Check-digit error-detection rates: China's Unified Social Credit Code (GB 32100-2015) versus the US Vehicle Identification Number (49 CFR 565)

Measured error-detection rates for the check characters of two national identifier schemes: China's 18-character Unified Social Credit Code, whose check character follows GB 32100-2015 (a modulus-31 weighted scheme), and the United States Vehicle Identification Number, whose check digit follows 49 CFR 565 (a modulus-11 scheme with character transliteration). Three error classes were measured over 2,000 generated codes per scheme: exhaustive single-character substitution at every position with every alternative character; exhaustive two-position transposition over every unordered pair of differing characters, reported both overall and restricted to the data positions; and fixed-seed Monte-Carlo random double substitutions. Detection means the mutated code fails the scheme's own check equation. Every single-substitution miss in the VIN scheme was attributed to a single cause: the transliteration table assigns equal numeric values to distinct characters, so substituting one for another leaves the checksum unchanged. The runs are reproducible: codes were generated with a fixed-seed PRNG (seed 20260809), the substitution and transposition passes are exhaustive rather than sampled, and the VIN implementation validates against the sample VIN published in NHTSA's manufacturer handbook. Limit that travels with the figures: a check character measures transcription integrity only. A code that passes is evidence of correct copying, not evidence that the company or vehicle exists — anyone who knows the algorithm can construct a passing code.