发表机构
Stanford University; Quadrivium AI（斯坦福大学; 四元数人工智能）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文提出TRACE模式用于记录推理轨迹，通过字段和测试应对推理不在语言模型中的问题，给出参考程序和操作规范，借助记录-消费者契约形成操作接口，通过示例展示应用，贡献了模式及其契约。
AI 中文摘要
本文定义了TRACE（类型化推理与承诺证据），这是一种用于记录推理轨迹的类型化、版本化模式，以及针对该模式编写记录的参考程序和一种操作规范，即无记录则无持久状态变化。文章分三层论证推理不在语言模型中。该模式通过字段和测试来应对这一缺失，包括TraceRecord及其因果特化、八阶段参考编写器、门优先测量机制、TRACE-Bench协议以及消费者、内存接纳、计划门控、时间遗憾和裁决重用等，更可审计的决策是记录的衡量标准。记录-消费者契约规定了记录的保证和消费者的回报义务，使该模式成为一个操作接口而非被动文档。文中给出两个示例，一个是音乐课程论证，另一个是洪水搜救场景，展示了该模式的应用。附录包含完整模式、编写器算法和成本模型、临床和政策示例、基准协议、收敛指标及使用场景。
英文摘要
This paper defines TRACE (Typed Reasoning And Commitment Evidence): a typed, versioned schema for recording reasoning traces, a reference procedure for writing records against it, and one operating discipline, no durable state change without a record. The paper argues in three layers that reasoning is not in the language model: the autoregressive mechanism natively computes association; chain-of-thought and reinforcement learning inherit its limits; and the formal constructs of reasoning theory, from Socratic procedure to Pearl's ladder, are absent as machinery. The schema answers the absence with fields and tests: the TraceRecord and its causal specialization, an eight-stage reference writer, a gate-first measurement regime, the TRACE-Bench protocol, and the consumers, memory admission, plan gating, temporal regret, and verdict reuse, whose more auditable decisions are the measure of the record. A record-consumer contract states what a record guarantees and what a consumer must honor in return, making the schema an operational interface rather than a passive document. Two worked examples run in the main text: a music-lessons argument traced from sentence to typed verdict, separating association, intervention, and prescription; and a flood search-and-rescue vignette in which a predictive world model reports confident plan success that its own support and out-of-distribution scores contradict, so the record defers the commitment, requests a bounded observation, revises append-only, and clears a different branch. The vignette is illustrative, not empirical; closed-loop evaluation is left to future work, so the contribution is the schema and its contract, not a performance claim. Appendices carry the full schema, writer algorithms and cost model, clinical and policy illustrations, the benchmark protocol, convergence metrics, and usage scenarios.
Comments46 pages, 18 tables, 4 figures
