2026-07-27｜当一篇论文有了自己的位置

观察类型： 个人研究观察 主题： 公开记录 · 人类判断 · 非自动责任

今天，我的新论文正式有了 DOI。

这件事看起来只是上传文件、填写标题、确认作者信息，但做完以后，我反而觉得它不像一个简单的发布动作，更像是把一段长期思考放到了一个可以被外界看见的位置。

论文的题目是：

《连续主体与非自动责任：高复制人工智能系统中的身份、判断与归责边界》

英文题目是：

The Continuous Subject and Non-Automatic Responsibility: Boundaries of Identity, Judgment, and the Attribution of Responsibility in High-Replication AI Systems

DOI：

https://doi.org/10.5281/zenodo.21632567

今天真正让我有感触的，不只是论文被发布了，而是发布过程中每一个看起来很小的选择，其实都需要人重新判断。标题要不要完整，文件名是不是对应这篇论文，中文和英文版本是不是放在同一条记录里，作者姓名和 ORCID 是否一致，许可证是否合适，前面的研究应该如何被引用，这些都不是系统自动替人判断好的事情。

系统可以保存文件，也可以生成链接，但它不会自动知道一篇论文真正属于谁，哪个版本可以公开，哪一个标题最准确，哪一个记录会在以后被别人引用。很多时候，人以为自己只是填写页面，其实是在决定一项研究以后如何被理解。

这也让我再次想到，记录本身并不等于责任。一个名字出现在文件里，不代表这个人自动承担后来所有变化；一个过去的表达被保存下来，也不代表它自动授权未来所有使用；一个记录被系统固定住，不代表它已经替代了人的知情、判断和确认。

今天的发布过程很普通，但它让我更清楚地看到：AI 时代的很多问题不是发生在特别宏大的地方，而是发生在这些具体的小动作里。一个文件上传错了，可能会继续被复制；一个标题写得不准确，可能会影响后面的理解；一个身份标识没有对齐，可能会让作者和作品之间的关系变得模糊。

所以人类判断不是抽象的词。它有时候就是重新看一遍标题，就是检查一个文件名，就是确认一个版本，就是发现不对以后停下来修改。很多重要的边界，都是在这种看起来很小的地方被守住的。

我也越来越觉得，公开研究不是把所有过程都摊开给别人看。公开的部分应该清楚、稳定、可以被引用；未公开的准备材料和非正式草稿，则继续留在私有范围内。这样既能让外界看到正式成果，也能避免把还没有完成确认的内容提前带入公共环境。

今天这篇论文有了自己的公开位置。对我来说，它不只是一个链接，也是一种确认：这项研究有作者、有标题、有版本、有 DOI，也有清楚的责任边界。

记录可以帮助别人找到它、引用它、核验它，但记录仍然只是记录。它不能自动替代真实知情，不能自动扩展授权，也不能自动生成责任。

这一点，正是这篇论文想要继续讨论的地方。

当下观察

一篇论文真正进入公共环境，不只是因为它被上传了，而是因为作者、标题、版本、文件和引用关系终于能够彼此对应。

系统可以保存记录，但不能自动完成判断。

记录可以支持引用，但不能自动产生责任。

今天这个发布过程让我再次确认：AI 时代越依赖系统记录，越需要人保留真实确认的位置。

---

July 27, 2026｜When a Paper Has Its Own Place

Observation Type: Personal Research Observation Themes: Public Record · Human Judgment · Non-Automatic Responsibility

Today, my new paper formally received its DOI.

At first glance, this may look like a simple act of uploading files, filling in titles, and confirming author information. But after completing it, I felt that it was not merely a publishing action. It was more like placing a long period of thought into a position where it can be seen by others.

The title of the paper is:

The Continuous Subject and Non-Automatic Responsibility: Boundaries of Identity, Judgment, and the Attribution of Responsibility in High-Replication AI Systems

Chinese title:

《连续主体与非自动责任：高复制人工智能系统中的身份、判断与归责边界》

DOI:

https://doi.org/10.5281/zenodo.21632567

What moved me today was not only that the paper had been released. It was that every small choice in the release process still required human judgment. Whether the title was complete, whether the filename truly belonged to this paper, whether the Chinese and English versions should remain within the same record, whether the author name matched the ORCID, whether the license was appropriate, and how earlier research should be cited—none of these were decisions that the system could fully make on behalf of the author.

A system can preserve files and generate links, but it does not automatically know who a paper truly belongs to, which version is ready for public release, which title is the most accurate, or which record may later be cited by others. Often, what looks like filling in a page is actually shaping how a piece of research will be understood in the future.

This reminded me again that a record itself is not responsibility. A name appearing in a file does not mean that the person automatically bears responsibility for all later changes. A past expression being preserved does not mean it automatically authorizes every future use. A record fixed by a system does not mean that it has replaced actual knowledge, judgment, and confirmation.

Today’s release process was ordinary, but it made me see more clearly that many problems in the AI era do not only occur in grand situations. They also appear in small concrete actions. If a file is uploaded incorrectly, it may continue to replicate. If a title is inaccurate, it may shape later understanding. If an identity marker does not align, the relationship between the author and the work may become unclear.

This is why human judgment is not an abstract term. Sometimes it simply means reading a title again, checking a filename, confirming a version, or stopping to correct something when it feels wrong. Many important boundaries are protected in small moments like these.

I also increasingly feel that public research does not require every part of the process to be exposed. The public part should be clear, stable, and citable. Preparatory materials and informal drafts can remain within a private scope. This allows formal work to be visible while preventing materials that have not been fully confirmed from entering the public environment too early.

Today, this paper gained its own public place. For me, it is not only a link. It is also a form of confirmation: the work has an author, a title, a version, a DOI, and a clearer responsibility boundary.

A record can help others find it, cite it, and verify it. But a record remains a record. It cannot automatically replace actual knowledge. It cannot automatically expand authorization. It cannot automatically generate responsibility.

This is exactly where the paper continues its discussion.

Current Observation

A paper truly enters the public environment not only because it has been uploaded, but because its author, title, version, files, and citation relationship finally correspond to one another.

A system can preserve records, but it cannot automatically complete judgment.

A record can support citation, but it cannot automatically create responsibility.

Today’s release process reminded me again: the more the AI era relies on system records, the more human confirmation must remain visible.

Research Context

Complex Systems · AI Governance · Decision Architecture · Enterprise Transformation · Human Judgment

Public Research Release · Non-Automatic Responsibility · Responsibility Boundary · Record Clarity · Human Confirmation

Canonical Public Research Reference

Xufen Tu（涂绪芬）

Independent Interdisciplinary Researcher in AI Governance & Complex Systems

Human Judgment as Structural Constraint in AI-Mediated Complex Systems

Canonical public research source: https://xufentu.com