What direct support is available for open-access Diamond journals? Funding models and arrangements for implementation

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.


Introduction 11
From the BOAI to the OA Diamond Journals report: 11 direct funding of journals for open-access publishing 11 Understanding the funding of Diamond journals more clearly 13 The starting point was the Diamond journals' financial situation. 17 Report outline: from models to implementation issues 19 The team, funding sources, and survey context 19 Funding models for Diamond journals What are these models funding? 31 Exploring publishing processes  Table 5. Benefits and pitfalls of direct funding of Diamond journals 82 Table 6. The five main themes of the questionnaire 84 Table 7. Payment to entities for publishing acts 87

Background
In 2020, we participated with the OPERAS consortium in the OA Diamond Journals study, the purpose of which was to quantify and qualify the worldwide Diamond journals ecosystem, focusing on journals and platforms located in Europe and South America. One of the many recommendations in the final report was the proposal to establish direct funding for Diamond journals by institutions which do not currently provide support. This work follows on from the recommendation. It complements the work done in 2020 by focusing on a specific aspect of the Diamond ecosystem, namely the ways in which they are funded. The aim in exploring in detail the current funding options for Diamond journals is to identify the specific forms that permanent funding models could take.

Method
This study is mainly based on a questionnaire survey which was sent to over a thousand

Four direct funding models for Diamond journals
Our questionnaire survey identified the various forms of support available. In addition to the development of publication infrastructures and the availability of services, we also focused on direct funding models. Our research has led to the creation of four models that are, to a greater or lesser degree, linked to the publication yield and are based on annual funding: • The white list. Journals meeting the criteria that make them eligible for inclusion in a funding list, would obtain access to a fixed sum of money, irrespective of the content of the articles and the publication levels or yield.
• The threshold. A set amount of money is allocated to the journal, subject to a minimum number of publications that are of interest to the funder.
• The slots. The sums of money paid to the journal do not depend linearly on the number of articles which are of interest to a funder, but on "publication tranches", within which the amount is fixed.
• The yield. A sum of money is allocated to the journal on the basis of the annual publication yield, in line with the notion of "fund as you publish" as used in some transformational agreements.

Who acts in which acts when a scientific journal is published?
In order to enable us to look in detail at the publication process, we have put together a list of 26 acts which delineate the production of a scientific article within a journal. These 26 acts are grouped into 3 major categories: certification (reception, assessment, response to author, etc.), physical production of the document (copy editing, proofreading, coding/conversion, etc.), dissemination (rights/contracts, assigning a DOI, assigning metadata, dissemination, archiving, etc.) By using a questionnaire survey, we were able to identify patterns within the Diamond journals that we investigated.
• The editors and their assistants "cover" all of the publication acts: they oversee all acts and execute all or part of most acts. They take the lead in the certification process (which is shared with members of the editorial board) and in the acts related to dissemination.
• The sub-editors and external service providers contribute in more specific ways to the physical production of the document. To a lesser extent, they also take part in the dissemination acts.
• Certain specific acts involve two entities: external assessors are the most involved (45% of respondents for reviewing, whereas the use of software is quite evident when used for checking plagiarism (plagiarism check, 22% of respondents) and assigning a DOI (19% of respondents).

Which acts in the publication process could be funded?
The survey highlights the importance of unremunerated work within the journals, but goes only as far as underlining all of the tasks involved. This situation is characteristic of scientific publishing, where scientific staff are paid by the institution they work for to carry out all the work that their research covers. Thus the work carried out in a journal is not subject to additional remuneration, as it is considered to be part of the work done by employees working in higher education and research (for which they already receive salaries). This peculiarity of academic journals is even more marked for Diamond journals, as suggested by the OA Diamond Journals survey. Additionally, certain acts in the publishing process that are performed by external service providers, are indeed subject to financial transactions. In this respect, this work allows us to identify direct funding focus areas: • The acts involving certifying manuscripts have the lowest percentage of monetary transactions. Even in an ideal world without financial constraints, Diamond journals would be fairly unwilling to finance this part of the publication process, in particular because they wish to retain their editorial independence in scientific publishing.
• The second category of publication acts relating to the physical production of documents, involves the highest monetary transaction percentages. It's also the category for which journals are the most willing to pay in an "ideal" situation" free of all financial constraints.
• As regards the dissemination acts, some of the journals are in favour of remunerating some of them.

The technical conditions for implementing direct funding models.
Regardless of the model selected, a certain number of technical conditions are required for the implementation of direct funding: • The ability to carry out monetary transactions ("transactional capability"). Around 80% of the journals surveyed state that they are capable of accepting money, either directly or indirectly. This percentage increases to 86% if potential capabilities are taken into account -the willingness of journals to implement an accounting system provided adequate funding is available.
• The ability to make funders visible. 51% of the journals surveyed already have the means to create reports in order to specify the contributions made by funders to the publication in question. Of those who are not currently able to do so, 73% state that they are ready to adopt a system of this kind, provided there is adequate funding.
• The regulatory opportunity for research funders to provide direct support for journals.

Benefits and limits of direct funding models.
Through the use of open questions, we were able to identify the benefits anticipated by the journals which would come from direct funding, as well as the potential pitfalls to be avoided.
These elements are summarised in the table below. recommendations for developing and perpetuating these journals within the scientific publishing economy.
The main findings of the report consist of the identification of a vast universe of potentially close to 30,000 journals, only a third of which were referenced in the DOAJ. In contrast to the commonly-held view, even if SSH (Social sciences and Humanities) disciplines are in the overwhelming majority, STM (Science, Technology, Medicine) disciplines are well represented (39% of the journals identified). Unsurprisingly in this world, there is much greater linguistic diversity than in the APC model journals. As regards the subjects at the heart of our own study, the report emphasizes the importance of those contributions to the publishing process that are not directly remunerated, thereby exposing the diverse nature of the financial contributions required in order for them to operate: grants, donations, crowdfunding, shared infrastructures, the institutional support model, and the freemium model. This is overwhelmingly a "small-scale economy", with the majority of journals having less than one full-time employee, and 70% of them posting less than €10,000 in costs per year.
Although research and teaching organisations have a key role to play in supporting the Diamond ecosystem, this currently turns out be much less so for research funders.
Based on results like these, recommendations for support beyond the funding of production, dissemination, and archiving, include the implementation of direct funding models for journals produced by diverse players in the research community. Funders are specifically targeted as sources of direct support. Although we have just reiterated the role they play at the heart of the APC system, they are today almost completely absent from the Diamond ecosystem. Our current work is in line with this recommendation because we declared in the OA Diamond journals report our desire to carry out this additional survey. It therefore constitutes a second study of the Diamond journals, and focuses on their funding arrangements. Based on a questionnaire survey, we will look at direct funding models and reflect on the practicalities of implementing them.
Although they have not been systematically documented, many different kinds of direct support for Diamond journals have clearly emerged following the BOAI. In addition to infrastructure support, the OA Diamond Journals study showed that the most frequent form of support was either institutional (through the provision of personnel), or participatory (through participation of the publishing entity in the production and dissemination of the journal instead of direct financial compensation). However, the payment or making available of monetary sums by authors has also been described -even if this was done voluntarily.
Moreover, some respondents to our survey didn't fail to point out the existence of direct funders. These are generally public entities (research institutions, universities), sometimes referred to by name (such as the Swiss SNF or 3 Canadian agencies). It appears that this financial -and not only human and institutional -support has had a limited impact on the Diamond ecosystem, notably because these supporting parties are attached to an institution, or a discipline, or at best a national institution, without any overall coordination despite initiatives like the Global Sustainability Coalition for Open Science Services (SCOSS). Following on from the 2020 survey, we are therefore suggesting ways of organising and mainstreaming direct funding models to which all Diamond journals could potentially apply, in the same way that APC-type journals are able to benefit almost unconditionally from research project funding.
One of the major problems confronting recommendations for financial support is the lack of knowledge of the administrative, accounting, and financial situation of the Diamond After cleaning the dataset and deleting duplicates, we have gathered a total of 260 usable responses. As in the OA Diamond Journals survey, we received responses coming from a wide range of journals because 55 countries were represented, with the UK, Italy, and France dominating. Similarly, one can run through a wide range of disciplines -from the humanities to information technology, encountering chemistry and medicine along the way. Nonetheless, the results obtained allow us to identify certain trends. In the appendix to the report, you can consult the descriptions on how the questionnaire was developed, how the data was gathered and processed, and how representative the responding journals were.

