As disclosed in the preprint/open peer-review settings when authors submit their paper, preprints cannot be "unpublished". "Publication" means to "make public", and once something is public on the Internet, we cannot go back in time and undo the publication.
Preprints are considered citable documents which is why they receive a DOI, and are submitted to various databases such as Crossref. This is why we can't simply "remove" them. This is in line with policies of other preprint servers [1-2].
Preprints (including submitted manuscripts) will remain on JMIR Preprints in perpetuity and will link to the version of record (VoR) when and if the paper is published formally in a peer-reviewed journal that uses Crossref.
Withdrawal requests are sometimes motivated by fear over the outdated Ingelfinger rule, i.e. another journal rejecting a submission due to prior publication.
While this was still a concern in 1999 when JMIR was founded , as of 2022 this concern is no longer justified, and we will not remove preprints for this reason. Today, the vast majority of journals and publishers encourage preprints and will not reject submissions for that reason, see https://en.wikipedia.org/wiki/List_of_academic_journals_by_preprint_policy.
In 2022 we have done a survey among all medical informatics journals and all of them support prior publication as preprint. Occasionally there may be individual editors who may err by saying it cannot be published due to prior publication, but in this case we recommend authors to escalate this to the editor-in-chief of that journal or the publisher. We also would like to hear about such cases.
Minor improvements made through a peer-review process do not justify a withdrawal of a preprint - our preprint server will automatically point to the published version after it is published, even if it is published on another site, and readers are encouraged to cite the Version of Record instead of the preprint.
The only valid cases where we consider a withdrawal request are
- major errors in the submitted manuscript (to the degree that they constitute a public health concern), or other severe issues that lead to the authors no longer standing behind their findings.
- identifiable information in the manuscript, e.g. patient names or photographs for which no permission was obtained, or other legal issues such as copyright violations that require the entire manuscript to be permanently removed from the public record
In the first case, we will add a new cover page to a preprint PDF that contains an explanation from the authors explaining why the preprint was withdrawn (explaining substantial errors that would render the preprint unpublishable). The originally submitted preprint will still be part of the PDF, but the withdrawal explanation will be on the first page. Vague explanations such as "we are doing additional experiments / analyses to confirm our findings" will not be accepted.
In the second (extremely rare) case, we will replace the entire preprint only with a "tombstone page" and keep the original version internally archived for legal and research misconduct investigation purposes. We will never completely delete the former preprint version but remove it from public view. The tombstone page will only contain the original metadata, who requested removal and when, and the explanation given by the author or the court ordering the takedown.
Please submit a detailed explanation and a signed PDF to firstname.lastname@example.org.
1. Ide K, Koshiba H, Hawke P, Fujita M. Guidelines Are Urgently Needed for the Use of Preprints as a Source of Information. J Epidemiol. 2021 Jan 5;31(1):97-99. doi: 10.2188/jea.JE20200506. Epub 2020 Nov 25. PMID: 33162425; PMCID: PMC7738646.
2. Wright, Jennifer, & Hosseini, Mohammad. (2021). Preprint withdrawals: Proposing a FAIRly template-based approach. FORCE21. Zenodo. https://doi.org/10.5281/zenodo.5765563
3. Eysenbach G. Challenges and changing roles for medical journals in the cyberspace age: Electronic pre-prints and e-papers. J Med Internet Res 1999;1(2):e9
URL: https://www.jmir.org/1999/2/e9 DOI: 10.2196/jmir.1.2.e9