Formal verification was abandoned for labor costs, not infeasibility
jd_pressman · x · 2026-10-23
Replying to lusichu, jdpressman argues the intuition that formal verification is impractical may be a cached, outdated one: academic advances and tooling have steadily accumulated since it was abandoned as a general approach — they were simply too labor-intensive to use. Even if some practical patterns can't be verified, narrowing the search space to only those programs (maybe 1% of code) would still be a huge win; but starting at the kernel would break compatibility with all downstream software.
More from Research
- Varformer: multimodal transformer prioritizes drug targets from population-scale variation — anshulkundaje · 2026-10-23
- Koi Paper: Self-Calibrating Planner Boosts Heterogeneous LLM Fleet Goodput 1.18x, Cuts $/token 20% — burny_tech · 2026-10-23
- Fudan's Code2Video-Compass benchmarks 15 MLLMs on code-based video generation across 2,104 tasks — Fudan-University · 2026-10-23
- Kimi bets RL on a 3T base: trajectory volume beats model capacity, DeepSeek disagrees — teortaxesTex · 2026-10-23
- HCI Methods Chapter Now Compares Frequentist, Estimation and Bayesian Statistics Side by Side — IanArawjo · 2026-10-23
- OSI Hackathon winner Consilith: an open multiplayer studio where researchers and their AI agents collaborate on science — BenBlaiszik · 2026-10-23