A Survival Certificate for Machine-Extracted Legal Logic

Summary

Researchers introduce a formal survival certificate for implications extracted from statutes by machine parsers. It measures extractor disagreement, runs 1,000 Monte Carlo trials, and certifies rules only when a one-sided Wilson 95% lower survival bound reaches 0.95. Tests across Missouri and Indian statutes passed preregistered gates, but 93.2% of held-out chapters fell below the informativeness floor under one global error model because calibration rates failed to transfer. The authors recommend chapter-level calibration or error-tolerant deployment.