Meta Rebalancer shows how assignment systems improve when policy, expression, solving, and debugging share a stable contract.
AI 数学成果需要分别通过正确性、重要性与公开传播三道门，并明确保留最终决策责任。
AI math results need separate gates for correctness, significance, and public communication, with decision authority kept explicit.
可信的 Agent 机构记忆需要把每条可行动主张与来源、有效期、依赖关系和失效条件绑定。
AI agent memory becomes trustworthy when every actionable claim keeps its source, validity, dependencies, and invalidation conditions.
补丁半衰期把重写争论变成可复算决策：记录同类故障复发时间、修补成本、迁移成本，以及验证和回退能力。
Patch half-life turns rewrite debates into a measurable decision: track recurring failures, repair durability, migration cost, and rollback readiness.
AI Kill Switch 需要明确权限主体、分级证据、精确隔离、持续留痕和经过测试的恢复路径。
An AI kill switch needs named authority, graded evidence, scoped containment, preserved telemetry, and a tested recovery path.
Anthropic 的生命科学 AI 新实践把模型能力、使用资格与湿实验结果分成三类证据。本文拆解三道门如何连接。
