What TLA+ can and cannot check
What happened
On September 30, 2026, an article on the TLA+ formal method reached the Hacker News front page, discussing what TLA+ can and cannot check. TLA+ can verify safety properties such as invariants, action properties and liveness, the article says, but it cannot express reachability properties or hyperproperties, such as proving a game is winnable or that an energy-saving mode uses less power than a normal one. It also cannot natively define properties spanning more than two steps, floating-point arithmetic or real time, only logical time. The author warns that formal methods are no one-shot fix for agent software development: if a property cannot be written as a logical formula, no formal method can help.
Written by AI from the coverage · updated 55 minutes ago
Coverage
Follow the reports to see the story from different sides.
- Hacker News front pageWhat TLA+ can and can't check
TLA+ 可验证不变量、动作属性和活性等安全性属性，但无法表达可达性属性与超属性，例如证明游戏可获胜、或节能模式比普通模式更省电。它也无法原生定义跨两步以上的属性、浮点运算和真实时间，只能基于逻辑时间。作者提醒，形式化方法无法一劳永逸解决智能体软件开发问题，若属性无法写成逻辑公式，任何形式化方法都无能为力。
Heat over time
Not enough continuous observations to draw a trend yet.