The starting point was the Diamond journals' financial situation.
The initial survey questions covered the range of current situations in which the Diamond journals are operating. The OA Diamond Journals report established that the main physical difference between the OAC Diamond Journals and the APC journals was the virtual absence in the Diamond population of very large journals publishing huge numbers of articles.
Unsurprisingly, the journals that responded are "small" ones, publishing a maximum of 288 articles a year (an average of 31 and a median of 22 articles), with a very low estimated annual budget (75% below €10k), and a very limited number of paid employees (85% with less than the equivalent of 2 people in full-time employment). The first two results of our survey should be read with this general economy in mind; that is small journals with limited budgets.
We note that the journals lack financial autonomy when they are included in, or are linked to an entity via an institutional affiliation (figure 1). In this respect, they are no different than the great majority of non-Diamond journals, whether they belong to a scientific society, a university press, or a commercial publisher.   Turning journals into a commodity is clearly a foreign concept in the world of Diamond journals of our survey, as only 2% of them have the aim of making a profit. A little over a quarter of respondents (26%) say that they are unaware of this, which reflects the highly heterogeneous nature of organisations within the Diamond ecosystem. This makes them ideal candidates for direct funding support from the point of view of funders who are seeking to ensure that their money is actually used for publishing.

Report outline: from models to implementation issues
On the basis of this assessment of the Diamond journals' financial situation, the report is This report concludes with the benefits associated with direct funding models for journals, as well as the challenges and problems that they may elicit. The appendix to the report examines the survey methods, specifically the way in which the questionnaire was drafted and circulated, as well as the ways in which the data was analysed.

The team, funding sources, and survey context
We are researchers from the Centre de Sociologie de l'Innovation (Centre of Innovation General models for funding publishing.
In this first section, we organise and analyse proposals regarding the general publishing support models from the Diamond journals themselves. We use the term "support" because all of these models are not necessarily financial in nature. Among the answers to our question

Providing support for publishing infrastructures
The first model therefore concerns supporting infrastructures -whether they be software or distribution platforms. We would like to point out that tools of this type already exist, such as software that exhibits OJS functionality, or platforms such as the Amelica/Redalyc, OpenEdition, Episciences, or OLH, which integrate software.

Making services available
The second model is about making services available. It may involve making staff available as FTE (full time equivalent) workers by public entities, or a "capacity centre" like the one that was highly recommended at the end of the OA Diamond Journals study.

Direct funding of Journals
We have identified a third model whereby funds would be directly allocated to the journals.
The variety of one-off sources of funding identified by the 2020 survey was already impressive, but when the journals are asked about new funding options, they reveal that these options can be applied in very diverse ways. This diversity can be broken down between models which are uncoupled from the actual yields published -by far the ones most frequently mentionedand coupled models.

Models that are not coupled to the publication yield
Models which are uncoupled from publication yields include, first and foremost, self-financing by the journals themselves through advertising, fundraising, or voluntary donations. Obviously that is not sustainable, but I think it would be a good source of funds ».
Next, the journals report on receiving fixed sums of money -a yearly, twice yearly, or monthly amount, thereby promoting budgetary predictability. The third arrangement consists of a variant on the previous example: receiving a fixed amount based on applications for subsidy (known as "grants"), which would guarantee financial stability for the journal for a number of years (from 3 to 5 years). Finally, some propose more vaguely the awarding of grants based on visibility or recognition criteria, somewhat corresponding to a subsidy system.

Models coupled to publication yield
As well as these models, other respondents suggest models that are coupled to the amounts published, i.e. models where funding partly depends on the number of articles published.
Therefore, there exist funding proposals which dispense funding in proportion to the publication yield posted by a reporting system. This is a much-mentioned borderline case between authors being obliged to pay (subject to different exceptions or waivers), and voluntary contributions from the authors to the journal's general finances if they are able to do so, coming for instance from the funding of a project.
In our view, the above-mentioned redistribution model falls outside the Diamond framework because some authors are required to pay. 27 Section 2 Four direct funding models We have referred to proposals for Diamond journals in connection with models that provide publishing support. In line with the OA Diamond Journals study which was published in 2020, we intend to focus in particular on the third proposal, which is direct funding. The aim of this approach is not to underestimate the contributions made by other means of support, but rather to develop the "direct funding" sector. In this section, we draw on the survey responses together with our prior work in order to organise and refine the direct funding models. The refined funding models should be in a better position to address the issues of predictability and lack of certainty (2.1). In the end, we propose four models where the link with the actual publication yield varies (2.2).

Funding of scientific publishing: from predictability to uncertainty
Our previous survey examined the "transformative" agreements. These agreements were signed from 2015 onwards between national consortiums of university libraries (at least in Europe) and publishers 12 . What distinguishes "transformative" agreements is that they bring together two fairly different approaches to research publishing within the same contract: the classic subscription-based approach providing access to content for a limited audience against a paid fee; and more recently, an open access approach where, by definition, access is free of charge but producing articles is charged to institutions in the form of publishing costs. Two lessons can be drawn from joining these two worlds via transformative agreements.
First of all, a confrontation between two major financial configurations can be identified within the "transformative" agreements. On the one hand, there is the configuration that pursues complete predictability in the amounts exchanged by both parties.
This originally comes from the print model, where a specific number of issues or articles is bought, with the material costs and the amount of labour being strongly correlated to this amount. This is characteristic of the way subscriptions work. On the other hand there is the "complete uncertainty" configuration, which is characteristic of open access (in the gold-APC variant), where a reduction in publishing services dragged the concomitant costs down to those of an individual article, ultimately making financial models entirely dependent on the amounts published. The discussions around PLOS ONE and its incredible success (with a peak of over 30,000 articles per annum that was followed by a reduction in the amount published) have clearly shown the financial consequences of uncertainty in the yield. In a similar vein, suspicions that some publishers would be prepared to sacrifice their editorial integrity in favour of the profit generated by accepting an article -rather than rejecting it and incurring costs but no revenue -have highlighted the risks associated with the financial incentives to publish. We should also remember here that the growth of the "predator publisher" category stems from the existence of the APC economic model. One of our respondents highlighted one risk that is inherent in payment per article: "I do not think individual articles should be supported, as this may lead to Diamond predatory journals". The Diamond journals' responses to this survey by sometimes proposing fixed funding and sometimes models which are coupled to the amounts published reflect this confrontation between predictability and uncertainty. We will propose direct funding models based on all of this.
Second, rarely do the "transformative" agreements present a head-on confrontation between predictability and uncertainty. Novel ways of arranging these financial configurations are usually found. Put otherwise, the agreements we have studied exhibit a continuum from complete predictability to complete uncertainty. Building on this continuum concept, we have put together, not two, but four direct funding models for Diamond journals. These models represent a refinement of the binary confrontation between predictability and uncertainty, specifically because they provide the option of progressing from one to the other. We will see that they are partly inspired by original ways of structuring predictability and uncertainty that we identified in several transformative agreements.

