Do Infrastructures Have Epistemologies? Studying an Open Access Infrastructure for SSH from Within

a. Origins: A Heritage from Cybernetics and Big Science

A first clue to trace the genealogy of French SSH infrastructures is to look to the United States. In Code: From Information Theory to French Theory, media historian Bernard Dionysius Geoghegan (2023) tells the prehistory of datafication in the social sciences, beginning in the interwar period in the United States with the first applications of Claude Shannon’s information theory—glossed by Warren Weaver—to social science disciplines. As early as the 1930s, “robber barons’ philanthropies”—notably the Macy and Rockefeller Foundations (see also Hauptmann 2024)—funded research programs, particularly in anthropology with Gregory Bateson and Margaret Mead, aimed at using the new audiovisual media to produce material inscriptions of social phenomena, which could then be edited and analyzed to identify recurring patterns. During the war, Norbert Wiener developed the founding concepts of cybernetics around these analytical practices, notably the (in)famous “feedback loop.” Within cybernetics the world thus became code and the social sciences decoding practices. The link with French social sciences was forged at the École libre des hautes études, founded in New York in 1942 and financed by the Rockefeller Foundation, which welcomed researchers in exile from Europe, notably Roman Jakobson and Claude Lévi-Strauss, whose structural anthropology owes much to this cybernetic period. After the war, around the École des hautes études en sciences sociales (EHESS), again financed by the Rockefeller Foundation, Lévi-Strauss developed a genuine data repository, the Human Relations Area Files, which stored two million index cards of ethnographic data on a global scale and was equipped with a mechanographic device for consulting it in order to decode myths, among other cultural phenomena (Plutniak 2019). Geoghegan also notes the technocratic aspect of this genealogy, whose actors all have backgrounds close to public and economic authorities, colonial administrations, or NGOs such as UNESCO. The datafied social sciences are not only tools for describing reality; they are also instruments for controlling it.

This history of the concomitant datafication and infrastructuration of (French) SSH along the lines of epistemic features borrowed from mathematics and the natural sciences is continued by the recent work of Jonathan Zurbach. His detailed history and analysis of the “discourse on infrastructures” for SSH in France shows how, from UNESCO in the 1960s to OECD in the 1990s to the European Commission in the 2000s, the institutional entrepreneurs of SSH borrowed the rhetoric of Big Science—what he calls a “practical syllogism” (Zurbach 2024, 60)—to legitimize their massification, hence their funding. Next to telescopes and particle accelerators, digital libraries and data repositories become the instruments of a “datafication of social sciences” (25) that aims at elevating SSH to a superior epistemic status. Indeed, “the defense of these ‘infrastructures’ for ‘data’ is the defense of a particular social science practice, or even a conception of science” (27), one for which the “statistical proof” is “proof par excellence” (757). Like Geoghegan, Zurbach highlights the ambivalence of French research infrastructures that aim both at “supporting [soutenir] and improving [améliorer] research” (31): “The tension between ‘supporting’ and ‘improving’ can be seen as a translation of the fundamental tension inherent in the process of infrastructuration. The tension of infrastructuration involves, on the one hand, codified and institutionalized traditional scientific identities and practices, which as such exert a constraint on actors, leading them to support the latter, and, on the other hand, emerging scientific identities and practices, carried by individuals or pressure groups, whose aim is to improve the traditional game, leading these actors to engage in reformative work” (754). The infrastructure, which is shaped by, as much as it shapes, research communities and their practices, becomes the crux of a broader “reform” of SSH along the aforementioned epistemic features.

b. Context: A New Regime of Knowledge

These forgotten origins of French SSH infrastructures now have to be reappraised in their specific context: a new “regime of production, regulation, and appropriation of knowledge.” This concept, forged by STS scholar Dominique Pestre, can be defined as a historical period resulting from the (relatively) stable articulation of situated scientific definitions and related issues, forms of research outputs, socio-epistemic practices, values and norms, organizations and institutions related to knowledge production, “and so much more” (Pestre 2006, 104–5).⁴ A regime of knowledge, in short, is a way to understand the socio-political nature of knowledge production and use as an arrangement of, on the one hand, ways of knowing (standards of proof, forms of knowledge, an order of knowledge, a hierarchy of disciplines) and, on the other hand, a social and political order (the role of public power, forms of government, a hierarchy of values, rules of ownership, economic valorization). Because it is (relatively) stable, a regime allows for the anticipation and regulation of knowledge production through diverse means of steering, pressure, and governance of socio-epistemic practices and their outputs.

