Why Pseudorandom Isn't Random: The Mathematical Proof That Classical Generative Art Can Never Be Truly Unique
The Claim: "It's Random Enough"
Quantum Genesis NFT #7 — ruído quântico em Origin Quantum
Quantum Genesis NFT #33 — padrões de entrelaçamento GHZ-3
Every generative art platform — Art Blocks, fxhash, custom pipelines — uses a pseudorandom number generator (PRNG). The defense is always the same: "It passes statistical tests. It's random enough for art."
This is true for aesthetics. It is false for uniqueness.
If you're selling uniqueness — especially at collector prices — "random enough" is a category error. Here's the mathematical proof.
How PRNGs Actually Work
A PRNG is a deterministic function:
state_{n+1} = f(state_n)
output_n = g(state_n)
That's it. Given state_0 (the seed), the entire sequence is predetermined. There is no randomness. There is only computational indistinguishability from random — meaning no efficient algorithm can distinguish the output from true random without knowing the seed.
Common PRNGs in generative art:
- xorshift / xoshiro: 64–256 bit state, period 2⁶⁴–2²⁵⁶
- PCG (Permuted Congruential Generator): 64–128 bit state
- ChaCha20 / AES-CTR (CSPRNG): 256-bit state, cryptographic security
- JS
Math.random(): typically xorshift128+, 128-bit state
The Reversibility Problem
This is the core issue: every PRNG is reversible.
Given enough consecutive outputs, you can reconstruct the internal state. For xorshift128+ (used in V8/Node.js Math.random()), 2 consecutive 64-bit outputs are sufficient to reconstruct the full 128-bit state. For PCG, ~3–4 outputs. For ChaCha20, it's computationally infeasible — but only because of the 256-bit key, not because of fundamental physics.
Once you have the state, you can:
- Predict all future outputs — the entire "random" sequence
- Reconstruct all past outputs — run the function backward
- Find the seed — if the seed space is small enough (often it is)
This isn't theoretical. Tools exist: xorshift-reverse, pcg-reverse, js-random-reverse. Given 2–3 NFTs from the same collection, you can often reverse the PRNG state and generate the entire collection yourself.
State Space vs. Output Space
The fundamental limit of any PRNG: the output space cannot exceed the state space.
| PRNG | State Bits | Max Unique Sequences |
|---|---|---|
| xorshift128+ | 128 | 2¹²⁸ ≈ 3.4×10³⁸ |
| PCG64 | 128 | 2¹²⁸ |
| ChaCha20 (CSPRNG) | 256 | 2²⁵⁶ |
| Quantum (12-qubit, 4096 shots) | Effective ∞ | Physics-limited, not state-limited |
Even a CSPRNG with 256-bit state has a finite, enumeratable state space. Given infinite compute (or a sufficiently motivated actor with a quantum computer running Grover's algorithm), the seed space is searchable.
Quantum randomness has no state. The measurement outcome is not generated by a function. It has no internal state to reverse. The Born rule is a fundamental law of physics, not an algorithm.
Quantum Randomness: The Born Rule
When a qubit in superposition α|0⟩ + β|1⟩ is measured, the probability of outcome |0⟩ is |α|², and |1⟩ is |β|². This is the Born rule — one of the most experimentally verified principles in physics.
Key properties:
- No hidden variables: Bell's theorem + experimental violations of Bell inequalities prove no local hidden variable theory can reproduce quantum predictions
- No algorithm: The outcome is not computed. It is a fundamental physical event.
- Irreproducible: The measurement collapses the wavefunction. The pre-measurement state is gone. No amount of compute can "rewind" it.
- Min-entropy = 1 bit per qubit per shot: For a perfectly prepared |+⟩ state, each measurement yields 1 bit of true min-entropy. No algorithm can compress it.
Implications for Generative Art & NFTs
| Property | Classical Generative (PRNG) | Quantum Genesis |
|---|---|---|
| Seed reversibility | Yes (2–4 outputs) | No (physics) |
| Seed predictability | Yes (state reconstruction) | No (Born rule) |
| State space limit | Finite (2¹²⁸–2²⁵⁶) | Physics-limited (∞) |
| Collision possibility | Non-zero (birthday bound) | Zero (physics) |
| Provenance type | Social / algorithmic | Physics-backed |
| Verifiable by | Code audit + seed | Physics + code + chain |
The Mathematical Proof: Min-Entropy Comparison
Definition: Min-entropy H∞(X) = -log₂(maxₓ P[X=x]). It measures the worst-case unpredictability.
For a PRNG with S-bit state: The output distribution is deterministic given the seed. The min-entropy of the output is at most S bits (the seed entropy). Actually less — the output is a function of the state, so H∞(output) ≤ H∞(state) ≤ S.
For a quantum measurement of a qubit in |+⟩ state: P[0] = P[1] = 0.5. Min-entropy = -log₂(0.5) = 1 bit per shot. For N shots, min-entropy = N bits (independent measurements). No algorithm can reduce this. It's a fundamental law.
For Quantum Genesis (12-qubit circuit, 4096 shots): Each shot measures 12 qubits → 12 bits of true min-entropy per shot. Over 4096 shots: 49,152 bits of true min-entropy in the measurement distribution. The SHA-256 seed derivation compresses this to 256 bits — but those 256 bits inherit the full min-entropy of the source.
No PRNG can match this. A PRNG with 256-bit state has at most 256 bits of min-entropy — and that's only if the seed is truly random (which it usually isn't; it's often timestamp + PID). Quantum Genesis seeds have 49,152 bits of physics-guaranteed min-entropy, compressed to 256 bits.
This is why we say: the seed space of Quantum Genesis exceeds the number of atoms in the observable universe, and every bit is physics-guaranteed.
Quantum Genesis #7 — generated from true quantum randomness on Origin Quantum WK_C180
Quantum Genesis — Pseudorandom is predictable. Quantum is physics.
View Collection → |
The Math →
Comentários
Postar um comentário