Four models: link to publication volumes and managing uncertainty
The four models we have constructed propose a continuum ranging from predictability to uncertainty, with the model increasingly coupled to the published yields. We assume an annual payment once the journal is published for two reasons: from an administrative point of view, we wish to limit the number of transactions -costly for the funder and the journal; from a political point of view, we want to clearly distinguish this kind of support for Diamond journals from the APC model, where payment is demanded, in advance, for each article.
Obviously it is possible to envisage more regular support, depending on the amounts of money involved or other constraints. each tranche has a fixed price. Using this model for Diamond journals, the highest tranche would be the maximum amount of support from a funder, which is granted if the publishing yield exceeds a certain level and if predefined interest criteria have been met. There is a higher level of uncertainty here than with the previous models but it remains framed by the tranche system and the existence of a maximum tranche representing the maximum amount of support for a given journal from a funder.

B2 The yield
This involves the journal totalling the number of articles of interest for a funder and receiving in return an amount of money that depends linearly on this number of articles. This makes it a direct equivalent of the APC for a Diamond journal. As an example, Amsterdam University allocates a given sum to all DOAJ journals publishing articles signed by its lecturers. There is a maximum level of uncertainty here because it depends on the actual publication yield. This configuration is close to the "Pay as you publish" model that we have established for transformative agreements. In the context of the funding of Diamond journals, we find ourselves in a symmetrical situation of "fund as you publish".

***
Obviously, it is possible to have other models than those mentioned here, particularly all of those which derive from a "funding" model reserved for Diamond journals where the annual sum is set in advance. In the same way as all funds already devoted to APCs, they apply rules of the "first come, first served" type or, conversely, calculating an allocation based on realised publications. In this sense, although they guarantee maximum predictability for funders who grant a total lump sum, they set journals against each other or, at least, make the amounts they actually receive uncertain.
-Part II -What are these models funding?
In the previous section, we presented the general publication support models. By cross-referencing responses to our questionnaire as well as elements from -which are likely to be "billed" by the journals. Although the focus here is on Diamond journals, the issue under discussion in this section is a more general reflection which applies to all scientific journals: the series of acts required to produce and circulate articles.
In a first part, we suggest a way of specifying the journal publication process by identifying a set of discrete acts and the entities which are likely to perform them (3.1). To do this, we have drawn up two lists (one with the acts and one with the entities) for which we indicate the methodological safeguards to be adopted and the limitations that arise. In the second part, we present the results of our survey using a breakdown that shows how different entities share the work -how different entities share execution of the Diamond journals publishing acts. (3.2).

A list of acts
The first list covers the publishing acts which are usually performed to ensure the publication of scientific articles in a journal. We have compiled it based on literature describing how scientific journals function. Although this literature is scarce, it does offer several taxonomies for the tasks performed within a scientific journal. The most extensive task consists of 102 operations 15 . Our work is based on these propositions, as well as our experience of the way in which several journals function internally (with two team members being part of the editorial board). The list of operations that we have put together has a very specific objective: to identify the very clearly circumscribed acts to which research funders can consider allocating funds.
In this respect, the list has several specific features. Firstly, it is designed for open access journals (Diamond or otherwise) given that research funders who will be contacted are those funders who are already providing funding or who would now like to be funding this publishing method. Therefore, it does not include aspects relating to paper publishing nor the specifics relating to subscriptions. Secondly, it focuses solely on operations which are directly involved in the progressive transformation from submitted manuscript to published article, voluntarily eliminating aspects that are definitely important for a journal to function but which go beyond the process of strictly producing scientific articles. These aspects include defining an editorial line, the make-up of the editorial board, developing assessment procedures depending on the kinds of texts, managing editorial flows, research and management of funding, long-term archiving, etc. It must be borne in mind that some of these aspects have already been considered within the OA Diamond Journals study. Thirdly, the list is designed to be a process of discrete tasks which flow in a linear fashion. The reality of the publishing process is definitely more convoluted and complex, so the challenge is to present things with enough clarity to make it easier for funders to become involved.
We have ended up with a list of 26 acts, ranging from receiving the submitted manuscript through its assessment up to its dissemination as a published text (e.g. scientific article, editorial, minutes, etc.), describing along the way how it is evaluated and how its physical format is produced. We have selected the acts which are widely shared by scientific journals, regardless of whether it is a Diamond journal or not. Furthermore, we will see that some acts are more common than others. As is the case with all selection processes, we could come up with a different categorisation. In order to identify any potential gaps, each of the questions give respondents the opportunity to respond freely, thus providing details that did not fit the formal structure of our list. Some acts, such as website maintenance or archiving, were left out. The aim of our work, however is not to exhaustively cover every situation nor the specific situation of each and every journal, but instead to define clearly identifiable publication acts. This should make it easier for funders to support the publishing of Diamond journals. In so doing, our approach disregards some specifics in favour of more general trends. The initial list of 26 acts has led to the development of three levels ranging from the most disaggregated to the most aggregated (Table 2). At the first level, each item corresponds to a single acts. This first level is designed to be a linear tracking the process of publishing texts in a scientific journal, stage by stage. This is the version of the list to which the surveys have had access.
At the second level, we have grouped the acts into seven distinct groupings: manuscript handling, manuscript assessment, author information gathering, document production, rights management, metadata assignment, and dissemination. This second level was not made visible to respondents, as it was above all a tool for processing responses.
Although it is less exact than the first level, it makes the list easier to work with when identifying thematic trends.
Finally a third level of aggregation around three major categories gradually became apparent, i.e. certification, physical production of the document, and distribution. The distinguishing feature of this third level is thatit is inferred inductively from the survey's empirical results. Indeed, we will see in the rest of this report that the distribution of responses to the questions on publication acts is systematically organised around these three major categories. Table 2 Three levels of aggregation of production acts

Act Act no. Group of acts Major categories
Reception of a manuscript ("reception") 1 I. Acceptance of the document Certification Formatting the manuscript before entering the editorial process ("preformatting") 2 Communication with authors ("author communication") 3 Finding reviewers and monitoring their work, scheduling, etc. ("finding reviewers") 4 Reviewing (definition of criteria, written evaluation format, traceability of exchanges)("reviewing") 5

A list of entities carrying out the acts
In parallel with the list of 26 acts, we have drawn up a list of the entities likely to undertake them. Hence our question: "For any given act, who does it?" The entities selected are as follows: • Editor-in-chief, assistant  Several journals also mention that masters and doctoral students generally assume the role of editorial assistants and sub-editors in their publishing operation We realise that in order to faithfully reflect the diversity of organisational arrangements, we should have conducted another survey with more specifically targeted questions. We have chosen to use generic categories (editor-in-chief, editorial board, etc.), in order to enable us to explore the entities involved, while still being able to easily link them to questions on financing.
Secondly, in their free responses to Q3.4.1, several journals pointed out some shortcomings. The most frequently mentioned is the author (37 occurrences), who is involved in a wide range of acts (translation, image rights, preformatting, copy editing, graphics, proofreading, etc.). Furthermore, many of the envisaged acts (image rights, assigning a DOI, preformatting, conflict of interest check, and plagiarism check) are sometimes not undertaken within the journals (29 occurrences). Almost systematically, these framing problems were resolved by respondents ticking the other or I don't know answers. As will be seen, the percentages for these categories are relatively small in relation to all responses, and are thus unlikely to alter the trends we observed. As well as the role of the authors and the "nonachievement", we should point out the occasional mentioning of the "university" and "publisher" entities. Although these remarks may be important regarding the limitations of our study, they should not detract from the main aim of these two lists: to explore the acts in the production of an article for which funding is possible. Our approach does not intend to reflect all of the aspects that make up the reality of a daily journal. Thirdly, the free responses to the question on the completion of tasks have enabled us to clarify certain aspects. Thus, among services providers there are also companies which are able to take over several acts, and freelancers who often specialise in one act, (for example copy editing, website management, etc.). Similarly, and unsurprisingly, we encountered the [familiar] names of the main infrastructures used by Diamond journals -Amelica, Erudit, JTCAM, Lodel, OJS, OLH, OpenEdition, Ubiquity -as well as certain providers of specific services such as CrossRef for article metadata, or Lockss/clocks for archiving and conservation.