It is generally acknowledged that since the 1970s in the United States, and soon after in Europe, a new regime of knowledge has been emerging (Bonneuil and Joly 2013). From a 19th- and early 20th-century regime in which science was deemed a public good serving the progress of a welfare state—and economic power, military domination, and identity construction—a gradual transition has occurred towards a new regime, sometimes referred to as “neoliberal” or the “knowledge economy.” In this new regime, with the ebb of public funding and the managerialization of research through new public management practices, the sciences become subject to market logics that promote the financialization of innovation and project-based funding, as well as competitive dynamics in the production of knowledge, underpinned by symbolic economies such as citation metrics and impact factors. In this context, where science is seen as a source of competitiveness, the advent of digital technology and the internet has helped to open up and break down a number of historical “black boxes” that used to act as mediators within the scientific field, giving rise to a set of socio-technical interdependencies between new components, proliferating new black boxes, and actors or stakeholders that were previously enclosed within integrated systems.

This change of regime also entailed a shift in the scientific ethos—the DECAY, or “differentialism, egoism, capitalism and advocacy” (Macfarlane 2023), of Robert Merton’s CUDOS (Communalism, Universalism, Disinterestedness, Organized Skepticism).⁵ Indeed, however normative or idealist, it is no wonder that Merton’s sociology of the scientists is at the heart of the discussion in the first publication on “open science.” In his 1985 seminal paper “Open Science and Closed Science: Tradeoffs in a Democracy,” sociologist and US policy advisor Daryl E. Chubin more specifically discussed the norm of communalism—“open communication [as] an imperative for scientific integrity” (73)—and its demise in the face of new knowledge enclosures resulting from the emergence of “corporate science,” itself fostered by “liberal democracy” in the 1970s and 1980s. Openness and closure become the two ends of a tradeoff: “liberal-democratic society has complicated the production, communication, and utilization of science. It has politicized scientists by discipline and by employment sector; it has caught science in a torrent of tradeoffs. Democracy has made ‘openness’ and ‘secrecy’ the subjects of policymaking and conjecture; it has forced consideration of how science can be closed by special interests and closed off to the public good” (74). In this ambit, digital knowledge infrastructures thus appeared as a technical means to rearticulate, for the benefit of their stakeholders, the redistribution of people, skills, tools, and knowledge, as well as balance the tradeoff between openness and closure (Mounier and Dumas Primbault 2023).

In the French case more specifically and building on Pestre’s concept of regime, Célya Gruson-Daniel studied in what ways “the open in the sciences and the meanings associated with it provide a better understanding of the current reconfigurations of the knowledge regime with digital technologies” (Gruson-Daniel 2018, 75). Focusing on the 2015 national public consultation that preceded the parliamentary debates on the Loi pour une République numérique (2016), she compiled, coded, and analyzed the discourses on science held by a diversity of actors on the dedicated platform, as well as on other media and social networks. In doing so, Gruson-Daniel identified four normative “conceptions” of “the open in the sciences,” ranging from the defense of past, pre-digital regimes of science (chaps. 5 and 6), to the embrace of digital networks in the service of scholarly communication (chap. 7), to the call for an “informational liberalism” grounded in free software and knowledge commons (chap. 8). She situates French public knowledge infrastructures in general—and, namely, OpenEdition—in the defense of a “civic-technoindustrial regime”: a means for public research to “regain control” over the production and circulation of knowledge for the public good through the creation of a network of public innovative infrastructures (chap. 7). Like Zurbach, Gruson-Daniel highlights that the type of SSH knowledge fostered by this regime comes under the banner of “digital humanities,” even the “institutional scientific humanities” (228–35).

