# Randomness is a lie. Let me explain

- Channel: [Veritasium](https://www.youtube.com/channel/UCHnyfMqiRRG1u-2MsSQLbXA)
- Published: 2026-11-18
- Source video: [Watch on YouTube](https://youtube.com/watch?v=uXuQF_PPg5U)
- Read on ReadTube: https://www.read.tube/p/videos/uXuQF_PPg5U
- Language: Original (source language)

## Summary

**The Hidden Patterns Behind Randomness**

The video shows that most computer randomness is deterministic pseudorandomness, making it useful for simulations but vulnerable when its algorithm or internal state can be studied. Weak generators have enabled Minecraft exploits and large-scale casino fraud, while even sophisticated generators can remain predictable without proper security design. Cryptographically secure systems add protection through secret keys, but those keys must begin with physical randomness. Quantum experiments using entangled photons offer one of the closest available approaches to randomness that can be scientifically certified as unpredictable.

The video argues that computer-generated randomness is usually only pseudorandom: it looks unpredictable but follows deterministic rules. This made it useful for simulating nuclear chain reactions, when Stanisław Ulam proposed sampling many possible neutron paths instead of calculating every path individually. John von Neumann implemented the approach on ENIAC, helping establish the Monte Carlo method, but his early middle-square generator quickly revealed the dangers of short cycles, hidden patterns, and predictable sequences.

Simple generators can fail even when their sequences are extremely long. Linear congruential generators, used in software such as Java and older versions of Minecraft, can create structural patterns and expose enough of their internal state to let someone calculate future outputs. In Minecraft, related seeds for clay and diamond placement allowed players to infer diamond locations from visible clay patches; the exploit was eventually fixed by changing the generator. More sophisticated systems such as the Mersenne Twister have vastly longer periods, but their fixed structure can still be reverse-engineered when enough outputs are observed.

These weaknesses can become serious security problems, as demonstrated by gamblers who exploited slot machines using Mersenne Twister. A hacker known as Alex reverse-engineered the machines, predicted favorable outcomes from recorded spins, and built a phone app that compensated for the player’s reaction time. Teams using the system reportedly won hundreds of thousands of dollars, showing why randomness used for games is not suitable for protecting messages, accounts, or financial transactions.

Cryptographically secure generators make prediction practically impossible by using encryption and a secret key, but they still need an unpredictable physical source for that key. NIST researchers are pursuing a stronger standard through entangled photons and Bell tests, whose measurement correlations can rule out locally predetermined results under stated experimental assumptions. Randomness extraction then compresses the certified unpredictability into usable bits, creating numbers that are not merely difficult to predict but whose unpredictability can be experimentally certified.
