The Case Against Formal Verification, 50 Years Later

Engineers are increasingly excited about formal verification for software, driven primarily by AI coding agents that create gaps in program understanding and make verification faster to implement. This renewed interest contrasts sharply with the field's historically niche status, prompting a re-examination of classic arguments against formal verification from a 1979 paper that claimed program verification was bound to fail. The article explores whether recent developments in AI-assisted coding and verification tools have addressed the fundamental obstacles that were identified decades ago.