c. Rationality: l’agir ingénieur

Finally, besides the origins and context of SSH infrastructures, some light can be shed on the practical ideology, or rationality, that animates these infrastructures by looking at other French public infrastructures dedicated to openness: government open data repositories. Studying, as an embedded researcher, diverse loci of the design and development of the policies and infrastructures for government open data, Sarah Labelle highlighted how much infrastructure building was “a cardinal element of political change” (2020b, 34). A “reform” of the State is thereby endeavored by mobilizing engineering know-how as a way to transform the world in a “contraction of politics and technology” (61). This is what she calls “engineering action [l’agir ingénieur]”: “data policies presuppose that subjects mobilize a model of action that considers the state of the world in terms of the engineering processes that can transform it: the relationship to the world is thought of and defined in terms of infrastructures, i.e., it is thought of as mediated. I propose the term engineering action [l’agir ingénieur] to designate the model of action that these subjects mobilize” (35). Yet, while the infrastructure—such as an open data repository—is acknowledged as a means to bring about political change, notably in a particular form of transparency, it is itself supposedly devoid of any political content. Somewhat paradoxically, the infrastructure would still lie outside politics and value conflicts, not only because of a certain conception of the nature of “data” but also because it merely calls for practical involvement of citizens with data, without constraining or prescribing said practice: “Engineering action is characterized by an indifference to the question of ends: it’s more a question of getting people to act than of achieving a certain action. […] Engineering action consists in producing and providing a framework whose plasticity favors practices, whatever they may be: it’s a question of producing devices that offer a potential for action, that ‘elicit practices’ but absolve them of any precise objective” (39). In this rationality that eschews the prescriptive and performative powers of the infrastructures on the socio-epistemic practices and communities they mediate,⁶ the embedded researcher is welcome as a colleague, yet another engineer whose expertise is expected to be instrumental: “I experience this expectation with a certain anxiety, because I don’t know if I can live up to it. This brings us back to the position of the [embedded] researcher in the field, a position that reveals the usefulness desired [utilité souhaitée] by the various interlocutors” (Labelle 2020a, 81).

For some actors and commentators of open science, this becoming-engineering of scientific research is desirable and must be fostered by national and international scientific policies, including open science policies. For René Von Schomberg, a professor of STS affiliated with the Directorate General for Research and Innovation of the European Commission for 24 years (1998–2022), “The sciences have evolved over the decades post-Merton and are now intertwined with societal and industrial interests. Merton’s image of science as primarily aimed at explaining or understanding natural and social phenomena has been subject to change, with many sciences adopting an engineering perspective” (Von Schomberg 2024, 16). From this perspective, open science policies can help steer scientific research along a “problem-solving” (28) path and become “societal-challenge driven mission-oriented research” revolving around a “socio-technical imaginary” such as “smart cities” (10). This instrumentalist steering of science as innovation- or policy-oriented research—or “mode 2” involving the “Quadruple Helix, including academia, industry, civil society, and public authorities” (Von Schomberg 2024)—preceded and grounded open science policies at the European level (Vanholsbeeck 2017).

What this genealogy shows is that, at the turn of the 21st century and at the crossroads of their origins, context, and rationality, nascent SSH infrastructures, particularly in France, could then be expected to embody and enforce, in part or in whole, some of the following four socio-epistemic features:

• A data-driven epistemology or data-intensive conception of scientific research⁷—that is, some flavor of “the end of theory” (Anderson 2008) or “the fourth paradigm” (Hey et al. 2009), which renews the promises of the so-called Scientific Revolution, such as radical empiricism, radical inductivism, representationality of exhaustive data, and neutrality of the computation tools.⁸

• A civic-technoindustrial regime of knowledge production, regulation, and appropriation that strives to regain control over public scientific research by the design and development of a national network of large-scale digital infrastructures made to support, improve, and circulate scientific research.

• A practical ideology in the form of a technical rationality, premised on a distinction between technology and politics, thereby favoring facts over values—and generating “spontaneity”—in the governance of research infrastructures (e.g., by setting a quantitative target presumed to be free of value conflicts⁹) and in the steering of research.

