TO REFORM SCIENTIFIC PUBLISHING, UNIVERSITIES MUST BE PUBLISHERS

Giving universities the benefits of, and responsibility for, their researchers’ published work

Scientific publication faces growing problems. It is too expensive. Fake and predatory journals abound. Paper mills and peer review cartels corrupt. Journal-based peer review fails to prevent publication of poor and unreliable research. AI is generating completely fabricated data sets and full manuscripts. Peer review is being overwhelmed.

Reforms are being tried; preprint servers, the eLife model, diamond open access, DORA and CoARA, better automated checking systems, but there are strong disincentives to change; publisher’s profits, and our own continuing loyalty to the publishers’ metrics.

The journal publishers argue that they provide essential quality control through their peer review processes and manuscript editing. There is truth to this but we also give it too much respect. When we say about a publication that “it’s been peer reviewed by (insert journal name)”, we imply that it has had its truthfulness and reliability checked by experts. We give it a stamp of quality control. But we fool ourselves. Unreliable research abounds nevertheless and recent studies have shown that few submitted papers are meaningfully changed during peer review (footnote 1).

A simpler system would be for research institutions to publish directly their employees’ research. Why should universities and researchers continue to make themselves dependent on third-party publishers at all, three decades after the invention of the World Wide Web, when there are few financial and technical barriers to publishing online (footnote 2)

Universities as publishers is not a new concept. In the past, many had their own printing houses to disseminate their academic works. The shift to commercial publishers only occurred in the last third of the 20^th century when university publishing houses struggled to keep up with the costs of printing and distributing paper journals and private companies saw opportunities for profit. Since the arrival of the internet however, there are no longer insurmountable barriers to universities publishing their researchers’ works.

If universities were directly responsible for publishing, it would more strongly focus them on their role as the guardians of the scholarship produced by their employees. They would be, (and should be), directly responsible for the validity, quality, integrity and the ethics of the published output of their researchers.

For this to work, they would need to have in place their own internal review and integrity processes. Some already do, requiring researchers to submit drafts for internal scrutiny and lodgement of primary material in accessible repositories before submission to journal. Most though do not take this role seriously and leave scrutiny of their employees’ manuscripts to the journals.

Scientific journals need not completely disappear. They could develop other business models, for example they could be the places where scholarly commentary and reviews of the primary published material are published. Scientific society journals could continue to collate and draw attention to important works in their professional field.

Why would universities do this? As well as to strengthen self-governance, for their reputation. Institutional prestige flows from the impactful, highly regarded research published by their staff and this prestige attracts students, attracts government and philanthropic funding, attracts patients to university associated medical centres, attracts commercial partners. Currently and counterproductively, the prestige of breakthrough research is transferred out to the journals that publish the work. With university hosted electronic journals, readers would be taken to the university’s website, automatically expanding the brand of the university doing the research. Universities would avoid paying authorship fees to others and there would be the same potential for additional revenue that now resides with the journals, such as advertising or pay for access models.

University direct publishing could help currently less prestigious or well-known institutes to compete directly with the traditionally more powerful. In the current journal system, decisions are heavily swayed in favour of more prestigious and better-known universities and investigators through their reputational influence (this is the Matthew effect, “To those who have, more will be given”).

One of the immediate objections will be that universities cannot be trusted. But direct university publication of papers should help here too. To protect their own reputations, they will need to be vigilant about what their researchers publish. The research journal of a university that publishes low quality or fraudulent work would rapidly lose its reputation and damage the university’s credibility. Forced retractions would have a long-lasting negative impact.

There are plenty other issues that would need to be worked out, such as changed roles for the preprint servers and indexing services, and where to publish when co-authors are from multiple institutions? None of these seem insurmountable to us; for example, multiple institutions could each publish a common agreed version and identifier.

