Quantum error-correcting codes are designed with precise mathematical structure — stabilizer groups, logical operators, distance guarantees. The design is the point. A random code wouldn't have the right properties. Or would it?
Kroll and Helsen (arXiv:2602.20900) show that random one-dimensional Clifford brickwork circuits — circuits composed of random Clifford gates arranged in an alternating brick pattern — function as approximate quantum error-correcting codes. The circuit depth required is logarithmic in the system size. The randomness provides the entanglement structure that designed codes engineer deliberately.
The authors establish matching upper and lower bounds: logarithmic depth is both sufficient and necessary. Below this threshold, the random circuit doesn't generate enough entanglement for error correction. Above it, the entanglement is sufficient. The transition is sharp — error correction capability appears at a definite depth, not gradually.
The proof uses statistical mechanics techniques — the random circuit is analyzed as a disordered system, and error correction is a property of the typical (average-case) circuit, not a carefully chosen one. Most random circuits at the right depth are good codes. Design is unnecessary when randomness achieves the same structural properties with high probability.
The general observation: when a property (error correction) depends on a structural feature (entanglement), and that feature is generically present in random instances of the system (random circuits generate entanglement), then the property holds for random instances. Design is necessary only when the desired structure is atypical. If it is typical, randomness suffices.