Data
Everything on the site can be downloaded, so you can check our work or disagree with it using the same material we used. Atlas files are generated when you download them and record the time; they leave out anything withheld under review and anything we hold privately, such as correspondence. Numbered releases with DOIs will start after the first verified sample.
Policy Atlas
| File | Format | Contents |
|---|---|---|
| atlas.csv | CSV | One row per newsroom: its classification, its position on each remedy, and when it was last checked. |
| atlas.json | JSON | Everything in the CSV plus who decides, each version of the policy we've seen, the sources, and the record's history. |
| evidence.csv | CSV | Every public source behind the records, with links, archived copies and the date we read it. |
| codebook.csv · .json | CSV / JSON | What each column and code means, including the ones that are easy to misread. |
To download a subset, filter the Atlas first; its download links export what you're looking at.
Where the list of newsrooms came from
The 1,805 newsrooms and organizations in the Atlas, and a first automated reading of each one's published policy, come from NSMPA, an open tool that searched each organization's own website for its policy and quoted what it found. It was built by the project's founder while preparing their own request to a student newspaper; that paper's record is held back until someone independent has reviewed it, and it is never contacted automatically.
Nobody has checked NSMPA's readings by hand yet, so they appear on each record separately, marked unverified, and never count toward Atlas totals. Records also list the policy pages NSMPA read, with the date it first saw each one, and note when an organization is on NSMPA's hand-compiled list of known policies. 2,548 short quotes from newsrooms' own policy pages are public. Another 3,829 (from news stories, other sites and search results) stay private, because they can quote people named in old coverage.
Import details and codes
Imported Oct 11, 2026 from NSMPA 0.8.5, commit 985f6c2; workbook SHA-256 8a88f11a0a2524e406058c1318cbee03c022f71550aa757bf01a90745cd934e0. License: CC BY 4.0 (compiled evidence); quoted excerpts remain the work of their publishers.
In the downloads, NSMPA's values are in the columns starting nsmpa_, and its quotes are the evidence rows with human_verified = false.
NO_RELEVANT_GUIDANCE: No relevant public policy locatedUNDETERMINED: Site could not be read well enough to tellSUPPORTS_CHANGED_CIRCUMSTANCES: Policy text permits some remedies when circumstances changeSUPPORTS_RELIEF: Policy text permits some remediesCASE_BY_CASE: Policy text: requests reviewed case by caseUPDATE_ONLY: Policy text: updates or follow-ups onlySTRICT_ARCHIVE: Policy text: no changes beyond correcting errorsMIXED: Policy text permits some remedies and rules out others
Founding documents
| Document | Version | Download |
|---|---|---|
| Complete founding set (mission, methodology, codebook, charter, model policy, citation) | 1.0 | Markdown |
| App. A Research Methodology and Institutional Classification Standards | 1.0 | Markdown |
| App. B Editorial Independence and Institutional Integrity Charter | 1.0 | Markdown |
| App. C The Archive Ethics Model Policy CC BY 4.0 | 1.0 | Markdown |
| Institutional Classification and Evidence Standards | 1.0 | Markdown |
| Mission, Identity, and Guiding Principle | 1.0 | Markdown |
| Open-Use and Citation Framework | 1.0 | Markdown |
| The Permanence Paradox: Research Prospectus and National Research Program Working paper, not yet reviewed outside the project | 1.0 | Markdown |
For a PDF, open the document and use "Print or save as PDF".
References
references.bib (BibTeX) · references.csv. None of these has been re-checked yet.
Numbered releases
None yet.
Citing a snapshot
The Archive Ethics Project. National Policy Atlas [data snapshot]. Retrieved Oct 12, 2026, from https://archiveethics.org/data/atlas.csv. Methodology version 1.0.
What's left out
The downloads never include names of people covered in old stories, private correspondence, anything a newsroom told us off the record, records held back for review, or full copies of articles and policies. Records quote briefly and link to the source.
Reuse
We haven't chosen a license for the Atlas data yet, because some of it quotes other people's work and we can't give away rights we don't hold. The model policy is free to reuse under CC BY 4.0.