• An instrumentalist steering of research as an innovation-oriented (R&D) or policy-oriented endeavor aligning with an engineering ethos and socio-epistemic features and practices—notably to directly and practically “impact” perceived “societal challenges” through the development of technical arrangements such as infrastructures.¹⁰

a. Behaving Like an Institution

Founded in 1999 as Revues.org by a PhD student at Avignon Université, OpenEdition began as a grassroots effort to improve the online visibility of SSH journals. Initially described as a “federation,” it gradually institutionalized over two decades. This institutionalization unfolded through (1) the progressive expansion of partnerships and funding, bringing in Centre national de la recherche scientifique (CNRS), Aix-Marseille Université (AMU), EHESS, and later support directly from the Ministry of Higher Education and Research; (2) a rebranding in 2011 to OpenEdition, reflecting a broader, international mission beyond journals to include books, event listings (Calenda), and academic blogs (Hypotheses); and (3) formal recognition as a public research infrastructure in 2016, positioning it alongside major scientific instruments such as telescopes and particle accelerators in the national roadmap. This marked a turning point as OpenEdition evolved from a decentralized network to a more structured and standardized entity. Today, despite not being a legally incorporated body, OpenEdition functions as a “joint unit” dependent on its four supporting institutions and operates under a complex governance structure in which staff are only employed by the parent institutions.¹²

This history of OpenEdition highlights the fact that the process of infrastructure building is at the same time a process of institutionalization (Zurbach 2024, 32ff.), and the consequences of this process could not be more clearly stated than by Pierre Mounier, one of the founders of OpenEdition:

As these infrastructures scale, however, they progressively enter a new dimension, moving further from the initial community-focused logic centred on a single or limited set of journals. Increasingly, infrastructural growth brings about constraints associated with human resource management, large-scale financial sustainability, technical maintenance, and user relations, as users begin to behave more as consumers than as community members. Operational streamlining, efficiency demands, rigorous oversight and, ultimately, bureaucracy, become prevalent and begin to transform the character of the infrastructure itself.

More significantly, upon reaching a certain scale, these infrastructures are compelled to develop closer relationships with external entities, such as government agencies, library consortia, and large funding bodies. These institutions impose specific requirements and often introduce their own organisational culture, shaping the infrastructure in ways that may diverge from its original communal values. Consequently, these infrastructures face moral dilemmas as they struggle to balance their scholarly foundations with an administrative and technical culture that, in some respects, contradicts the very notion of community.

Ultimately, infrastructures initially rooted in a specific scholarly community find themselves compelled to serve multiple other communities, necessitating a gradual departure from their communitarian essence in order to fulfil these broader obligations. (Gatti et al. 2025, 28–29)

Overall, OpenEdition’s case highlights the aporias of institutionalization: Increased recognition and stability come at the cost of bureaucratic complexity and limited operational independence. Furthermore, the need to conform to each institution’s norms hinders autonomy, illustrating how institutional support both enables and constrains OpenEdition’s operations. It also shapes its socio-epistemic features: With Gruson-Daniel (2018), we see that, as a public research infrastructure, OpenEdition finds its place in a broader network of national (and international) innovative infrastructures dedicated to regaining control over public knowledge production in a civic-technoindustrial regime, which, specifically in the French case, takes shape as a “public service mission.”

Although in the process of being institutionalized—that is, supported and steered by four public research institutions—OpenEdition is not itself an institution (Tournay 2011): (1) from a historical perspective, its short history, rebranding, and multiple reorganizations attest to a continuous adaptation to an ever-moving “ecosystem” rather than a fixed point in a landscape; (2) from a legal perspective, OpenEdition has no legal personality, which means it cannot manage its own finances or employ its own staff, nor can it be held responsible for civil or criminal infractions; (3) from the perspective of its scale, although the infrastructure staff grew from 49 to 70 over the last 10 years, the vast majority of its agents are employed on fixed-term contracts that legally cannot cumulate more than six years, leading to an important turnover rate with a mean duration of stay of little more than three years¹³; and (4) finally, from the perspective of its public recognition, surveys and interviews led by members of the Lab with the research and publishing communities showed that only backend users—that is, editors and publishers—unequivocally and explicitly recognize OpenEdition as a public infrastructure and know about its missions and functioning. Other users, in an answer to a survey or a response to an interview on the matter, frequently mistake it for other platforms—either CAIRN or Persée—or even with open access at large.¹⁴

