Publishing, DOIs
and what changes later
Three states, one direction, and a rule about editing. Everything here is written so that nobody discovers a constraint after they have committed to it.
Three states, and what each one costs
Transitions run one direction only. A page people have linked cannot be unpublished, and a DOI cannot be un-minted.
| State | What you gain | What you accept |
|---|---|---|
| Draft unlisted, noindex, random URL | Share for review without committing. Change anything, including the conclusion. | It is still a public URL. Anyone with the link can read and forward
it. noindex is a request, not enforcement. A cautious journal
may count it as disclosure. Nothing is archived. |
| Published no DOI | Finished, indexed, findable. Still correctable — edits overwrite. | Not citable in a reference list, and not archived independently. An edit destroys the previous state, so a reader who quoted a number cannot discover that it changed. |
| Published + DOI | Citable. Archived independently. Versioned — edits accumulate instead of overwriting. Belongs on a CV. | Permanent. The archived record cannot be withdrawn, so mistakes are on it too. A correction means a new version with a visible supersession trail. Posting before a journal makes it a preprint. |
Drafts are unlisted, not private. There is no login here and there will not be one: authentication means accounts, password resets and a class of security bug, for a site whose whole value is being simple and public.
So the rule is: if it needs to be genuinely private, it is not on this site yet. Share a file or a screen instead.
No DOI by default
A DOI is a promise of permanence, and most instruments do not need one. An instrument with no domain author is a demonstration and an invitation — a URL is enough. When someone puts their name and their interpretation on it, permanence starts to matter, and then it is offered.
| Situation | DOI | Why |
|---|---|---|
| Instrument published, no domain author yet | no | A demonstration. Nothing depends on it being permanent. |
| You write the interpretation and take lead credit | offered | You may want to cite it later, and now you have something at stake |
| The instrument is the record — no paper behind it | yes | Nothing else holds its identity |
Where a DOI is minted, the record is archived independently of this site — so if the site stops, your work does not. That is the main reason to take one.
Allowed — and the ceremony scales with the change
| What changed | What it needs |
|---|---|
| Typo, broken link, clearer wording | Nothing. Edit freely — no reader's understanding changes. |
| A number moves | A visible "updated" date and one line saying what changed. Someone may have quoted the old figure. |
| The conclusion changes | Not a silent edit. A new version — and usually the moment to mint a DOI, because the work has just become worth citing. |
The middle row is the one that gets skipped, and it is the one that matters. A number that changes without a date is how a reader ends up citing something that is no longer on the page.
When new data changes the answer
Instruments are re-runnable, so they get re-run — and sometimes the finding does not survive newer data. That is not a maintenance problem, it is a result: this held in 2024 data and does not hold in 2026 data is worth publishing, and almost nobody tests it systematically.
The earlier version stays reachable, stamped superseded. It is a state that happens to an article rather than one anybody selects.
You can still publish in a journal
If the instrument goes up before your paper, it is functionally a preprint, and what governs is your target journal's preprint policy. Most journals accept preprints; policies vary by publisher and by field, and this site cannot track them. Check Sherpa Romeo and your journal's own author guidelines before posting.
| Order | Risk |
|---|---|
| Journal first, instrument afterwards | None. The paper is the record; this is the companion. |
| Instrument first, journal later | Real, and field-dependent. Check first. |
Two framing choices reduce the risk to almost nothing, and both are permanent policy here: this site never claims peer review of record, and each case is presented as a research artifact rather than as a paper.