Which entities for which acts?
The way in which the acts are distributed in Diamond journals, arising from responses to question 3.1, shows trends which can be grouped into three major categories. As shown in  Finally, the third and final category (acts 19 to 26) also shows a preponderance of editors-in-chief. However, there are several entities working elbow to elbow, such as service providers, members of the editorial board, as well as the categories "other" and "I don't know", members of which have taken on the unfinished tasks. This category bundles the dissemination acts, i.e. the acts performed once the scientific article has been produced such as rights management (rights/contracts, licence management, etc.), the addition of metadata and the distribution itself. In the following, we propose to organise the way in which results are presented based on this emerging categorisation. This will then enable us to explore more specifically the trends within each of these three categories. 39

Category 1: certification
By focussing on the first category, it is possible to get a clearer picture of the way in which the acts related to publication are distributed (for acts 1 to 10, see figure 3a). This can be broken down into the processing of the manuscript, assessing it and collecting information about authors. Broadly speaking, the majority of actors who mainly carry out these acts are internal to the journal: the editor-in-chief or his/her assistants top the list of almost all responses (between 38% and 69%), followed by members of the editorial board, especially for author communication (27%) and finding reviewers (38%). Furthermore, the percentage of contracting is very low (from 0% to 4%).
In addition to the prominence of the editor-in-chief for the handling of the manuscript, we note the importance of the other category for the second act (2. Preformatting). From reading the comments in the free fields, we see that it is mainly the authors who do the preformatting themselves for many journals.
The role of external reviewers predominates for reviewing manuscripts (46%), even if the editor-in-chief and editorial board members occupy a significant place (28% and 17%). In the response to author act, the decision to accept or reject texts and replying to authors are generally done internally, first by the editor-in-chief (66%), then by editorial board members (27%). Finally, the plagiarism check is shared between the different entities: while the editor-in-chief predominates in this act (30%), it is delegated to a software package in 22% of cases, and in 13% of cases it is performed by entities that are specified in the responses given in the other category which, based on the free responses, again often correspond to the authors themselves.
Primary information processing on manuscripts (8. Plagiarism check, 9. Author identification is always one of the acts which is mainly carried out by the editors-in-chief and members of the editorial board (respectively 50% and 20%). Here too, the "author" role can be seen in the other responses, because it varies between 7% and 12% In a similar way to other certification-related acts, the percentage of journals that perform the contracting act remains low. The first five acts are dominated by a main activity of the journals' internal staff: copy editing and language editing are mostly done by sub-editors (44% and 38%), followed by the editor-in-chief (23% and 24%) and member of the editorial board (13% et 13%). Template compliance, graphics, and proofreading are primarily undertaken by editors-in-chief (37%, 26% and 37%), or by subeditors (28%, 25%, 25%). We note that graphics is taken on by the other category at 19%, where once again the free responses reveal the author role. Translation plays a special role: journals' internal members (editor-in-chief, sub-editors and board members) account for slightly less than 50% of the responses, while the other category peaks at 29%. The free responses indicate that this percentage is shared between the authors who directly undertake the translation work, and situations where translation is not done by the journal. Figure 3b. Distribution of the physical production acts A greater spread of responses is seen in the formatting of the text (to html, pdf, xml via coding/conversion) and in reference semantization. Coding/conversion is undertaken by the editorsin-chief (27%), by an external service provider, (22%) or by the sub-editors (20%). As for reference semantization, this is done by editors-in-chief (26%) and sub-editors (19%), although service providers and authors (as indicated in the other response) play a significant role here (10% and 13%).
We also point out that 13% of respondents are unaware of who undertakes this act.
Generally speaking, acts involving the physical production of the texts are more likely to require external service providers than those in the first major category relating to certification. Acts that are outsourced to service providers often represent between 5% and 13% of responses, and can even rise to almost a quarter for coding/conversion (22%), as already highlighted by us.

Category 3: Dissemination
The last major category, which bundles acts 19 to 26, is dissemination (figure 3c): the acts associated with rights management, assigning metadata, and distribution. For the most part, rights management, which includes image rights, licence management, and rights/contracts, is undertaken by the editors-in-chief (between 40% and 44%). This group of acts is also managed internally by the editorial board (at a level somewhat above 10%) and sub-editors (between 7% and 12%). There is significant sharing of these three acts with the other response category (between 16 and 23%). Free responses indicate that these acts are sometimes undertaken by authors, or are not done at all. Another significant trend is that the share of respondents who don't know the answer varies between 9% and 14%. We also show that the share of services subcontracted to external service providers is low: the highest percentage reaches 4%.

Figure 3c. Distribution of dissemination acts
The text publishing process also involves assigning metadata: textual metadata (title, abstract), assignment of a DOI, and incorporation into a broader document. Here again, the editorin-chief's role is the most important (respectively 35%, 24% and 49%), with other members of the journal (editorial board and sub-editor) together accounting for between 19% and 26% of journals for each of these acts. Software has a clear role in integrating metadata and DOI (11% and 19%), but the role played by external service providers is also more prominent, as they undertake 12% of this part of the distribution work. Finally, the prevalence of the other category (23%) for assigning a DOI (according to the free responses) reflects the statement made by some journals that they do not assign DOIs.
Text distribution involves putting it online and publishing its metadata. Once again, this work is dominated by the editor-in-chief (44% and 37%), and supported by the combined input of the editorial board and the sub-editor (28% and 20%). Service providers are involved in these operations to the respective levels of 14% and 10%, as well as the software developer (8% for both acts). Finally, in 9% of cases, respondents are unaware of who undertakes to publish metadata. The free responses particularly show that this work is sometimes not carried out.
To conclude this inventory, we should point out that we are unaware of actually who has answered this set of very precise questions. We have used the contact provided when the OA Diamond Journals study was carried out in 2020, but we are unable to be sure whether it is the same person, someone else, even a collective who actually responded. Nonetheless, based on the status given by the respondents of this previous study, over 80% of them were editors-in-chief or members of the editorial board, whereas only 15% of them came from the "technical" side of the teams, particularly the sub-editor, with the remainder coming from the distribution platforms or editors. Therefore, it is possible that these responses reflect the vision of these same stakeholders and that they have elevated their role to the detriment of others. Despite this uncertainty about the exact nature of the respondents within the journals, our survey reveals a rich, diverse field of stakeholders who participate in the publishing process. 45 Section 4

Monetisation of the publishing acts
We have described scientific publishing as a production process that sets in motion differently organised entities in order to perform a series of specific acts. This first exploration now allows us to examine more closely how the monetary transactions are distributed within Diamond journals, in order to assess their current funding arrangements.
The first section looks at the distribution of the sums of money by publishing acts. It reports the answers given to the question: "For each act, is there a monetary transaction or not?" (4.1). The second section looks at the different entities which are remunerated when a given act is the subject of a monetary transaction (4.2).
We have chosen to use the term "monetary transaction" rather than "payment" in order to clarify what appeared to be ambiguous, after the results of the 2020 OA Diamond Journals survey were posted. Some of the respondents made cost calculations, calculated monetary equivalents in terms of support or consolidated budgets, for example, but this did not necessarily represent the a dissemination of money required for the functioning of the journal in its current organisation. The concept of a monetary transaction dispenses with these issues, since it refers to the effective dissemination of sums of money, regardless of where they come from.

