[The different models of scientific journals].

Introduction: Scientific journals are the main source of scientific data, ensuring their registration, validation, distribution and archiving. With over 2.6 million scientific articles published each year, the turnover of scientific journals exceeds $25 billion annually. Five publishers share nearly half of this lucrative market. Scientists are the key players in the process, but other stakeholders have gradually been introduced, building various business models whose similarities and differences are described here.

Concepts underlying scientific publication: Open access to scientific papers dates back to the scientific and technical revolution of the 17^th century. However, its evolution has been considerably boosted by the development of the Internet and the recognition of science as "commons".Scientific integrity is under the control of research institutions to ensure the prevention of fraud and misconduct in the course of scientific production. Usually, the scientific integrity is questioned during the manuscript reviewing process which may result in identification of flaws.

Models of scientific publications: In the historical model, readers pay for access to the document. Authors are not remunerated and renounce copyright on their articles to the publisher. The limits of the historical model became clear in the 90s, facing the cost of publishing, reduction in the number of subscribers, development of the Internet and willingness to improve manuscript evaluation.With the development of the Internet and the paradigm of open access, publishers proposed a new model in the 2000s, replacing the cost of access to articles for the reader with the payment of Article Processing Charges (APCs) paid by the author or its institution ("author pays" model). In this model, the content of the article can be freely reproduced and used, provided that the original author is credited. In addition to the evaluation of the manuscript which remains a critical factor, the cost of publication appears inequitable. However, all or part of the APCs may be waived, particularly for authors from low- and middle-income countries.For the past 15 years or so, publishers, learned societies and academic or research institutions (including libraries) have been seeking to publish reliable, open access manuscripts that respect scientific integrity while being affordable for the author.Predatory journals emerged in the late 2000s, taking advantage of the success of the authorpays model to capture APCs. Lacking a proper evaluation process resulting in poor-quality publications, these journals are rejected by most scientific institutions. On the other hand, they are particularly attractive in low- and middle-income countries because of their aggressive commercial practices (insistent invitation to submit a manuscript, low rejection rate, rapid publication, reduced APC, etc.).The purpose of each journal is to secure its economic model. This goes through ensuring its visibility, which is determined by the number of citations (online and social media citations) rather than the quality of the articles published.

Peer review: This very old concept has not been widely used until the 20^th century. In the historical model, manuscript evaluation is generally carried out by members of the learned society that publishes the journal. Evaluation can be either unblinded, single-blinded (referee is anonymous) or double-blinded (author and referee are anonymous). Several studies have shown that blind procedures do not alter the quality of the evaluation. Since the early 90s, post-publication evaluation has emerged, of which there are several variants. The aim is to shorten times to publication and open up the evaluation process more widely in order to limit the bias. Apart from the fact that this system does not guarantee a better evaluation of the manuscript, its main disadvantage is that the article is accessible without validation of the data collection and analysis throughout the entire process, which can be lengthy.

Cost and funding of scientific journals: The cost of an article depends on charges that vary according to the conditions and places of production. Reviewers are volunteers. On the other hand, manuscript management, editing and distribution are carried out by professionals, which entail financial charges. Some of these costs are lower in low- and medium-resource countries, where economies of scale and higher benefits are possible.

Conclusion: The limits of the historical model have led to the development of several business models of scientific journals, that are in constant evolution, especially the author pays model which promises open access to publications but impacts scientific production. However, the evaluation of scientific production is heterogeneous due to a limited pool of reviewers inadequately selected. Scientific publishing is looking for solutions to find a virtuous model that respects open science, open access to data and scientific integrity. The "Diamond open access" model, free of charge for both readers and authors with the guarantee of an irrevocable license to reproduce the content of the article provided that the original source is cited, perfectly fits.