MARGINALIA · QUOTED

A generator can produce convincing output from a small and reproducible state.

“The use of pseudo-random processes to generate secret quantities can result in pseudo-security.”

RFC 4086: Randomness Requirements for Security, 2005.

The warning is not an argument against cryptographic pseudo-random generators. It concerns the state from which they begin.

Output can be evenly distributed and still be predictable when the seed space is small, duplicated or recoverable from the environment. Passing statistical tests says little about what an adversary knows. The relevant property is how much uncertainty remains after accounting for boot state, manufacturing data, clocks, cloning, snapshots and previous output.

A review should therefore begin before the first generated byte: which sources exist at startup, when the generator becomes ready, whether its state survives reset, what cloning reproduces and how failure is reported. Randomness is not established by appearance.

Related research: Entropy arrives late.