Publishing acts and financial transactions
The prevalence of work without any financial transaction.
The analysis of the monetary transactions is primarily based on the responses to our question Q3.2.
The number of responses regarding the completion of each act varies in the same manner as it does for Q3.1 (between 203 and 231), but remains high. For a given act, the responses are broadly split in binary fashion between "yes" for a act subject to payment and "no" for a act executed without explicit direct financial support. The third possible answer ("don't know") is found fairly rarely, up to 5% for the first 15 acts and between 2% and 7% for the following acts. As Figure 4 shows, an initial general result seems to emerge: in our sample of Diamond journals, the majority of publishing acts are carried out without any monetary transaction. Here we are in a situation which is typical of the general economic setup of scientific publishing where the work performed by the academics within the journals is an integral part of research work and is not directly monetised. Conversely, publishing work is indirectly remunerated via the salaries paid by institutions to researchers. This is supported by the free responses given in relation to this question (Q.3.4.2) which particularly stress the importance of unpaid work. Thus, several respondents stress the fact that the work is done completely on a voluntary basis, and that there is no kind of monetisation at play in the production process. Further to this position, the voluntary, non-monetised aspect of work in journals is presented in a particularly radical way by one respondent who says to be "philosophically opposed" to any kind of payment for scientific publishing.

ID 12781517408: "I am philosophically opposed to paying to publish and so would not be interested in externalising (for payment) any act of the process".
Nonetheless, caution must be exercised here since even the use of the words "volunteering" or "volunteers" is not widely shared. In the OA Diamond Journals survey, only 60% of journals declare that they make use of volunteers, but it is likely that almost all of the people holding a position do not receive any money from the journal or its owner. Certain posts, particularly those held by academics with a permanent status, are part of the profession's "moral economy", hence the lack of a reference to "volunteering". By having our study focus on the opposite concept of monetisation, we are hoping to obtain responses from a similar perspective. Another particular response concerns a journal that publishes conference papers. In this particular case, publishing is funded by the institutions organising the conference and they are responsible for publishing the issue.

ID 12753821473 : « Proceedings is funded by special issue organizers ».
How are the monetary transactions distributed between the different acts in the publication process. The breakdown of negative and positive responses shown in Figure 4 makes it possible to further explore the specific characteristics of our sample and to qualify the first overall impression.
In order to clarify the terms used, we have organised the presentation around the three major categories previously identified.
In the certification acts (acts 1 to 10), Figure 4a shows that the negative responses are particularly high (ranging from 74% to 90%). They peak at the finding reviewers, the review itself and the response to author (respectively 88%, 90% and 90%). Conversely, the positive responses range between 9% and 24%. The strongest trends are plagiarism check (24%), preformatting (20%),  Acts in physical production of documents (acts 11 to 18) have the highest percentage of responses ranging between 25% and 43% (figure 4b). Acts declared as being the most monetised relate to coding/conversion at 43%, copy editing (38%), language editing (36%), as well as graphic work on the text (graphics) (34%). These results are consistent with the possible outsourcing of part of physical document production to service providers. Having said that, there is still a majority of negative responses ranging from 54% to 71%. The least monetised physical production acts are template compliance (27%), proofreading (28%) and translation (25%) There is also a low number of respondents declaring that they are unaware that financial transactions exist for this group of acts, ranging between 1% and 7%. Figure 4b. Monetisation of the physical production acts Finally, for acts 19 to 26 that cover dissemination (rights management, assignment of metadata, and dissemination), the percentage of negative responses is between 64% and 81% (figure 4c). Two sub-groups can be clearly distinguished here: on the one hand, rights management, which gets the highest negative responses (between 77% and 80%); on the other hand, all of the metadata and dissemination acts, where negative responses vary between 64% and 69%. Monetary transactions linked to publishing acts are therefore higher for this second sub-group, assigning metadata (30%), assigning a DOI (31%), integration (30%) and posting online (34%). Once again, a low number of respondents is unaware that a monetary transaction exists (between 2% and 8%).

Which entities are remunerated for which publishing acts?
Gaining a clearer picture of the general breakdown of the monetisation of the acts makes it possible to identify which entities are involved when respondents say they pay for certain publishing acts.
To do this, we have cross-referenced the answers to questions 3.1 (Who carries out the acts?) and

(Do these acts lead to a monetary transaction?)
A first possible way of visualising this cross-reference is to break down the percentages of "yes" in the total population by the entities carrying out the acts. (Figure 5). The benefit of this visualisation is that it serves as a reminder that the number of positive responses to the question "Do the publishing acts lead to monetary transactions?" remains low compared to work without any monetary transactions. We see in particular that the entities which are funded to carry out the acts never exceed 21% of the total population, whereas the majority of responses are between 0% and 10%. 53 Figure 5. The breakdown of monetisation of acts in relation to the total population A second possible way of visualising this is to focus on the breakdown of monetary payments by entity within the more limited population of "yes" responses, i.e. the population of paid entities ( Figure 6). Obviously, this kind of presentation distorts the graph compared to the previous one, as the relative weightings of the responses in relation to the total population are abandoned. Let us take an example. In the previous graph, the editor-in-chief is paid in 8% of cases for act 6 (board decision). Furthermore, the sub-editor is paid for graphical work in 13% of cases (14. Graphical work). However, editors-in-chief alone dominate act 6, whereas act 14 also gives prominence to the contractor (also 12%). Thus, in relation to the "yes" population, the editor-in-chief, who is almost alone in act 6, is paid in 90% of the paid population cases. Conversely, the sub-editors and external service providers are neck and neck when it comes to graphical work: in relation to the size of the "yes" population, each of them is paid in 36% and 35% of cases.
Despite the magnifying glass effect, some significant trends emerge. Let's start with the certification acts category. This is dominated by editors-in-chief. The highest remuneration percentages occur for author communication (69%), finding reviewers (81%) and the response to author (90%). In contrast, remuneration of editors-in-chief is lower for preformatting (38%) a major part of which is delegated to the sub-editor (27%), reviewing (32%) shared with external reviewers (27%), and plagiarism check (23%); for this act, "software" constitutes the biggest number of responses (42%).
Physical production acts (11 to 18) are marked by the significant presence of sub-editors who are paid for copy editing (57%) and language editing (54%), and that of the service providers who are instead paid for translating (43%) and formatting documents (49%).
Finally, acts relating to distribution (acts 19 to 26) show an even more diversified breakdown.
While rights management is dominated by the sub-editor for image rights (37%), both the editorsin-chief (who perform the licence management and rights/contracts acts respectively in 41% and 39% of cases) and the service providers dominate in the remuneration for acts relating to metadata (assigning metadata) and dissemination (between 28% and 37%). 55 Figure 6. The breakdown of the monetisation of acts between stakeholders Funding publishing acts in an ideal world This study's aim is to explore possible means of funding for Diamond journals, following on from the recommendation of the first OA Diamond Journals survey carried out in 2020 within the OPERAS consortium. In this light, we would like to clarify the funding needs of Diamond journals. We are going one act further compared to the previous section. As well as looking at those acts which currently involve a monetary transaction, the aim is now to look at what could be funded, if funding were available. It is not immediately apparent what the funding needs of the journals' publishing acts actually are. To achieve our aim, we turned to the concept of the "ideal world" which is defined as a situation without financial constraints where unlimited funds would be available to journals.
On the basis of this hypothetical situation, we questioned the Diamond journals about the likelihood of them paying for a given act in the publishing process. Of course, an ideal world where there are no financial constraints should not be viewed as a realistic possibility for the scientific publishing economic setup. Instead, it serves as a useful fiction for drawing attention to the precise acts for which the journals would like to have additional funding. In short, the ideal world is a proxy to help us grasp Diamond journals' needs.
The first part presents the results regarding the needs expressed by the journals in an ideal world (5.1). The second part of this chapter focuses on the expenses being kept on paid for certain publishing acts in an ideal world (5.2).

