Feature Request: Prevent deletion of unpublished/draft datasets by non-admin users when the dataset DOI has already been registered with DataCite.

[sergejzr]

Overview of the Feature Request

Prevent deletion of unpublished/draft datasets by non-admin users when the dataset DOI has already been registered with DataCite.

Dear dataverse team,
in Dataverse installations using DataCite, a dataset DOI may be reserved or registered before the dataset is published. This allows researchers to submit the DOI to a journal or manuscript before the dataset is publicly released.

However, if the user later deletes the draft dataset, the DOI may no longer resolve to a Dataverse landing page. This can create a broken/non-existent DOI from the perspective of a journal, reviewer, or reader, even though the DOI may already have been shared externally.

The requested feature is: when a dataset is still in draft/unpublished state, but its DOI has already reached a non-deletable DataCite state, such as Registered or Findable, Dataverse should prevent ordinary users/depositors from deleting the dataset.

Deletion should either be blocked entirely or restricted to superusers/admins, with a clear warning explaining that the DOI has already been registered externally and must continue to resolve to a valid landing page.

What kind of user is the feature intended for?

Depositor, Curator, Superuser, Sysadmin.

What inspired the request?

We encountered a real-world support case in our Dataverse installation, bonndata, which runs Harvard Dataverse.

A user created a draft dataset and asked the repository team to reserve/register a DOI. The DOI was then submitted to a journal. Later, the user accidently deleted the draft dataset. As a result, the journal had a DOI that no longer pointed to an existing Dataverse dataset landing page.

This creates a difficult recovery situation for repository administrators and can damage trust in DOI-based citation workflows.

What existing behavior do you want changed?

Currently, a user can delete a draft dataset even if the dataset DOI has already been registered with DataCite or otherwise communicated outside Dataverse. (AFIK, we can not remove this permission from dataset creators)

This is risky because the DOI may already have been submitted to a journal, included in a manuscript, shared with reviewers, or used in other scholarly communication contexts.

The current behavior should be changed so that Dataverse checks the DOI state before allowing deletion of an unpublished dataset.

If the DOI is only locally reserved or still deletable as a DataCite Draft record, deletion may remain possible, depending on local policy.

If the DOI is already Registered or Findable at DataCite, Dataverse should not allow ordinary users/depositors to delete the dataset (or at least warn them).

Any brand new behavior do you want to add to Dataverse?

Add a DOI/PID state check before deleting a draft dataset.

Suggested behavior:

1. When a user attempts to delete a draft dataset, Dataverse checks the dataset PID provider and PID state.

2. For DataCite DOIs, Dataverse checks whether the DOI is in Draft, Registered, or Findable state.

3. If the DOI is Registered or Findable, block deletion for ordinary users/depositors.

4. Show a clear message, for example:

   “This draft dataset cannot be deleted because its DOI has already been registered with DataCite. Registered DOIs are persistent and must continue to resolve to a valid landing page. Please contact repository support or a Dataverse administrator.”

5. Allow superusers/admins to handle exceptional cases, either by:

   • preventing deletion for everyone and requiring an administrative recovery workflow, or
   • allowing superusers to override with an explicit confirmation and warning.

Possible implementation details:

• Reuse existing PID/DataCite state-checking functionality if available.
• Make the behavior configurable for installations that may want different local policies.
• Consider applying similar protection to file-level DOIs if file DOIs are enabled and already registered.

Any open or closed issues related to this feature request?

I did not find an exact existing feature request for preventing deletion of a draft dataset after its DOI has been registered with DataCite.

Possibly related issues:

• Implement ability to reserve DOIs through DataCite #5093 — Implement ability to reserve DOIs through DataCite

Are you thinking about creating a pull request for this feature?

Not at this time.

We are reporting this based on an operational support case in our Dataverse installation. We would be interested in discussing the desired behavior and whether this should be implemented in Dataverse core, especially for installations using DataCite DOI registration before publication.

Thank you!
Sergej