An attempt to explain the partial 'silent' withdrawal or retraction of a SAGE Advance preprint

In recent years, there has been a rise in the use and acceptance of preprints, as well as in the availability of preprint servers to which academics can post their initial or crude intellectual ideas and work, advancing and speeding up the open and wider dissemination of information^[1−4]. Despite this, there is still concern that preprints might be a source of unreliable information, even though they are labelled as not being peer reviewed^{[5, 6]}. That is a valid risk worth considering, especially when referring to studies pertaining to human health, as was evidenced in the COVID-19 pandemic, although misinformation in preprints was almost also present in their peer-reviewed counterpart papers^[7]. How then can unscrutinized work enter the academic literature stream so easily if it poses any risk to the integrity of the information and knowledge stream? The answer is three-fold, in my view: First, and a pragmatically simple structural problem, the logic of preprint servers is to allow crudely vetted papers to pass a quick screening procedure before being posted online within a few days, although there is wide variation in moderation policies between available preprint servers^[8], as well as inconsistencies in the implementation of such policies^{[9, 10]}, although a greater consistency was noted for policies related to the acceptance of opinion papers^[11]. Second, preprints are being aggressively branded and marketed as a solution to the replication crisis^[7], placing pressure on journals and publishers to accept them as part of the publication process. Third, preprints are not labeled as duplicate publications, i.e., they are being offered exceptional ethical status.

There are some caveats to the positive factors of preprints that are being branded. For example, preprints, when existing alongside peer-reviewed papers, may offer additional citations for the same intellectual content, potentially – but not always – representing a citation advantage^[12]. Another problem is the incompatibility between the lack of anonymization of preprints in which authors' details are identified and anonymization requirements of peer-reviewed journals^[13]. A third problem is the existence of linguistically degraded terminology or jargon, referred to as 'tortured phrases'^[14], in preprints^[15].

It is thus good that more scrutiny is being paid to the weaknesses of preprints and preprint servers because they are being used as a mechanism to draw intellect to peer-reviewed journals, resulting in gains for the publisher, either as copyright and subscription-related sales, or as open access (OA)-related APCs. ASAPbio maintains a list of preprint servers that mostly claim the permanence of the published record, when a preprint is assigned a digital object identifier (DOI), which is recognized by Crossref^[8].

Readers' attention is drawn to a disturbing aspect of intellectual record-keeping that involves the opaque and partial or fully 'silent' withdrawal or retraction of peer-reviewed papers^[16] or preprints^[17], even more so when they are associated with a DOI. This is because the DOI supposedly represents a permanent form of archiving information for scholarly posterity. When a 'skeleton' trace of a withdrawn or retracted (these terms are hereafter treated as equivalent procedures in this paper) preprint can be identified, then there is still the opportunity to hold the authors and/or publisher (i.e., preprint server) accountable. The more serious case is when papers or preprints are completely deleted from the academic record, not allowing any information to be drawn from vestigial websites as to how, why, or when a preprint might have been retracted. The author is of the opinion that 'silently' retracted preprints (and peer-reviewed papers) thus reflect extremely poorly on preprint servers, in the case of preprints, and on journals and their publishers, in the case of peer-reviewed papers, because they reflect opacity (or the lack of transparency) by the preprint server or publisher. In addition, they call into doubt the veracity of due process and ethical processing that was involved in the removal of such DOI-indexed information, and they leave a bibliometric gap in the knowledge stream that disrupts and induces doubt on the integrity of information retrieval, storage, and management, aspects that are fundamental to information integrity, such as applied to citations^[18].

This topic is thus of particular interest to academics involved in library and information science, as well as science integrity. These processes are supposed to occur transparently in journals and publishers that are members of the Committee on Publication Ethics (COPE), which has established a pre-determined set of guidelines for retractions (i.e., withdrawals) that is supposed to be followed by its members^[19]. Even though – surprisingly – none of the ASAPbio preprint servers are COPE members, it has been argued that both preprints and peer-reviewed papers must be observed with the same ethical rigor and scrutiny, by treating them as 'equals'^[20]. Even more so for publishers that are COPE members, such as SAGE Publishing, that have their own preprint servers. In the case of SAGE, it is Advance (https://advance.sagepub.com/), which has a set of submission guidelines, including some broadly defined ethical guidelines (https://advance.sagepub.com/f/submission-guidelines), and also terms of use (https://advance.sagepub.com/f/terms)