Financial requirements of journals
We asked the journals to indicate which acts they would be prepared to pay for if there were no financial constraints at all (Q3.3). This question gives them the opportunity to reveal Diamond journals' actual funding needs, noting specifically where the funding is needed. As for the previous questions 3.1 (Who carries out this act?) and 3.2 (Do you pay for this act?), the total number of responses ranges between 202 and 231. As Figure 7 shows, the responses show marked trends between the 26 acts, which again break down into three major categories as described above: certification (acts 1 to 10), physical production of the document (11 to 18), and dissemination (19 to 26).

Certification
The majority of journals declare that they would not be inclined to pay for the certification acts, even in a world where there were no funding constraints (Figure 7a). In this respect, the decision to accept or reject (6. board decision) reaches the highest level with 74% saying "no". Then there is the author communication, finding reviewers and reviewing, as well as responding to author and author identification, with negative response percentages ranging between 59% and 65%. While less marked, conflict of interests check (55%) and preformatting (47%) are considered to be acts that do not require funding. One notable result, the checking of plagiarism (8. Plagiarism check) is the only where the negative responses are lower than the positive responses (37% "no" compared to 45% "yes"). This result is unsurprising, given that in 22% of cases 8 is performed by software (see section 2.1., Figure 3a).  If we look closely at the "maybe" responses which break down to 8% and 15%, we can see that the spread of potentially supported acts follows the same trend (with lower scores) as the acts already funded (Figure 7b). Figure 7b. Breakdown of funding opportunities ("yes" or "maybe") When the results obtained from both responses are added together, we can see the potential funding score per certification publishing . Two examples of this are: preformatting which becomes a for which funding is more apparent (49%), while plagiarism check has increased (to 58%).

Physical document production
For the process of the physical production of texts, the trend is reversed compared to certification, with positive responses dominating. This result is consistent with the two previous results. On one hand, this part of the publishing process is where the majority of journals report paying for external service providers. On the other hand, the open responses linked to the 'ideal world' point in the same direction. For instance, respondents state that these are technical acts which are easy to outsource unlike the rest of the editorial work. There are also arguments in favour of professionalisation and therefore better quality of service when resorting to specialist service providers.
ID 12747407738 : « Externalization is unavoidable. At the present, we must teach people for these services, but in an ideal world we could ask for professional services already built for this purpose ».
In an ideal world without any funding constraints (Figure 7c), the acts which the journals would be most willing to pay for are copy editing (60%), language editing (59%) and coding/conversion (56%). The rest of the responses range between 43% and 50% for checking compliance, graphics, proofreading, translation and reference semantization. Negative responses vary between 29% and 39% for acts 13 to 18 (checking compliance, graphics, proofreading, translation, coding/conversion, reference semantization). If the "yes" and "maybe" responses are added together; the majority of responses are positive, between 58% and 73%. Journals are more against the idea of a monetary transaction for image rights (46%), licence management (48%) and rights/contracts (53%). The trend for integrating metadata is even more nuanced, with 41% of both yes and no responses Assigning a DOI gets more negative responses (45%), as does integration (49%). The responses regarding article dissemination are also mixed, with journals stating that they are fairly willing to pay for this (43%). On the other hand, publishing metadata receives more negative responses (45% against 37%). The "Don't know" response percentages vary between 3% and 11%. As with the two other major categories of publishing acts, the "maybe" responses are relatively stable, ranging between 9% and 14%. Adding the "yes" and "maybe" results together boosts the share of positive responses  Thus, image rights, assigning a DOI, integration, or article data publication all have more or less similar scores (48% against 45%; 48% against 49% and 48% against 45%). By contrast, licence management and rights/contracts always show a large majority of negative responses (48% and 53%). Conversely, assigning metadata is more dominated by positive responses (55%), as is posting online (55%).

Continuing payment in an ideal world
Having presented journals' willingness to pay for publication in an ideal world, we now turn our attention to the small sub-group that responded to question 3.2. Doing so enables us to look more closely at the conditions for continuing the current payment in a world without any budgetary constraints. To put it another way, it enables us to grasp the extent to which some current expenditure is incurred but not necessarily welcomed by the journals.
The visualisation below (Figure 8) shows the number of journals which, out of the group of current payers, would continue to pay for a given in an ideal world. By way of a methodological caveat, we would like to point out that this visualisation focuses on a small part of the overall population of respondents (solely the population of journals that pays for a given ). One predominant result can be seen in the graph: for the great majority of the acts, the current payers would generally like to continue with this payment in an ideal world where there were no financial constraints. Let's look at this graph from the point of view of the 3 major categories of acts. and 53% of positive responses. The range of negative responses for these same acts is between 32 and 37%. However, we should point out several specific cases: the editorial board's decision is always overwhelmingly associated with a lack of a monetary transaction (62% "no" responses against 29% "yes" responses), whereas finding reviewers and author identification get a small majority of "no" responses (44 and 46%). Finally, the journals are split as regards responding to authors: the number of journals that are in favour of continuing payment for this is, in an ideal world, the same as the number of journals that are against it (42%). The "don't know" responses still constitute a small share of all responses: between 5 and 10%. Finally, the percentage of those who are undecided ("maybe") varies between 5% and 12%.
The second major category of acts covers the production of the document content. This part of the graph shows an overwhelming majority in favour of continuing payments. The highest score is 78% for language editing, whereas the lowest score is 63% for proofreading. There is a relatively low level of negative responses, between 16% (copy editing, language editing) and 25% (compliance). "Don't know" answers remain low (between 2% and 8%). "Maybe" responses are between 6% and 8%. By adding them to the total number of Yes responses, this would only increase the prominence of Yes responses in this part of the graph.
The third and last category is dissemination. Here we can observe a similar trend to that seen in the previous category, i.e. a clear preponderance of journals in favour of continuing monetary transactions, albeit in a more measured way. Positive responses vary between 54% (image rights) and 66% (DOI), whereas negative responses range from 19% (integration) to 51% (licence management). The same percentage of responses as for production acts can be found for the "I don't know" responses (between 9 and 10%) as well as for the "maybe" responses (between 4% and 9%). The ability of journals to carry out monetary transactions.
This section is devoted to the Diamond journals' ability to transact -a crucial point when implementing direct funding mechanisms. In particular, we differentiate between the direct (6.1), indirect (6.2), overall (6.3), and potential ability to transact (6.4).

