Make the record more useful.

Study first. Share what you learn.

Independent researchers, developers, historians, students, and preservation volunteers are welcome. The collection expects a reasonable, good-faith effort to contribute useful methods, findings, corrections, or unsuccessful attempts after research, when possible. A favorable result, affiliation, or publishable discovery is unnecessary.

There is no advance application or contact requirement. ZIP and torrent access follow the same purpose-of-use acknowledgment already used by this collection. No contributor account or identity verification has been added.

1. State a question another person can check.

Choose an existing project or describe a different concrete question. State what could count against your interpretation before examining the result. “How many include targets change between these identified copies?” is more testable than a general claim that a release is important.

Exploration is welcome. If you do not know which question the material will answer, report the scope you examined and the point at which a question emerged. Do not turn a speculative idea into a completed result.

2. Explain the inputs and alternatives.

Record the reference identifier, selected member identifiers, tool versions, and scope. Distinguish a container digest from member identity. Say whether public JSON/CSV, a manifest, selected definitions, or a lawfully held installation was sufficient. Explain the uncertainty left by narrower alternatives before claiming a whole installation was needed.

The published data permits several checks without any client download. A need to inspect a particular definition is different from a need to execute it; execution introduces environmental, rights, and service questions that a static study cannot settle.

3. Preserve the route from method to finding.

Describe repeatable steps, controls, observations, and interpretation separately. Mark documentary statements, direct measurements, inferences, and untested hypotheses. Report an unexpected, negative, or inconclusive result with its stopping point and alternative explanations.

For example, if Inventory’s root-only graph mask produces no matched edge, state that it is a calculation on a recorded graph. It does not show that a running client failed. If a ZIP digest differs, compare member identities before declaring that the historical payload changed.

A useful correction identifies the affected claim, the evidence, a reproducible check, and the narrower wording the evidence supports. It need not agree with the archive’s current interpretation.

4. Prepare a clean public record.

Remove personal names attached to private activity, account details, home-directory paths, IP addresses, credentials, logs, personal configurations, and participant identifiers. Prefer relative paths, aggregate results, and original authored examples. Do not attach the client, copyrighted artwork, or entire interface payloads where an explanation or measurement suffices.

Human-subject work needs an appropriate consent and research process. This collection supplies neither participant recruitment nor consent management. A proposed study involving people must stay proposed until it has a suitable basis and has actually been performed.

A small, portable research note.

Download the contribution template · JSON ↓

The template contains no name, email, or affiliation field. It covers the question, actual inputs, alternatives, method, findings, uncertainties, reproduction steps, corrections, and a privacy check. Fill it locally; downloading it does not submit anything.

Research participation and an access acknowledgment do not grant third-party software rights, establish permission to redistribute, or prove a legal exception. Keep the scope of a research finding distinct from the basis for any copying or distribution.

Continue with the collection.

Research overview · Project register · ZIP and torrent downloads