Turing Organization: Isomorphic Interface, Heterogeneous Governance

We propose the Turing Organization: a collaboration paradigm for human-agent organizations built on two axioms — Isomorphic Interface (humans and agents are indistinguishable at the collaboration surface) and Heterogeneous Governance (identity, authority, contribution, and accountability are strictly differentiated beneath the surface). From the axiom that an agent can never be an accountability subject, we derive three iron laws enforced at the code level: authority originates from humans; agent contributions are citable but not accountable; and every judgment is structured, anchored, and audit-logged with mandatory reasons. A systematic audit of the harness-engineering literature and the agent-protocol landscape (A2A, MCP) shows the organizational layer remains unaddressed and is self-acknowledged as open by the most recent survey. We validate feasibility with a ~600-line zero-dependency governance kernel (event sourcing, hash-chained audit, declarative process specifications) and a Web-IM adapter in which a live LLM agent collaborates in a group chat where governance buttons literally do not exist on its client. Companion specification (RFC-0) and reference implementation: github.com/jemarisan/turing-organization. Chinese version included as an additional file. ``` ## 拿到 DOI 后回填三处 1. `docs/site/index.html` 的 #paper 链接 2. 视频简介(见 video-script-3min.md 发布参数) 3. `docs/publish-checklist.md` 第一步 1a 的 DOI 空位