Capacity to carry out direct transactions
We asked the journals two separate questions regarding their ability to execute monetary transactions. Are they able to receive money? And are they capable of making expenses? As illustrated in Figures 9a and 9b, the response patterns to these two questions are extremely close, which would suggest that the answer to one of them generally implies the same answer for the other. Indeed, the large majority of the 254 respondents is capable of receiving money (70%), and a very similar result is found, albeit slightly higher, for the ability to spend (73%). The flip side of this is that a sizeable minority (21%) is not able to receive money, while 19% of journals state that they are unable to spend money. We should also point out the small proportion of journals which do not know what their journal's transactional capabilities are. 9% for receiving money, 8% for spending it. Therefore, we asked the journals that did not give a positive response to the two previous questions if 1) an intermediary was able to collect money on their behalf, and 2), if an intermediary could make expenses on their behalf. We received 72 responses to these two conditional questions. What we notice first of all in this subset is that the response patterns between the two conditional questions are not the same. The proportion of "yes" responses is higher for receiving money and, conversely, the proportion of "no" responses is higher for spending money. There is a majority of "no" responses for both questions (Figures 10a and 10b): a majority of these journals are unable to receive money via an intermediary (42%) nor are they able to spend it (50%). Slightly more than a third of respondents answered "yes" for the capability of receiving money (35%) and less than a quarter for spending it (22%). Finally, 24% of the respondents do not know if the journal can receive money via an intermediary, and 28% do not know if the journal can spend it. we increase from 73% to 79% positive responses. Table 3 presents these aggregated elements.

Potential transactional capability
Of the 72 journals with no direct transaction capability, 50 of them have indicated they would be prepared to implement a transaction accounting system if the funding allocated to them was sufficient for the purpose. As shown in Figure 11, 30% of the journals would be ready to implement this kind of accounting system, 42% would refuse to so, and 28% say that they don't know. Figure 11. Potential financial capabilities As the population size here is fairly small, the results regarding the willingness of those journals without an accounting system to adopt one should be treated with some caution. However, the 30% of positive responses, or 15 journals, can be considered to be a conclusive result as to the ability of these journals to put in place accounting arrangements in order to receive money under certain conditions. Therefore, when this number (n=15) is added to the journals which have declared that they are able to receive money (n=203), the total number of journals with the potential ability to perform monetary transactions reaches 218, or 86% of the journals in the survey population (n=254). These elements, exhibited by a large majority of journals, are shown in table 4. What is the visibility of research funders?
The development of funding channels in a world structured around the principles of transparency and openness goes hand in hand with the ability of journals to make their various funding sources visible. This section probes the readiness of journals to ensure this traceability of research funders.
The first part provides an overview of the current abilities of journals to provide guarantees of visibility to research funders, particularly through reporting practices (7.1). The second part focuses on the journals that lack a reporting capability, and reflects on the possible incentives for journals to do so (7.2).

Reporting capability of journals
The  While it is important to ensure the traceability of funding by article, it does not fully address the visibility of journal funders. To ensure this visibility, it is usually necessary to produce a funder report that collects, per funder, the articles it helped produce. As shown in Figure 13, the trend is reversed here: out of the 136 responses to Q4.2, only 36% of journals state that they were able to produce a report of this kind, whereas 40% state that they are unable to do so. The large proportion of respondents who do not know whether it is possible to provide reports to funders is worthy of note here (24%).   dollars for the lowest threshold, to a range between 8,000 and 20,000 dollars for the majority of responses. These amounts of money are subject to several associated conditions: the importance of not foisting additional work and costs onto the publishing process. Some journals therefore advocate providing technical support in order to facilitate the generation of the report. The option of employing a person who is capable of producing a report, for example a "copy editor", is also a strongly preferred option.
Despite the opportunity to receive a regular income for the journal, a small minority refuses to consider any kind of reporting (a total of 13 journals or 7% of responses to Q18). 9 of these were willing to provide explanations by replying to the open question Q4.5. We mainly want to underscore the diversity of the disciplines (post-colonial studies, social sciences, geography, biology, material science, etc.) and the nationalities that make up this sub-sample group (France, Italy, the USA and several international journals, including a group of Middle Eastern countries). There are three main arguments put forward to explain the refusal to implement a reporting system. First, some journals consider that they do not need extra funding and that they only rarely deal with funders, or that the research underlying the article proposals is not very dependent on subsidies from research funders. Moreover, some respondents highlight the financially sustainable nature of the journal, such as a journal which is integrated into OLH. Second, the reporting prospect is not attractive for some journals because of the extra expected administrative burden, and the risk of losing the journal's independence. Third, one respondent argues that the researchers rather than the journal should be funded.
The ability of funders to provide direct support for journals As was pointed out in the OA Diamond Journals report, there are many different stakeholders who all contribute to the Diamond ecosystem, and the funding mechanisms that we are proposing are generic and could be implemented by many of them. Nevertheless, as indicated above, we chose first to explore the contributions of research funding institutions, because a number of them (1) already fund open access with APC, and (2) imposed transparency requirements for publications funded by their fellowships. To do this, we have been in contact with Coalition S via Johan Rooryck and we have designed a two-stage approach.
We first provided a series of preliminary questions with a view to understanding the legal and regulatory constraints weighing on funders. These issues have been studied within Coalition S and have been added to other issues that relate in particular to books, and a survey should be prepared among the various partners on the current status of their funding for publication, and their regulatory constraints. Once this first stage will be completed, we should work together to build an information workshop and a survey for its members, which should be based on the following framework. First, we should present the empirical results of our research, as described in this report in summarised form. Next, we would present the proposed funding mechanisms, taking into account of course the results gathered from the funders, and in particular the need or not to go through third parties to finance publications. Finally, the second part of the workshop would be devoted to questions and answers from funders in order to assess the realistic and desirable nature of this direct funding, in addition to the funding of Diamond platforms which already exists today for a certain number of them.
Conclusion: advantages and challenges of the direct funding models After we have presented the funding models -the specific destination of funding flows and the technical conditions of implementation -we will give a short conclusion on the advantages and limitations of direct funding support to journals. This conclusion will fully leave the floor to the journals. This has enabled us to identify many advantages but also several pitfalls to avoid regarding the implementation of direct funding.

The advantages of direct funding models for journals
One general notion that emerges is that this funding would provide support for journals and would ensure their continuity, whether carried out by funding software such as OJS, a "capacity centre" to access services, or by subcontracting certain acts such as copy editing, proofreading, etc. Second, the outsourcing of physical production acts, thereby allowing editorial teams to refocus on the scientific content, which would improve content quality.

