Formal methods assurance levels debated: TLAPS-proven properties can still fail in real systems
moarbugs · x · 2026-11-05
A technical exchange clarifies the assurance boundary of formal verification: TLAPS-Bench doesn't concern the implementation, so even if TLAPS proves property P in a system's model, P can still be violated in the actual system.
maxvonhippel is working on an illustration of the assurance hierarchy across formal methods — where "actual code" means proof semantics equal execution semantics (as in ACL2 and SBLC), alongside weaker guarantees like differentially fuzzed "slop semantics" and "slop translations." moarbugs adds the caveat that every claimed silver bullet is either BS or marketing.
More from Research
- Weight space programming spawns a Rube Goldberg machine where parts have tiny brains — _ScottCondron · 2026-11-05
- Claude read all 3,096 ASHG abstracts to surface 33 human genetics highlights — anshulkundaje · 2026-11-05
- Stanford's James Zou on AI agents entering the wet lab, in Nature — james_y_zou · 2026-11-05
- Hamel Husain: Benchmarks are used like credit scores, so criticism is warranted — HamelHusain · 2026-11-05
- ArkenOS now runs on Rust, freeing room for more parallel agents — _AustinCalvert_ · 2026-11-05
- CoRL paper GUILD: actively guiding robot data collection toward useful auxiliary tasks — RobobertoMM · 2026-11-05