Yet, despite this state of affairs, one way for OpenEdition to accommodate all the tensions and pressures exerted by its many stakeholders at the crossroads of scientific institutions and policies, scholarly communities, other infrastructures, and its own staff is to behave like an institution. Indeed, only by setting itself apart from the vagaries of time and outside the socio-political mundanities of worldly affairs can OpenEdition exhibit the ideal-typical properties of an infrastructure—that is, to perform its technical function in a transparent and efficient manner, becoming “visible [only] upon breakdown” (Star and Ruhleder 1996).¹⁵ This points to one of the socio-epistemic features manifested by the infrastructure: Its functioning is grounded in a radical distinction between technology and politics, or between facts and values, and thus between workers and individuals.¹⁶

b. Epistemic Frictions

In the course of OpenEdition’s history, and under the aegis of its founder and first director, Marin Dacos, a research service was founded within the infrastructure: OpenEdition Lab. During its first phase (2011–2018), OpenEdition Lab was intended as a space to design, prototype, and implement data-scientific tools, such as for reference detection, resource recommendation, named entity recognition, or log analysis:

OpenEdition Lab is the research, development, and innovation cell of Cléo,¹⁷ which proposes to develop and experiment with a range of tools exploiting the possibilities offered by digital technology for collaboration and access to scientific information. It is a space for research and innovation around the search for information in these different deposits [gisement]: books, journal articles, research notebooks, and scientific events.

The aim is to enable the deployment of innovative services, through new devices facilitating the dissemination of scientific information, as well as the implementation of research and development programs aimed at interlinking all the documentary resources disseminated (articles, chapters, etc.).¹⁸

Initially funded by two Google grants for the digital humanities, OpenEdition Lab was populated with engineers and developers, as well as PhDs and postdocs in information technology (IT), data science, and software engineering. As clearly stated in this excerpt, this innovation-oriented R&D service is grounded in a form of technological determinism for which technology comes first and “offers possibilities” that researchers and engineers have to seize in order to deploy services. In this (spontaneous) epistemology, “scientific information” (books, articles, notebooks) becomes the natural resources that need to be mined. Embedded in a broader agir ingénieur that pervades the whole infrastructure, we can therefore situate the first phase of OpenEdition Lab at the crossroads of the “institutional scientific humanities,” an instrumentalist steering of research, and a somewhat data-oriented conception of science.

In 2023, after a few years of inactivity, OpenEdition Lab began a second phase with an emphasis on fundamental scientific research. Since then, its aim has been to grasp and document the socio-technical dynamics of open science for SSH in all their depths and “thickness”: from individual practices to infrastructures to open science policies and philosophies of the “open” more broadly. At the intersection between STS, library and information science, and data sciences, this second phase of the Lab focuses on “research on research” and, among other issues, questions the role played by digital technology—software, platforms, infrastructures, code—in the contemporary transformation of research practices, tools, and policies. As such, although it recognizes the prescriptive and performative powers of technology, it cannot yield to a technological determinism. Rather, it is premised on a socio-technical constructivism that (1) does not sharply distinguish technology from politics, nor favors facts over values, but posits that artifacts do have politics and (2) does not conceive of technical development—such as an infrastructure—as a merely “natural” or “spontaneous” endeavor, but rather claims that it results from processes of socio-political construction—even recursive processes since said technology has an effect on socio-political dynamics.

Consequently, to the complex equation of the four epistemic features that we identified in the first part and detailed in the case of OpenEdition, we must add a fifth term: a socio-technical constructivist premise. This premise is highlighted since the beginning of the new phase of the Lab but is nonetheless shared by other actors—both individuals and organizations—across and beyond the infrastructure. While behaving like an institution partly allowed the infrastructure to accommodate some tensions, year in, year out, the irruption of this fifth term generates epistemic frictions wherever constraints collide and pressures build up.¹⁹ These epistemic frictions, which can either be made productive or result in deadlocks, may manifest as:

• Frictions between professional cultures: for example, between the symbolic economy of scholarly authorship that requires individuals to be recognized as knowledge producers in a broader community—and, more largely, OpenEdition Lab as a fundamental research team rather than the R&D branch of the unit—and the corporate communication strategy for which the infrastructure speaks of in its own depersonalized and disembodied voice.

