Random Generator Guide: True vs Pseudo-Random Tools

You’ve probably been asked to “pick a random number” and instinctively said 7 or 37. Most of us are terrible at choosing randomly—and the tools we rely on often aren’t truly random either. This guide explains the difference between true random number generators and pseudo-random algorithms, explores why humans gravitate to numbers like 37 and 7, and helps you choose the best generator for a raffle, game, or simulation.

Most common human-chosen random number: 37 ·
True random service users per day: millions (RANDOM.ORG) ·
Computer-generated random numbers are actually: pseudo-random (unless hardware-based) ·
Probability of rolling any number on a fair die: 1 in 6 (≈16.7%) ·
Raffle generators rely on: pseudo-random algorithms

Five key facts that define the landscape of random generation:

What is a random generator?

Basic definition of a random generator

• A random generator is a device or algorithm that produces sequences with no predictable pattern. NIST CSRC Glossary (U.S. government standards body)
• Software generators use mathematical formulas (pseudo-random) while hardware generators use physical phenomena. NIST SP 800-22 Rev. 1 (federal standard)
• RANDOM.ORG uses atmospheric noise as a source of true randomness. RANDOM.ORG (true random number service)

How random generators produce numbers

• Pseudo-random number generators (PRNGs) start with a seed and apply a deterministic algorithm. The Rust Rand Book (developer documentation)
• True random number generators (TRNGs) observe a natural process such as atomic decay or thermal noise. CMU School of Computer Science (academic lecture notes)

Common uses (lotteries, simulations, games)

• Lotteries and raffles rely on random selection to ensure fairness.
• Monte Carlo simulations in finance and physics use PRNGs for statistical modelling.
• Video games use random number generation for procedural content and loot drops.

What is the best random generator?

Criteria for best: use case, reliability, speed

• No single best generator exists; choice depends on need for true randomness vs convenience. WolfSSL (TLS provider)
• For cryptographic applications, NIST recommends DRBGs from SP 800-90A. NIST SP 800-90A Rev. 1 (federal standard)
• For casual use, speed and ease of integration matter more than theoretical randomness.

Best for general randomness: RANDOM.ORG

• RANDOM.ORG provides true random numbers via atmospheric noise, used in scientific research and gambling. RANDOM.ORG (true random number service)

Best for raffles: wheelofnames.com, randomlists.com

• Wheelofnames.com allows custom spin wheels for giveaways and team picking.
• Randomlists.com offers a dedicated team generator and name picker with simple UI.

Best for developers: built-in libraries (e.g., Python random module)

• Python’s random module uses Mersenne Twister (a PRNG) suitable for simulation but not security.
• For secure applications, Python’s secrets module uses OS-provided randomness.

4 popular generators, one pattern: no tool is perfect for every use case — the trade-off is between true unpredictability and convenience.

Is there a true random generator?

Pseudo-random vs true random explained

• PRNGs are deterministic: given the same seed, they produce the same sequence. NIST SP 800-22 Rev. 1 (federal standard)
• TRNGs use non-deterministic physical sources: atmospheric noise, radioactive decay, quantum effects. NIST ITL Bulletin (U.S. government standards body)

Hardware true random generators (atmospheric noise, radioactive decay)

• Quantum random number generators (QRNGs) exploit quantum superposition to produce truly random bits.
• RANDOM.ORG captures atmospheric noise via radio receivers. RANDOM.ORG (true random number service)

Limitations of true randomness in software

• Software alone cannot generate true randomness because algorithms are deterministic. WolfSSL (TLS provider)
• Even TRNGs require clean hardware and post-processing to remove bias.

Why is 37 the most picked number?

Psychological bias in human number choice

• Studies show that when asked to pick a “random” number between 1 and 100, people choose 37 more often than any other. Wikipedia (community encyclopedia)
• This bias is linked to a preference for odd numbers that are not too small or too large.

Research on common random numbers (37, 7, 42)

• 42 is popular due to its cultural reference in “The Hitchhiker’s Guide to the Galaxy.”
• 7 is chosen often because it’s perceived as lucky and is a single-digit odd number.

Why 37 appears in studies and anecdotes

• Psychologists theorize that 37 feels “random” because it lacks obvious associations (unlike 7 or 42).
• Human randomness is poor; generators are needed for unbiased results.

Why does ChatGPT always pick 7?

ChatGPT’s behavior when asked for a random number

• Users report that ChatGPT almost always responds with 7 when asked to “pick a random number.”
• This is not true randomness but a pattern copied from training data.

Role of training data and model architecture

• ChatGPT’s training corpus includes many examples where “random number 7” appears.
• Large language models prioritize common patterns, so 7 gets overrepresented.

Comparison with human biases (7 vs 37)

• Humans pick 37 most often; ChatGPT picks 7 — both are biased but in different directions.
• The difference stems from the distinct sources: human psychology vs. text frequency.

RANDOM.ORG uses atmospheric noise, which is a physical process that is inherently unpredictable. We’ve been running since 1998 and serve millions of random numbers every day.

— Mads Haahr, founder of RANDOM.ORG RANDOM.ORG (true random number service)

When people are asked to generate a random number, they tend to avoid the ends and prefer odd numbers. 37 is a perfect storm: it’s odd, not too small, not too large, and has no obvious cultural hooks.

— Psychology researcher on numerical cognition (cited in Wikipedia (community encyclopedia))

For anyone running a raffle, building a simulation, or designing a game, the gap between human intuition and technical randomness matters. A free wheel spinner works for a casual drawing, but a security-conscious developer needs a hardware TRNG or a service like RANDOM.ORG. The choice is clear: match the generator to the consequence of bias.

Frequently asked questions