Once published, is a paper permanent?
Imagine that you publish a paper in an academic journal. It passes peer review, is accepted, receives a DOI and appears online. Researchers read it, some cite it, and others may disagree with it. Perhaps somebody publishes a response challenging your conclusions.
Then, months or even years later, the journal decides that the paper should no longer be available. It is not corrected, accompanied by an editorial note or even retracted in the conventional sense. It is removed.
Is that something a journal should be able to do? At first sight, the answer might seem straightforward. Journals control their publishing platforms and surely must be able to decide what appears on them. But scholarly publishing is not quite like publishing a newspaper article, maintaining a corporate website or deleting an old social-media post. Publication creates a scholarly record, and one of the fundamental characteristics of that record is supposed to be that it records what was published, including those occasions when what was published later turns out to be wrong. In many ways, this permanence is what gives meaning to the idea of a scholarly record.
This creates an interesting problem. If scholarly publishing is partly about creating a permanent record, under what circumstances should we allow somebody to remove part of that record?
Retraction does not usually mean deletion
It is worth starting with a distinction that is sometimes overlooked. A correction changes or supplements a paper when an error needs to be addressed but the work remains substantially reliable. An expression of concern warns readers that there may be a problem while an investigation continues. A retraction tells readers that the paper, or its conclusions, should no longer be relied upon. Importantly, however, a retracted paper does not normally disappear.
COPE’s (Committee on Publication Ethics) retraction guidelines describe retraction as a mechanism for correcting the literature and alerting readers to work whose findings cannot be relied upon. Crucially, COPE says that the main purpose is to correct the literature and preserve its integrity, rather than to punish the authors. Retraction notices “… should link to the retracted article, clearly identify it with title and authors, and be published promptly and be freely accessible to all readers.”[1]
That principle matters because an article may contain fabricated data, it may contain plagiarism or its peer-review process may have been manipulated. In these cases (and more) the article remains accessible, which is important as this, in itself, is part of the scientific record. The article may be flawed but the article remains available, albeit prominently identified as retracted.
The fact that the paper was published is itself part of the scholarly record. Deleting it would not alter history. It would simply make that history more difficult to see.
The scholarly record includes our mistakes
Elsevier expresses this principle particularly clearly in its current correction, retraction and removal policy. It says that published articles should remain “extant, exact and unaltered as far as is possible” and describes the scholarly literature as a permanent and historic record of scholarship.[2] It goes on to say “Given the importance of maintaining the scholarly record as a permanent and – as far as possible – unaltered record of the transactions of scholarship, removal of an article is rare.”
Springer Nature similarly notes that, immediately upon publication, the published version becomes the official record and is distributed onwards to numerous partners. Changes therefore have to be managed through formal post-publication processes.[3]
This matters because research does not develop by continually replacing yesterday’s literature with today’s preferred version. It develops through challenge, correction and sometimes rejection. An incorrect paper may lead to a correction. A fraudulent paper may be retracted. A disputed interpretation may provoke a commentary or another paper demonstrating why the original authors were wrong. Researchers examining the history of a field should be able to see not only the current status, but also how it got there.
The scholarly record is supposed to preserve our failures as well as our successes. That does not mean preserving them without comment. Quite the opposite. It means preserving the evidence while adding the information needed for future readers to understand what went wrong.
So can a published article ever be removed?
There are clearly circumstances where leaving the full text available may create risks that cannot adequately be dealt with through a retraction notice. Elsevier, for example, says removal should occur only in an “extremely limited number of cases”.2 Its examples include situations where an article is defamatory or infringes legal rights and retraction would not provide a sufficient remedy, where the article is subject to a court order, or where acting upon its contents could create a serious health risk. Even then, the title and author metadata are retained and readers are told that the article has been removed.
This seems a sensible principle. There are circumstances where continued access to particular information may itself cause serious harm. An article might unlawfully reveal confidential information about research participants, for example. There may be legal restrictions on continuing to distribute material. An article might provide dangerously incorrect instructions where simply labelling the work as retracted would not sufficiently mitigate the risk.
The important point is that removal is treated as exceptional. Once that is accepted, the more difficult question becomes how high the threshold should be before a journal moves from correcting or retracting the record to removing access to the published work itself.
How high should the threshold be?
Suppose a paper contains fabricated results. Retraction would normally be the appropriate response. Suppose substantial plagiarism is discovered, or serious errors mean that the conclusions can no longer be trusted. Again, retraction is likely to be the appropriate mechanism. If other researchers strongly disagree with the conclusions, the obvious scholarly response is criticism, rebuttal and further research rather than deletion.
More difficult cases arise when the objections are not about fraud or reliability, but about values, language, interpretation or perceived harm. What if a paper uses terminology that is now regarded as inappropriate? What if it conflicts with the values of the organisation that publishes the journal? What if members of a professional association regard its conclusions as offensive, or the publisher believes that continued availability could cause reputational damage?
These cases are much harder because “harm” can encompass very different things. There is an obvious distinction between an article exposing the identity of a vulnerable research participant and an article containing arguments that some readers find objectionable or distressing. Both might be described as causing harm, but it is far from obvious that they should justify the same publishing response. If removal is to remain exceptional, we need to be much clearer about the threshold that turns concern about a publication into justification for removing the publication itself.
Publication is not endorsement
There is another principle that becomes particularly important in these cases. A journal publishing a paper does not mean that the editor agrees with it, nor does it mean that the editorial board, publisher, university or professional association endorses every conclusion it contains. Peer review certainly does not provide a certificate saying that a paper is true.
Publication means, more modestly, that the journal has decided that the work meets its criteria for entering the scholarly conversation. Other researchers remain free to challenge virtually every aspect of it. Indeed, scholarly publishing would be considerably less useful if journals only published arguments with which their editors and owners agreed with.