The greatest barrier to any change may be we scientists ourselves. We can feel that we have little alternative to the current publishing system, despite its problems, because our careers are largely determined by the perceived prestige of journals and by the almost universal overt or covert use of publishers’ citation metrics. The result is that unaccountable third parties have non-contestable powers over our careers. Returning to a university publishing model would give universities stronger self-governance and give researchers a more direct role in determining what is published. Researchers’ resistance to change would change quickly if some of the leading universities were to start publishing and funders changed assessment policies.

Universities employ the people that do the research, in facilities that they provide. They should therefore also publish the research. Why hand it over to a commercial body?

Paul M. O’Connor is a Professor in the Department of Physiology, Medical College of Georgia)

Footnotes (quoted from here)

Footnote 1.

Klein and colleagues tested academic publishers’ claims of the value they add to scholarly articles. They used four different metrics to compare ‘similarity’ between the preprint and published versions of almost 10,000 full published articles with their preprints, and almost 1500 full text articles with preprint versions. They found ‘that the text contents of the scientific papers generally changed very little from their pre-print to final published versions. Looking specifically at a subset of the preprints that were first posted on bioRxiv, they found ‘…that bioRxiv pre-prints and their final published versions do not exhibit many differentiating features, neither on the editorial nor on the semantic level.’’ Klein, M., Broadwell, P., Farb, S. E., et al. (2018). Comparing published scientific journal articles to their pre-print versions. International Journal on Digital Libraries, 20(4), 335–350. https://doi.org/10.1007/s00799-018-0234-1



7 comments on PubPeer (by: Timothy H. Vines, Anolis Jacare, Hoya Camphorifolia, Strabena Modesta, Lyophyllum Pulvis-horrei, Polybia Scrobalis, Pandanus Peyrierasii)

Brierley and colleagues examined publications in the early stages of the COVID pandemic, comparing the abstracts in the preprint and published paper versions. They concluded ‘Our data analysing abstracts suggests that the main conclusions of 93% of non-COVID-related life sciences articles do not change from their preprint to final published versions, with only one out of 184 papers in our analysis contradicting a conclusion made by its pre-print’. ‘…our data provides confidence in the use of preprints for dissemination of research’. Brierley, L., Nanni, F., Polka, J. K., Dey, G., Pálfy, M., Fraser, N., & Coates, J. A. (2022). Tracking changes between preprint posting and journal publication during a pandemic. PLoS Biology, 20(2), e3001285. https://doi.org/10.1371/journal. pbio.3001285

Janda and colleagues examined ‘547 clinical studies that were initially posted to medRxiv (in September 2020 during the pandemic) and later published in peer-reviewed journals’. They found that ‘Overall, 526 pairs (96.2%) had concordant study interpretations’; that is the journal review and approval process had not changed the overall conclusion. This held even when there was a difference(‘discordance’) in the primary end point results between the preprint and the later publication. They wrote that ‘This study suggests that most clinical studies posted as preprints on medRxiv and subsequently published in peer-reviewed journals had concordant study characteristics, results, and final interpretations’. Janda, G., Khetpal, V., Shi, X., Ross, J. S., & Wallach, J. D. (2022). Comparison of Clinical Study Results Reported in medRxiv Preprints vs Peer-reviewed Journal Articles. JAMA Network Open, 5(12), e2245847. https://doi.org/10.1001/jamanetworko-pen.2022.45847

Footnote 2.

Ozonoff has pointed out that research articles are not truly coupled any longer to a journal,141 writing that we have had ‘a seis-mic shift in scientific publishing and it has happened in a relatively short time without most of us being conscious of it. Just as music streaming services uncoupled song tracks from the record album or CD upon which they originally appeared, no-cost search engines like Google or the biomedical research database PubMed have uncoupled individual research articles from the journals where they originally appeared’. With PubMed, Medline and Google Scholar the journal of origin becomes irrelevant if we just want to read the paper – but only too relevant when the main point is journal prestige and career advancement.