A first-principles critique of the AI Code Review Bench methodology, arguing that AI code review has split into two distinct problems — human comprehension and machine verification — that no single benchmark can yet capture.

The AI race focuses on intelligence. History suggests we should focus on organization.