Ownership Record: Who Made What in This Project

An honest accounting of the distribution of work between the human author and Claude across the development of these workshop materials. Written in the spirit of the disclosure principles the materials themselves advocate.

This document applies the framework from B.ownership.md to the project that produced it.

Why this document exists

These materials teach researchers to assess and disclose AI's role in their work. Practising that on the materials themselves is the honest thing to do — and makes the point more credibly than any lecture about it.

This is also a useful record for when the materials are published (GitHub Pages, mini-books): attribution should be accurate, not gestural.

Modes of collaboration used in this project

Rather than scoring each document, it is more useful to describe the distinct modes of collaboration that produced them.

Mode 1: User-authored

The user wrote the document independently. Claude was available but not the drafter.

Documents:

Ownership: User — ideas, structure, language.

Mode 2: Claude-drafted from user brief

The user identified the topic and provided a brief (sometimes a sentence, sometimes a list of points to cover); Claude drafted the document; user reviewed and accepted with light or no editing.

Documents:

Ownership: Ideas substantially co-generated; language primarily Claude's; structure primarily Claude's with user steering.

Mode 3: Research agent + Claude synthesis

The user commissioned web research with a defined brief; a background Claude agent ran the searches and produced raw findings; Claude synthesised findings into a structured document; the user exercised editorial oversight and accepted.

Documents:

Ownership: Research brief and editorial direction are the user's. The actual research — finding the sources, identifying the key figures, selecting the relevant findings — was performed by the agent. The synthesis structure and language are Claude's. The facts derive from sources the agent found; the framing of what matters derives from Claude's interpretation of the user's brief.

This is the mode with the most diffuse ownership. The user cannot independently claim credit for finding Pepinsky, Hall, or Berry — the agent found them. The user can claim credit for commissioning the right questions and judging what was useful.

Mode 4: Structural and editorial decisions (throughout)

Across all modes, the following remained entirely the user's:

Overall ownership assessment

Applying the dimensions from B.ownership.md:

Dimension User Claude
Project framing, audience definition, scope Primary Minimal
Structural decisions (what to include, phase structure) Primary Advisory
Core ideas and concepts Shared — user named the problems; Claude structured the frameworks Shared
Research findings (A11–A13, E-docs) Commissioned and curated Executed
Language and prose ~30% (Mode 1 docs) ~70% (Mode 2 and 3 docs)
Evaluative judgement (what is good enough) Primary None

Plain statement: The user directed this project and wrote roughly a third of the documents independently. The remaining two-thirds were drafted by Claude from user-provided briefs, commissions, or topic identifications — and accepted with light editorial revision. The ideas in the Claude-drafted documents are genuinely collaborative: the user framed the questions that shaped the documents, and Claude structured the answers. The language is predominantly Claude's.

What this means for attribution and disclosure

If these materials are published under the user's name:

Accurate disclosure: "These workshop materials were developed in collaboration with Claude (Anthropic). The project structure, editorial decisions, and approximately one-third of the documents were produced by the author independently. The remaining documents were drafted by Claude from the author's briefs and commissions, and accepted following editorial review. All research findings (external sources, policy documents) were obtained via AI-assisted web research agents working from author-defined briefs."

What the user can claim without qualification: The project concept, audience framing, scope decisions, the E2 "own the result" framing, the three-way tool comparison, and all Mode 1 documents.

What requires qualified attribution: The Mode 2 documents — the ideas are collaborative, the language is Claude's. The Mode 3 documents — the research was AI-executed; the user commissioned and curated.

A note on the meta-level

This document is itself a Mode 2 product: the user identified the need for it and the question it should answer; Claude drafted it. The ownership assessment it contains is Claude's interpretation of the conversation record and _history.md. The user should verify that the Mode 1 document list is accurate and that the overall characterisation matches their experience of the collaboration.

If it does not — that discrepancy is itself informative.

Related