to 50 entries. When every celebrity in the currently filtered categories has appeared, the history automatically resets for those categories, and a toast notification informs you. This guarantees the tool never gets stuck in an infinite retry loop.

Yes. The history is session-scoped and persisted in LocalStorage, but it caps at 50 entries to prevent unbounded memory growth. If you clear your browser storage or exceed the cap, previously seen celebrities become eligible again. Cross-session deduplication would require a different data structure with timestamp tracking.

The dataset is curated from globally recognized English-language sources (encyclopedias, award databases, historical records). This inherently skews toward Western figures post-1800. The History and Science categories attempt to correct this with figures from Africa, Asia, and the Middle East, but full demographic parity would require a dataset 10× larger with regional editorial review.

Each celebrity is assigned exactly one primary category based on their most recognized contribution. For example, Arnold Schwarzenegger is categorized under Film rather than Politics or Sports. This single-assignment rule prevents double-counting in probability calculations and ensures uniform sampling within any filtered subset.

Yes. Selecting fewer categories reduces the pool size n, which increases each individual's selection probability to 1/n. For example, filtering to only Science (25 entries) gives each scientist a 4% chance per draw, versus 0.5% when all 200+ are active.