That distinction becomes important if a journal is owned by a professional society, university or other organisation with its own mission and values. There is a difference between asking, “Is this work suitable for publication in this scholarly journal?” and asking, “Does this work reflect the values of our organisation?” If those two tests become indistinguishable, the journal starts to move from being a venue for scholarly discussion towards becoming an expression of institutional positions.
There may be publications for which that is entirely intentional, but if so, authors and readers should understand the basis on which material is being selected and retained. It becomes much more problematic when the distinction only becomes visible after publication, when previously accepted material starts to sit uncomfortably with the organisation that owns the journal.
What happens when values change?
There is also a temporal problem. Values change, as do terminology, ethical standards, research methods and social attitudes. We should welcome many of those changes because scholarship should not be frozen according to the standards of previous generations. The difficult question is whether the historical record should itself be rewritten to conform to current standards.
Imagine applying that principle retrospectively across a university library. There would undoubtedly be books, papers and arguments that we would not publish today. Some would contain ideas that modern readers find deeply objectionable, yet their existence still tells us something about the development of scholarship and society.
There is an important difference between saying, “We now reject this argument” and behaving as though the argument was never published. The ability to make that distinction is one reason why preserving the scholarly record matters. We can criticise the past without pretending it did not exist.
Who gets to decide?
There is a governance question. Who should have the authority to remove an already published paper: the editor, the editorial board, the publisher, the owner of the journal, a professional society, a university, lawyers or an ethics committee?
There will be circumstances where several of these groups legitimately need to be involved. A serious legal issue cannot simply be ignored because an editor wants absolute independence, and a research-integrity investigation may require expertise that the editor does not possess. Nevertheless, the decision is significant enough that the process should be transparent and governed by clear principles.
If a paper is removed, readers should be able to establish that something previously existed, why it has been removed, who made that decision and what process was followed. The alternative is particularly problematic. A DOI resolves differently (if at all), a PDF disappears and a citation points to something that readers can no longer inspect. Future researchers may have little idea what happened or why.
That is not merely a technical problem. It affects the transparency of the scholarly record itself.
This is not merely hypothetical
A recent controversy in New Zealand illustrates why these questions matter. A peer-reviewed article published in 2025 by the Journal of the New Zealand College of Clinical Psychologists subsequently attracted criticism, underwent corrections and was later removed. The College said retaining the article could perpetuate harm to M?ori and was inconsistent with its values. The case has prompted debate about academic freedom, editorial independence and the difference between retraction and removal. You can read more about this at Retraction Watch.[4]
I do not intend to comment on that dispute here. The underlying issues are complex and are not the subject of this edition. What interests me here is the publishing-integrity question and that this case demonstrates that removal is not merely a theoretical possibility. Journals and their owners can face genuinely difficult decisions about what to do when published scholarship becomes highly controversial, and the principles we apply in those situations matter.
More scholarship may be better than less record
Removing an article is not the only way that a journal can distance itself from previously published material. It can publish a correction, issue an editorial note, publish an expression of concern, commission commentary, publish rebuttals, invite the original authors to respond or, when the established criteria are satisfied, retract the article while retaining the original as part of the record.
These mechanisms do something that deletion cannot. They allow future readers to see both the original scholarship and what subsequently happened to it. Criticism can sometimes be more informative when the object of that criticism remains available because readers can inspect the original argument, read the challenge and make their own assessment.
Scholarship advances through precisely that process. The record becomes more informative when disagreement, error and correction remain visible, not less.
Preserving an uncomfortable record
There will be exceptional circumstances where removing access to a published article is justified, and I do not think an absolute prohibition on removal would be sensible. But the threshold should be very high, because the normal presumption should surely be that once something has formally entered the scholarly record, we preserve evidence that it was there. Even if something is deleted, a record should remain that this was done.
Essentially, if it is wrong, correct it; if it is unreliable, retract it; if it is disputed, debate it; if additional context is required, provide that context. Making published scholarship disappear should be different because it does more than tell readers that something was problematic. It removes their ability to inspect the thing that became problematic in the first place.
The integrity of the scholarly record matters most when preserving that record is uncomfortable. It is relatively easy to defend the permanence of papers with which nobody has a problem. The real test comes when we would rather they were not there, and perhaps that is precisely when we should be most cautious about pressing delete.
Did you find this edition useful?
I publish two free weekly newsletters exploring different aspects of higher education, research and scholarly publishing. If you enjoyed this edition, you may also find value in the following.
Governance, Leadership & Research
A weekly brief for higher education leaders and researchers examining governance, leadership and research in practice.
https://buff.ly/GhsZqVl
Publishing with Integrity
Examining scholarly publishing through an ethical lens, challenging assumptions and supporting academic career development.
https://buff.ly/WiG1yFb
About the Author
Professor Graham Kendall is Vice-Chancellor of GlobalNxt University, Malaysia, and an Emeritus Professor at the University of Nottingham. Over the past 25 years, he has published more than 300 peer-reviewed papers and has served as Editor-in-Chief and Associate Editor of several international journals. Through Publishing with Integrity, he explores the ethical, governance and practical challenges facing scholarly publishing, encouraging greater transparency, integrity and informed debate across the global research community.
[1] https://publicationethics.org/sites/default/files/retraction-guidelines-cope.pdf, accessed 25 Aug 2026
[2] https://www.elsevier.com/about/policies-and-standards/article-withdrawal, accessed 25 Aug 2026
[3] https://support.springernature.com/en/support/solutions/articles/6000281736-post-publication-journal-corrections, accessed 25 Aug 2026
[4] https://retractionwatch.com/2026/08/19/new-zealand-journal-clinical-psychology-maori/, accessed 25 Aug 2026