• Frictions between modes of governance: notably between forms of collegiality or collective decision-making necessary for the social construction of facts, horizontal management that relies on third-party actors such as consulting firms to externalize the processes if not the responsibility in collective decision-making, and a vertical hierarchy emanating from more powerful stakeholders in a dissymmetric ecosystem.

• Frictions between facts and values: or, rather, muffled value conflicts that cannot find a space to be debated and become productive, for they are usually discredited as personal opinions and eschewed in technical rationality by supposed facts.

c. The Becoming-Engineering of Research

One last specific and prevailing friction can be found in the tension between fundamental research as a primarily descriptive practice, on the one hand, and innovation- or policy-oriented research and development as a normative endeavor, on the other. Although no sharp distinction can definitively be made between fundamental and applied research²⁰—since the “mere” description of a phenomenon may be intended to have some effective and concrete consequences on it—practices may radically differ depending on their final cause: whether to better understand the world by producing knowledge about it or to produce effective action that will have material consequences on it. As we have seen with Zurbach, this tension is particularly salient in the case of infrastructures: “The tension between “supporting” and “improving” [research] can be seen as a translation of the fundamental tension inherent in the process of infrastructuration” (2024, 754).

Indeed, many other infrastructures—either public or private, in France and abroad—host R&D, innovation, data, or policy “labs” dedicated to applied research in service of the infrastructure.²¹ When they do not have the luxury of an in-house research team, they may have recourse to dedicated service providers such as cooperative startups, consulting firms, or software companies.²² Although this would need a more thorough inquiry, much of the spontaneous epistemology of OpenEdition Lab in its first phase can be discerned, mutatis mutandis, in the discourse and dynamics of other such in-house R&D labs. During workshops, conferences, and encounters, I have witnessed a recurring pattern of technological determinism—which prioritizes existing state-of-the-art technical devices over actual practices, sometimes even claiming to meet “needs” that do not exist yet—accompanied by a normative discourse about what scientific research should be; what “good practices” may support this conception of science; or what scientists, as individuals, must become in this context.

This spontaneous epistemology for a technology-driven applied research—largely amplified by a knowledge regime grounded in the financialization of innovation, notably through project funding and management—enters in friction with fundamental research. And it is all the more salient when said fundamental research is premised on a critical stance on technology. Possible frictions may run along the following lines:

• Objects—frictions between the technical-rational demands of an infrastructure and “thicker,” yet infrastructure-specific, fundamental research issues such as the discoverability of open access content or cognitive justice for a fair open science.

• Scope—frictions between the role assigned to such an in-house Lab in the development of the infrastructure, reduced to prototyping mere technical devices, and the role it could have in opening spaces to debate the values of the infrastructure or the meaning of open science.

• Methods—frictions between rather formatted methods tailored for needs and impact assessments (e.g., UX/UI, A/B testing, user persona, cost-benefit analysis, SWOT analysis, indicators and metrics) and a host of both qualitative and quantitative methods for SSH (e.g., observation, participant observation, interviews, probes, archival research, surveys, network analysis).

• Temporality and Management—frictions between the longer timescales of open-ended and exploratory fundamental research and the tight planification of applied research, teleologically oriented towards concrete deliverables.

In this complex context due to the infrastructure’s spontaneous epistemology, itself amplified by a broader regime of knowledge, the research activities of OpenEdition Lab come under a set of epistemic pressures that cause frictions at certain interfaces both within and outside OpenEdition. Two main frictions have to be highlighted here: first, between the technical rationality of the infrastructure’s ideal-typical functioning and the socio-technical constructivist premise that grounds the Lab’s epistemology as highlighted above; and, second, between l’agir ingénieur that drives innovation-oriented research and open-ended, exploratory fundamental research.

