# Contributor roles (CRediT)

Date: 2026-10-01. Contributors: **Fernando Hoces de la Guardia** (human) and **Claude Code (Opus, Sonnet)** (AI).

Contributor roles follow [CRediT](https://credit.niso.org/) (ANSI/NISO Z39.104-2022); every entry was given by Fernando Hoces de la Guardia in a questionnaire, not inferred from commit history.

The objects: **Reconciliation page** (index.html at the site root, this page), **Bridge analysis code** (comparison/ R and Python twin, tests, Open Materials), **Disputes table** (comparison/DISPUTES.md, Open Materials), **Hoopes comparison** (the Hoopes (2026) section of this page, this page).

The author answered the 14 roles once, for the BSZ Explorer, and asked for the same answers to be applied to every object on all three pages; the human-verification question was answered separately for each object.

The raw answers are in `credit-answers.json`, the file this page and the landing table are rebuilt from.

## Matrix

| Role | Reconciliation page | Bridge analysis code | Disputes table | Hoopes comparison |
|---|---|---|---|---|
| Conceptualization | Human (lead) | Human (lead) | Human (lead) | Human (lead) |
| Data curation | AI only | AI only | AI only | AI only |
| Formal analysis | AI only | AI only | AI only | AI only |
| Funding acquisition | N/A | N/A | N/A | N/A |
| Investigation | AI only | AI only | AI only | AI only |
| Methodology | AI only | AI only | AI only | AI only |
| Project administration | N/A | N/A | N/A | N/A |
| Resources | N/A | N/A | N/A | N/A |
| Software | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| Supervision | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| Validation | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| Visualization | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| Writing – original draft | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| Writing – review & editing | AI (lead) | AI (lead) | AI (lead) | AI (lead) |
| **Human verification** | Lightly reviewed | Not reviewed | Not reviewed | Not reviewed |

Human only / AI only: one side did the role and the other was not involved. Human (lead) / AI (lead): that side led and the other supported. Equal: both contributed equally. N/A: no one did this for the object.

For Software, Supervision, Validation, Visualization and both Writing roles the answer was "the AI led"; the follow-up question that separates *AI only* from *AI (lead)* was not asked, so these cells read *AI (lead)*: the AI led, with Fernando Hoces de la Guardia supporting (directing, reviewing) or not involved.

Amended 2026-10-01 at the author's request: Conceptualization changed from Equal to Human (lead) on every object (the idea for each object was the author's).

## Statements by object

**Reconciliation page.** Fernando Hoces de la Guardia: Conceptualization (lead), Software (supporting), Supervision (supporting), Validation (supporting), Visualization (supporting), Writing – original draft (supporting), Writing – review & editing (supporting). Claude Code (Opus, Sonnet): Conceptualization (supporting), Data curation (only), Formal analysis (only), Investigation (only), Methodology (only), Software (lead), Supervision (lead), Validation (lead), Visualization (lead), Writing – original draft (lead), Writing – review & editing (lead). Human verification: lightly reviewed.

**Bridge analysis code.** Fernando Hoces de la Guardia: Conceptualization (lead), Software (supporting), Supervision (supporting), Validation (supporting), Visualization (supporting), Writing – original draft (supporting), Writing – review & editing (supporting). Claude Code (Opus, Sonnet): Conceptualization (supporting), Data curation (only), Formal analysis (only), Investigation (only), Methodology (only), Software (lead), Supervision (lead), Validation (lead), Visualization (lead), Writing – original draft (lead), Writing – review & editing (lead). Human verification: not reviewed.

**Disputes table.** Fernando Hoces de la Guardia: Conceptualization (lead), Software (supporting), Supervision (supporting), Validation (supporting), Visualization (supporting), Writing – original draft (supporting), Writing – review & editing (supporting). Claude Code (Opus, Sonnet): Conceptualization (supporting), Data curation (only), Formal analysis (only), Investigation (only), Methodology (only), Software (lead), Supervision (lead), Validation (lead), Visualization (lead), Writing – original draft (lead), Writing – review & editing (lead). Human verification: not reviewed.

**Hoopes comparison.** Fernando Hoces de la Guardia: Conceptualization (lead), Software (supporting), Supervision (supporting), Validation (supporting), Visualization (supporting), Writing – original draft (supporting), Writing – review & editing (supporting). Claude Code (Opus, Sonnet): Conceptualization (supporting), Data curation (only), Formal analysis (only), Investigation (only), Methodology (only), Software (lead), Supervision (lead), Validation (lead), Visualization (lead), Writing – original draft (lead), Writing – review & editing (lead). Human verification: not reviewed.

## How to read this

CRediT roles describe who contributed what, not who is responsible for it.

The **Human verification** row is not part of CRediT. It records how well a human understands each output, separately from who produced it, on a three-point scale: *Not reviewed* (no human eyes have reviewed the output), *Lightly reviewed* (checked, but cannot yet be clearly explained) and *Reviewed* (can be explained to others).