Potential pitfalls associated with direct funding models
Despite these benefits, respondents also identify a certain number of disadvantages that would make the implementation of potential direct funding more complex. These are expressed in two different ways: either as a condition to be fulfilled in order for the funding to be genuinely beneficial to journals, or as an impediment that it is difficult to circumvent -this is the position in which we find the 7 journals that are opposed to direct funding, one of which is deeply pessimistic about implementing a project of this kind.
The first aspect mentioned by the journals is that of scientific independence: the risk that the funder may influence the editorial group's scientific policy. Here, several respondents insist on the need for a funding channel maintained by public institutions. Third, some respondents fear the rise of "predatory journals", that would unfairly capture funding. Fourth, seven respondents consider that public funding is not an option, since their national institutions are not able to directly fund journals. Finally, the fact that the majority of the work that is done for journal articles is not funded by research entities. They would then not be able to benefit from funding.
ID 12777951732 : « We think is a good idea. However, this would probably not be applicable to our journal as most articles we receive/publish are not externally funded ».
All of the advantages and pitfalls mentioned here are summarised in table 5 below.  19 . The final version of this survey lists 26 acts which we have arranged into 7 groups and 3 major categories. At the start of Chapter 3, we provide a detailed presentation of this list when introducing the issues relating to the publishing process. Here we also refer to the theoretical and methodological challenges inherent in compiling this kind of list, along with its limitations.
In the second, we finalised a version of the questionnaire on funding methods for Diamond journals. The questionnaire begins in a traditional manner by presenting the study. It was then structured into five sections covering the following areas: basic information about the journal, the financial position, the acts involved in the publishing process, the relationship with research funders, and options for a direct funding mechanism (Table 6). 2. Your journal economic configuration Position of the journal within an a wider economic entity, its ability to receive/spend money 3. Tasks on a given manuscript Who carries out the publishing acts? Which acts are paid for? Which acts would be paid for in an "ideal world" without any financial constraints?
4. Funding and grant report Ability to identify the research funders by article and to provide reporting 5. Opinion on funding mechanisms Opinion of journals on the principle of direct funding, and on the forms that such funding would take.
Out of the five different sections, the third one is devoted to mapping work. Based on the table of acts referred to above, we asked the journals in the survey to define the actors that carry out each , where the resources used come from and what the journal would do if it could access new funding. The first draft of the questionnaire was produced as a text file. Once the questions were finalised, we imported the questionnaire into an Excel file in order to facilitate the future testing stage.
The third stage involved a collective discussion of the questionnaire's content. We tested the questionnaire twice in order to ensure that the questions were relevant. First, we undertook a number of discussions with members of the OPERAS consortium. They gave us some feedback and recommendations via a shared document. Second, we had the questionnaire tested by a "testing journal" who agreed to fill in the Excel version. These two tests enabled us to identify elements related to a proper understanding of the questions and to the choice of vocabulary used.
The fourth involves transcribing the questionnaire into Survey Monkey, an online software for creating questionnaires, sending them out to respondents, collecting the data, and analysing the data generated. This change in format enabled us to improve the questionnaire technically by introducing conditional questions. The transition to Survey Monkey also provided an opportunity to rethink the graphic and ergonomic presentation for the respondents.

Dissemination and the collection of data.
The OA Diamond Journals survey which was carried out in 2020 enabled us to identify a sample of 1,252 journals which agreed to being contacted again for subsequent surveys. This sample group has one major advantage. Since we already had information about the specifics of these journals (such as the discipline involved, the country and how long the journal has been in existence -all of which was gathered in the course of the OA Diamond Journals survey), we were able to focus our survey on the specific questions related to our research subject. The questionnaire was made available on 16 June 2021 via an e-mail sent out to the journals in the sample. After two email reminders, the questionnaire was closed on 12 July 2021. 296 individuals opened and began filling in the questionnaire. After we extracted the data using Excel spreadsheets, we then cleaned it by deleting those participants who ultimately failed to provide responses and by deleting some duplicates. In the end, we gathered a total of 260 journals whose responses could be used for analysis. There number of responses to the different questions varied between 200 and 260 for general questions, then went down to less than 70 for conditional questions, and further down to only several respondents in the case of subsidiary questions.
We were able to match data from the previous survey by the title of the journal and the ISSN provided by our respondents. This way, we were able to identify 252 journals, enabling us to have specific data for the journals (disciplines, localisation, size), as well as operational data which was gathered beforehand (consolidated budget, annual number of articles, etc.). The 8 remaining journals could not be matched. They probably responded to our questionnaire due to our correspondence being forwarded to them. We conclude this section on questionnaire dissemination and data gathering by reiterating a warning mentioned in the introduction. When we state that we were able to analyse 260 responses, the range of different people potentially responding on behalf of their journal needs to be borne in mind. There may be a variation from one individual to another, from one respondent to another among the different job titles within a journal (editor-in-chief, member of the editorial board, sub-editor), leading to a variation in the level of knowledge of the publishing process and the types of funding. The responses to the open questions are a testimony to this diversity, as this is where extremely detailed comments as well as brief, vague, or even non-existent remarks can be found. If the respondents are the same as for the OA Diamond Journals survey (we contacted the journals using the email addresses that were provided in that survey), then 80% of respondents are editors-in-chief or members of the editorial board, and this may influence the responses provided.

Data analysis
In this last section, we describe in detail how we processed the data we gathered. We begin by explaining our methods regarding quantitative processing, before developing the qualitative aspect.

Quantitative treatment
Most of the questionnaire is structured around closed questions which impose a limited number of predefined responses. In this way, much data easily lends itself to a simple quantitative analysis that can de graphically portrayed. The online Survey Monkey software provides graphical formatting of responses to closed questions from the start. That said, we preferred to work with the raw data by exporting it directly into Excel. There were two reasons for this: the first was in order to do data cleansing (as mentioned earlier), and the second was to ensure a uniform graphical presentation.
Some visualisations in Survey Monkey were unusable since the large amount of information generated dense, unreadable images.
In most cases during quantitative processing, we retrieved the dataset relating to a question, generated a table compiling the information and produced a graphical representation of that. We note that the graphs are usually histograms. The number of responses may vary from one question to another, particularly in part 3 of the questionnaire involving the list of publishing acts, so we systematically chose to display percentage levels rather than absolute values. For each histogram found in the report, the number of respondents is given in the body of the text.
We have also done some cross-referencing of datasets generated by the responses to several questions. Let us take an example Question 3.1 looks at which entities carry out the acts in the production process. Question 3.2 asks whether or not funding exists for each act. Therefore it is possible to know, for each respondent and for a given act, who performs it and if the involves a monetary transaction. For example, for the first acts (1. reception), out of the population of paid actors, 20 of them are editors-in-chief, only one of them is a member of the editorial board, 2 are sub-editors, and so on -as can be seen in Table 6, which presents the information on the entities paid for the different publishing acts. For all cross-referencing, it is possible to consider an overall population (in our example these are the entities that complete the different acts), a smaller population which is governed by stricter criteria (in our example, entities who carry out the acts AND who are paid for doing them). Under these conditions, expressing these results in percentage terms means they can be compared to data from the overall population or to data from a more restricted population. When presenting crossreferencing of this kind in the remainder of the report, we specify what methodological precautions need to be taken when interpreting the results in percentage terms.

Qualitative processing
There were several open questions within the questionnaire which required free answers and which cannot be compiled and tallied. We have differentiated between two kinds of open questions which we have processed in different ways.
The first kind involves questions with a large number of answers (at least 200). This is the case for the last two questions (5.1 and 5.2) which probe the respondents' opinion on ways of directly funding Diamond journals, as well as for the free questions associated with publishing acts (3. 4.1, 3.4.2, 3.4.3). In order process this data, we exported the responses to a text file and used qualitative data analysis software called ATLAS.ti. ATLAS.ti makes it possible to perform inductive coding while reading, and ultimately organise the material into several different categories and subcategories. For example, question 5.2 examines the Diamond funding models that journals were considering. The coding within this software enabled us to identify several general categories (a model for a funding infrastructure, a model for service provision, a direct funding model, implementation conditions for a model, origins of funds, desired funding). These categories are then broken down into several sub-categories (for example, a direct funding model with advertising, fundraising, voluntary contributions, fixed allocations, yield-based funding of publications. Once the categories are established and the coding has been done for all of the text, it is possible to retrieve all the sections of text associated with a particular code and thus rapidly gather examples which empirically support certain elements of the report's analysis. The second kind of open question are those that elicited few responses, generally less than 10. These are ancillary questions which we sometimes ask about sums of money that the respondents are considering (to finance some activity). For example, conditional question 2.7 asks the journals lacking transactional capability (receiving money/undertaking expenditure) if the existence of sufficient funding would make them willing to set up an accounting system. The possible answers are "yes", "no", "don't know", and by responding by "yes", they would then be asked to enter in a free text field the amount of money they think would be sufficient for the purpose. We received four relatively short responses, so we processed these manually, reading the responses directly on the Excel spreadsheet and interpreting them. The value of this kind of response varies hugely, so they are not all mentioned in this report.