An overturn rate above 50% is not a quirk in New York's data — it is a structural admission that the initial decision process produces wrong answers more often than it produces right ones on contested cases. The JAMA study's trajectory from 38% to almost 53% overturned between 2019 and 2025 [1] documents that this is a trend accelerating alongside the adoption of automated screening, not a legacy artifact. Regulators who focus on denial volume miss this entirely: volume measures how aggressive the system is; the overturn rate measures how…