A set of comments (www.pubpeer.com/publications/EB6B06831411E6EC13E53589CC84EC; www.jamespowell.org/Resignation/) surrounding a peer-reviewed paper^[21] published in a SAGE journal title, Research Ethics, provided clues that a preprint existed for that paper. However, surprisingly, a search at Advance revealed no such preprint for this author (https://advance.sagepub.com/search?q=James+Lawrence+Powell). After having been sent (by Dr. Mark Boslough) a Word document of the preprint and a link of the original URL where the preprint was originally housed (Two preprint URLs carry its title 'Sodom and Skepticism', although only the second one is relevant: https://advance.sagepub.com/articles/preprint/Sodom_and_Skepticism/193800; https://advance.sagepub.com/articles/preprint/Sodom_and_Skepticism/19380077/1), it was deemed that this preprint had been 'silently' retracted, and in place of all bibliometric information (author's name and affiliation, email, PDF file of the preprint, HTML text such as the abstract, DOI), only an opaque note can be found that states: 'sorry, this page is no longer available. This content has been intentionally removed or had its access disabled. Reason: The paper has been removed from Advance'. (Fig. 1a). No precise reason for the retraction has been provided, and no publication and retraction dates appear anywhere on that Advance preprint's page. Were this to be a peer-reviewed paper in a COPE member journal or of a COPE member publisher, this preprint's retraction would not be COPE-compliant^[19]. A search at Dimension.ai revealed no publication entry or DOI for the preprint (https://app.dimensions.ai/discover/publication?search_mode=content&search_text=Sodom%20and%20Skepticism&search_type=kws&search_field=full_search&or_facet_publication_type=preprint), and even though an entry at Crossref revealed that the preprint had been published on March 18, 2022 (https://api.crossref.org/v1/works/10.31124/advance.19380077.v1), simultaneously revealing the DOI (10.31124/advance.19380077), there is no information about its retraction. The DOI, when entered at Scite.ai, confirmed the existence of this preprint ( https://scite.ai/reports/sodom-and-skepticism-zRg6lK03). Moreover, the retracted preprint and the author are not indexed on the Retraction Watch Database^[22], even though they should be. ASAPbio makes three pertinent notes about Advance preprints: indexed at/by 'Google Scholar, Crossref', 'Content within scope, text overlap detection, ethical compliance, legal compliance', and 'Permanent with some removal options in exceptional circumstances'^[8]. Even though the Powell Advance preprint^[23] was published under a CC BY 4.0 license, no OA document could be identified at its 'skeleton' entry at Google Scholar (Fig. 1b).

Figure 1. 

(a) Message on page of a silently retracted SAGE Advance preprint (screenshot date: 24 March 2023): https://advance.sagepub.com/articles/preprint/Sodom_and_Skepticism/19380077/1. The stated reason is in fact not a reason. (b) No open access file could be identified on Google Scholar (https://scholar.google.co.jp/scholar?hl=en&as_sdt=0%2C5&q=Sodom+and+skepticism+powell&btnG=), despite the CC BY 4.0 license, as documented at Crossref. Screenshot date: 2 July 2023.

After having sent a formal request for an explanation to the SAGE Publishing Advance Editor, Social Science Journals, Julia Slater, on January 2, 2023, with a reminder on January 12, 2023, a response ('We are currently reviewing the editorial practices and standards around preprints published on our preprint platform, Advance, including withdrawals and the possibility to retain the metadata for such papers once they are removed from the website. Due to some technical limitations on our preprint platform, metadata is not currently retained for withdrawals, no matter the reason of the withdrawal. We are already looking into improving this aspect of our platform in order to bring preprints more in line with best practices recommended for peer-reviewed research and scholarly content in general, which we support and adhere to as a member of COPE. As part of this review, in due time we will update the Author guidelines published on the website at the link you mention below'. That website is the second URL in footnote 2 in this paper.) was received on 24 January 2023 from SAGE Advance, but signed by the 'Advance Preprints Team'. In addition, on 4 February 2023, Dr. Boslough sent a link to a copy of the now-retracted preprint that had been published, but which is hosted on an independent platform (https://cosmictusk.com/wp-content/uploads/2022/12/ReplytoBosFINAL.pdf; https://web.archive.org/web/20230203205657/; https://cosmictusk.com/wp-content/uploads/2022/12/ReplytoBosFINAL.pdf). Despite the formal confirmation by SAGE Advance, neither the preprint nor any of its bibliometric parameters have yet been reinstated (Fig. 1a).

