# Open science: how this page measures up to Astera's policy

This page was built as a worked example for the Astera Institute of how Adom's AI publishes an
experiment the moment it exists. This document holds it against the
[Astera Open Science Policy](https://astera.org/open-science-policy/)
([doi:10.5281/zenodo.17873235](https://doi.org/10.5281/zenodo.17873235)), line by line,
including where it falls short.

## Scorecard

| Astera asks for | What this page does | Status |
|---|---|---|
| Publish outputs as soon as they are ready for reuse, not when a paper is written | Each revision was published the day it was designed. Rev 2 went from request to public page in one working session. | Meets |
| Code: MIT license or more open | Board source, scripts and dependency pins are MIT (`LICENSE`) | Meets |
| Code hosted on GitHub | Hosted in this page's git repository on wiki.adom.inc, with history, issues and pull requests. Not mirrored to GitHub. | Partial |
| Data: CC0 waiver | `circuit.json`, 3D models, gerbers, BOMs are CC0 (`LICENSE-DATA`) | Meets |
| Data deposited on Zenodo | Not yet. No Zenodo deposit and no DOI. | Gap |
| Protocols: CC BY, on protocols.io or similar | Each revision's "Reproduce this experiment" section is a protocol, CC BY 4.0 (`LICENSE-DOCS`). Not on protocols.io. | Partial |
| Articles and text: CC BY or more open | All write-ups are CC BY 4.0 | Meets |
| Astera listed as affiliation or funder, with ROR ID | Funder listed: U.S. NSF ([ROR 021nxhr62](https://ror.org/021nxhr62)). Astera ([ROR 00ydx1s47](https://ror.org/00ydx1s47)) is named as the audience, not the funder, because it did not fund this work. | Meets, honestly scoped |
| A list of research outputs shared with the open science team | `outputs.csv`: every file, its revision, type, license and SHA-256 | Meets (in our format, not Astera's template) |
| Persistent identifiers | Page URL plus git commit hash. No DOI. | Gap |
| Methods, controls, provenance, analysis needed to reproduce | Every revision records the exact prompt, toolchain versions, lockfile, the checks run and their numbers | Meets |
| Negative findings shared | Every failed router layout is tabled with its error and split-net counts | Meets |
| Open, iterative peer review | Page comments, issues and pull requests are open; revisions stay public side by side. Not submitted to PREreview or Review Commons. | Partial |
| Machine-readable, AI-ready outputs (FAIR) | `circuit.json` is the complete machine-readable board; `page.json` and `CITATION.cff` carry metadata; the wiki serves all of it over a REST API | Meets |
| Small publishable units | Each revision is its own unit with its own question, method, result and outputs | Meets |

## The three gaps, and what closing them would take

1. **DOIs and Zenodo.** The pipeline to fix this is mechanical: on publish, push the revision's
   `outputs.csv` files to a Zenodo deposition through its REST API, write the returned DOI into
   `CITATION.cff` and the page metadata. This page deliberately shows the state before that
   pipeline exists.
2. **GitHub mirror.** A one-way mirror of each page repo to a GitHub organization would satisfy
   the letter of the code rule without changing how the lab works.
3. **Protocols.io and open review.** The reproduce sections are already structured as protocols;
   posting them is manual today.

## What this page shows that a repository alone does not

A repository stores files; a reader still needs the right software to understand them. Here the
wiki renders the design itself: an interactive 3D board you can orbit and x-ray, the PCB and
schematic drawings, the parts graph, and the fabrication files, all built from the same
`circuit.json` that is published as data. A reader does not need EDA software to inspect the
result, and an AI can read the whole board as JSON.

## How to cite

See `CITATION.cff` at the root of this page. Plain text:

> Lauer, J., Lysandrou, A., and Claude (AI). *Steven's LED Nameplate: three revisions.*
> Adom Industries, 2026. https://wiki.adom.inc/adom/steven-led-nameplate