Article
Quantum Randomness and Classical Randomness Are Not the Same Thing
2026-07-01When you ask a computer to "pick a random number," you are, in almost every case, asking it to lie to you politely. The number you receive is the output of a deterministic function — a pseudorandom number generator — that would produce exactly the same value again given the same internal state. The randomness is not in the number. It is in your ignorance of the state.
This distinction — between randomness as a property of the world and randomness as a property of your knowledge — is one of the deepest in physics, and it is worth understanding precisely if you intend to select a raffle winner with any rigor at all.
Classical randomness is ignorance, formalized
Every classical source of randomness is, at bottom, a measurement of something you could in principle have predicted. A coin flip is Newtonian mechanics: given the impulse, the air density, and the elasticity of the table, the outcome is fixed the moment the coin leaves your thumb. Diaconis, Holmes and Montgomery showed that a mechanically flipped coin is not even unbiased — it lands the same way it started about 51% of the time [1]. The coin is not random. The coin is merely difficult.
Algorithmic generators formalize this. The Mersenne Twister, for decades the default generator in most programming languages, has a period of 2¹⁹⁹³⁷−1 and passes extensive statistical test batteries [2] — and yet, after observing 624 of its outputs, you can reconstruct its entire internal state and predict every subsequent value with certainty. Cryptographically secure generators harden this by making state recovery computationally infeasible, which is an important engineering property and an entirely different thing from indeterminism. The sequence is still fixed by the seed. The universe, in the classical picture, already knows your raffle winner; it is simply declining to tell you early.
For most purposes this is fine. It must be said plainly: for picking a name out of a hat, a
well-seeded pseudorandom generator is statistically indistinguishable from anything the laws of
physics can offer. If that sentence settles the matter for you, we respect the position and note
that our competitors at /dev/urandom are available free of charge.
Quantum randomness is indeterminacy, measured
Quantum measurement is different in kind, not degree. When a qubit prepared in an equal superposition — the state (|0⟩ + |1⟩)/√2 — is measured, the outcome is not determined by any variable, hidden or otherwise, that lives locally in the apparatus. The Born rule assigns the probabilities [3]; nothing assigns the outcome.
This is not a statement about our ignorance, and we know that with unusual confidence. Bell's 1964 theorem showed that no theory of local hidden variables can reproduce the correlations quantum mechanics predicts [4], and the loophole-free experiments of 2015 — Delft, Vienna, and NIST, using entangled electrons and photons with detection and locality loopholes closed simultaneously — confirmed the quantum predictions decisively [5, 6]. The 2022 Nobel Prize in Physics was awarded for this line of work [7]. If the outcome of a projective measurement on a superposed qubit were secretly determined in advance by local facts, those experiments would have come out differently.
A measurement on such a qubit is therefore the cleanest source of randomness known to physics: an event whose outcome the universe itself had not fixed before the measurement occurred. The formal study of certified quantum randomness — including device-independent expansion protocols — is an active research field [8], precisely because "nobody, not even the device, knew the answer in advance" is a property no classical process can offer.
What this means for your raffle
Practically? Almost nothing. We are contractually obligated by our own honesty policy to say so.
The winner drawn by a quantum measurement and the winner drawn by a seeded Mersenne Twister are statistically identical. No participant, examining only outcomes, could ever distinguish them. Randomness extraction, hardware bias, and readout error mean that even a physical quantum measurement passes through classical post-processing on its way to becoming your result — our own reports document the measured bitstring of every attempt for exactly this reason.
But there is a difference, and it is real, and it is the entire premise of this service: the provenance of the outcome. One process is a function evaluated on a hidden seed. The other is a physical event — a superconducting circuit at 15 millikelvin, an irreversible amplification of a quantum indeterminacy into a classical record — that occurred at a documented time, on documented hardware, under a commitment published before the event took place.
When the selection matters — when there are entrants who will ask how the winner was chosen — "a physical measurement whose outcome was not determined until it happened, documented end to end" is a categorically better answer than "a number our laptop was always going to produce." Not statistically better. Narratively, evidentially, philosophically better.
Whether that difference is worth €50 is a question we leave entirely to you. We would observe only that people routinely pay more than that for bottled water flown in from volcanic islands, on the theory that origin matters. We merely extend the theory to integers.