In the absence of the proper processing of information, and the accurate recording of changes to the bibliometric scholarly record, science cannot progress efficiently. Even though the Advance preprint guidelines state that a preprint can be removed, when this process takes place opaquely and when all information, including bibliometric identifiers, are scrubbed clean from the preprint's URL, this reflects poorly on the ability of the preprint server to accurately retain information about the public record, and may even constitute – in the author's opinion – a form of tampering or manipulation of the literary and academic record, independent of the reason. Of greater concern, the preprint server's publisher is SAGE, a COPE member publisher, so greater transparency would have been expected. Both SAGE and the author of the preprint (James Lawrence Powell, University of Southern California) were contacted (January 2, 2023) to provide an explanation regarding the removal ('silent' retraction) of that preprint, why all bibliometric information was also removed, and why no transparent explanation has been provided to the public. Unfortunately, to date, no response was received nor was any explanation provided by Powell.

Usually, when a peer-reviewed paper is retracted, the retraction notice also has a DOI assigned to it^[24], but in the case of 'silent' retractions, these occur opaquely, and the existence of a DOI-registered retraction notice is missing. There is something akin to a crisis regarding the lack of transparency and background details that appear in many retraction notices^{[25, 26]}. The lack of indexing of this preprint and the author at the Retraction Watch Database^[22] also suggests that there is a problem with indexing at Crossref, and a failure to transmit retraction-related information automatically to this database, if indeed this is a mechanism by which metadata related to retractions is gathered for this database.

Preprints that have been transparently retracted, i.e., not 'silently', have established guidelines and reasons are reasonably well documented^[27]. Even so, authors are usually left to their own devices when assessing the advantages and risks of posting preprints^{[28, 29]}. This case revolves around the integrity of published literature, whether it be peer-reviewed, or not (as in preprints). It also focuses on the integrity of the process surrounding bibliometric indexing of published and retracted/withdrawn literature. Considering these aspects, the following advice is suggested, advice that could form the basis of much needed and more concrete guidance, following debate by experts, peers, and policy groups such as COPE, the ICMJE and the Council of Science Editors, who publish white-papers and guidelines and/or recommendations related to publishing ethics, documents that are frequently updated:

1) There needs to be formal recognition of the phenomenon of 'silent retractions', for preprints and peer-reviewed papers; without formal recognition of this phenomenon, solutions cannot be found;

2) Dual terms (retraction; withdrawal) for the same phenomenon, namely the removal of a paper or document from the literature, should be standardized or treated as equal terms; using two terms with the same consequential result will cause confusion among academics;

3) Preprints must be treated as 'ethical' equals, i.e., whatever consequences apply for ethical infractions of authors of peer-reviewed papers must apply as equally stringently to preprints; having a dual-level of ethical principles, even though both preprints and peer-reviewed papers might be published by the same publisher, in this case SAGE Publications, sends the wrong signal to academia, and induces confusion, frustration and an unfair 'playing field';

4) Even if the text is removed because it contained threatening language, insults, or other text that is the subject of legal threats and legal proceedings, the publisher or owner of the preprint server have the responsibility of not merely deleting all background information, but to leave basic bibliometric elements public and for posterity (e.g., authors' names, title, abstract, DOI);

5) No matter the reason for the withdrawal/retraction of a preprint, and independent of its removal by the author or by the publisher or owner of the preprint server, that/those reason(s) need to be fully, transparently and clearly stated; at the end of the day, information cannot simply disappear without transparently stated reasons;

6) Authors and publishers who fail to follow guidelines – when these are eventually created – should face punitive consequences, as equally as those who make ethical transgressions in the publishing enterprise;

7) Guidelines should be implemented retrospectively, i.e., 'silently' retracted preprints should be properly indicated as such for all such removed preprints, and since the first published preprint, by all preprints servers that are indexed on websites like ASAPbio, which serves as some sort of an 'industry standard'.

The author thanks Dr. Mark Boslough for providing useful background regarding this preprint, formal confirmation of its existence, as well as a link to where an independent copy exists and has been archived.