Slightly expanding on this second tension to conclude this part, it is worth going back to historian of science Paul A. David’s seminal paper on “The Historical Origins of ‘Open Science.’” Adopting a decidedly presentist posture in 1991 (later enriched in 2008), he acknowledges the recent appearance of knowledge enclosures in R&D—“commodification of knowledge”—as opposed to what he then dubbed “open science.” For David—as much as for Daryl Chubin—“open science” does not merely amount to opening the access to publications and data but is, in contrast with private R&D, fundamental research funded by public patronage. It is therefore a “balancing act” of public policies between these two “systems of knowledge”:

Considered at the macro-level, “open science” and commercially oriented R&D based upon proprietary information together form a complementary pair of institutionally distinct sub-systems. The public policy challenge that needs to be faced, consequently, is to keep the two sub-systems in proper productive balance, so that the special capabilities of each may amplify the productivity of the other. But the former of these sub-systems, being based on cooperative behavior of researchers whose work is dependent on public and private patronage for support, is the more fragile of the pair; and more vulnerable to being undermined by the incursion of information disclosure restrictions motivated by the goal of privately appropriating rents from possession of new scientific and technical information. (David 2008, 4)

While dividing lines have shifted since then—as there are private firms that produce open knowledge while publicly funded research can be privatized and commercialized—fundamental research, or “open science” in David’s terms, still seems to be the poor relation of scientific policies and infrastructures, especially in SSH.

a. Being Inside and Outside Openness

A host of different actors, individuals and organizations, produce discourses about open science. That said, few among them take it as the object of a critically informed fundamental research. For Nathanael Tkacz, the only way to distinguish oneself from institutional entrepreneurs, promoters of the open, or compliance auditors, is to “leave the rhetoric of open behind”:

There is a need to look more closely at the specific projects that operate under its name—at their details, emergent relations, consistencies, modes of organising and stabilising, points of difference, and forms of exclusion and inclusion. If we wish to understand the divergent political realities of things described as open, and to make visible their distributions of agency and organising forces, we cannot “go native,” as a young, anthropologically-minded Bruno Latour once wrote, meaning that we cannot adopt the language used in the practices we wish to study. To describe the political organisation of all things open requires leaving the rhetoric of open behind. (Tkacz 2021, 404)

Yet there is an urgent need precisely to fully grasp “the language used in the practices we wish to study” to understand not only what it means but in what ways it articulates with said practices as much as with “dispositifs and events, artefacts and experiences” (Labelle 2020b, 17) and how this assemblage constitutes a genuine policy or infrastructure. This, I believe, and more particularly in my own case, implies to somewhat concur with the rhetoric of open, at least temporarily or alternately.

This is precisely what Labelle aimed at as an embedded researcher constantly reappropriating her fieldwork, the concepts, and the people involved. The “permanent shuttling back-and-forth between the space of inquiry and the constitution of the research object” (Labelle 2020b, 19) is what she calls “multipositionality”:

A double constraint arises: you have to be spontaneous and free, but at the same time contribute (actively) to getting the message across. This contributes to the creation of a disjunction between the communicative ideal being promoted (transparency) and the conditions under which one’s point of view is exercised. Being an embedded researcher does not imply demonstrating “you do what you say,” which is unfortunately the prerogative of many surveys that adopt a comprehensive sociology of agreement to the values and norms of openness. The result is the creation of an in-between situation: it becomes necessary to situate oneself in the interstice in order to hold the inquiry and still be able to produce knowledge. (Labelle 2020a, 84)

Writing in English is not the only trick to inhabit interstices. Professional trips—for teaching, conferences, or merely remote working—are also one way for the researcher to be “multipositioned.”

But, more importantly, the “discomfort of the field” is for ethnographer Martin de La Soudière (1988) a condition of its felicity. Thus, frictions can be seen as ways to maintain the researcher in such a liminal space, both or alternately inside and outside openness, that needs to be constantly adjusted.²³ Indeed, frictions—either observed, experienced, or even provoked—can be used as methodological tools to maintain critical distance while objectifying it and, in doing so, shedding light on professional practices, the organization’s functioning, diverse conceptions of research, or competing philosophies of the open. Eventually, one way to be both inside and outside openness is precisely to look the rhetoric of open in the eye, together with its actors.

b. When Boundary Objects Collapse

“Open access,” “open science,” “ecosystem,” “data,” or “research”—a panoply of terms and expressions circulate within the infrastructure and beyond it through its interfaces with other organizations. These terms, widely mobilized yet seldom explicitly and unequivocally defined, seem to bear many meanings according to their context of utterance or inscription, either in a project proposal, a general assembly, or an activity report. Building on the work of Susan Leigh Star, James R. Griesemer, and Geoffrey C. Bowker, we can conceive of these expressions as verbal “boundary objects”:

We define boundary objects as those objects that both inhabit several communities of practice and satisfy the informational requirements of each of them. In working practice, they are objects that are able both to travel across borders and maintain some sort of constant identity. They can be tailored to meet the needs of any one community (they are plastic in this sense, or customizable). At the same time, they have common identities across settings. This is achieved by allowing the objects to be weakly structured in common use, imposing stronger structures in the individual site tailored use. They are thus both ambiguous and constant; they may be abstract or concrete. (Bowker and Star 1999, 16; see also Star and Griesemer 1989)

Boundary objects are essential to the dynamics of openness. Indeed, while studying the dynamics of knowledge production on Wikipedia, Nathanael Tkacz highlighted the importance of “boundary statements” (Tkacz 2015, 81), defined as discursive objects designed to bridge the many competing, contradictory, even irreconcilable frames involved in a dispute about an article. For Samuel Moore, the vitality—undecidability, even messiness (Adema 2014)—of open access is premised on it being a potent boundary object: While “theorising it as a boundary object allows us to conceptualise OA as a community-led process without fixed meaning and continually open to interpretation,” he warns us against the “enclosure” of such boundary objects by policymakers, emptying them of their power to bind diverse communities of practice (Moore 2017). Finally, Gruson-Daniel, drawing on Alice Krieg-Planque’s work, conceived of “the open” as a formula, a “verbal sequence,” easily recognizable and relatively stable (Gruson-Daniel 2018, 32). By circulating, a formula becomes a condition of possibility for collective negotiation: “Its very use contributes to the dynamics of the debates associated with it, associating it with a given meaning” (33).

While it could be argued that “open science” is a well-defined concept thanks to a variety of policy frameworks and institutional documents, we cannot take for granted that, at the level of the infrastructure, there is no definitional ambiguity and play around it. Indeed, some agents at OpenEdition do not really care about open, and for them working for OpenEdition is “just another job”; for others, who have been here a long time, open science has a protracted history with shifting meanings, and it is uncertain whether the last official meaning is the best suited for their praxis; others relate more to ideals of free software, which very seldom appears, if at all, in its ideological complexity in current definitions of open science; others yet relate more directly to citizen and participatory science, which may appear in the UNESCO definition but are missing from the official French one. Hence, the infrastructure cannot be reduced to a mere facilitator for the enforcement of higher policies and definitions, and its agents do not merely “enact” one of the institutional or political definitions that are out there on the market. Rather, they partake, however discretely, in the defining process, starting with the appropriation of a, sometimes contradictory, diversity of definitions that they confront with their own values and professional praxis. Therefore, not taking for granted the institutional or political definitions and not assuming that the definitional issue is settled, the infrastructure appears as a place where definitions are constantly renegotiated in practice rather than merely enacted from above, as a nexus where what open science means could be negotiated by confronting policies with practices. Being essential to the dynamic of all things open, boundary statements, boundary objects, or formulae are privileged observatories for the study of open science and its negotiations. Beneath the potential enclosures that policymakers generate by forcing boundary objects to precipitate a legal definition, there are, within infrastructures and at their interfaces, many fleeting moments when boundary objects temporarily collapse. The ephemeral collapse of a boundary object can become the locus of a discrete moment of negotiation about what open science is. Whenever, in a given context, one or several actors are faced with the necessity to clarify, in part or in whole, a boundary object, the technical rationality of the infrastructure shatters and reveals the value conflicts, the epistemic frictions, and the many tensions that inhabit and move it.

There are at least two ways to seize such moments when a boundary object temporarily collapses: either to witness these fleeting moments happening in a “natural” setting—a meeting (and there are many), a seminar, an exchange of emails, a workshop, a jury—or to spark them off in an experimental setting, such as a survey or a semi-directed interview. In the first case, ethnographic observation conducted on-site allows the researcher to document discrete moments when boundary objects—such as “open science,” or even “research” in the case of the present contribution—are being negotiated. In the second case, such collapses can be intentionally provoked in surveys and interviews where the respondents are explicitly asked to define, either freely or choosing among a set of inclusive or exclusive possibilities, the meaning of a given expression, how it describes their practice, in what ways it is related to other terms, in what context they mobilize the term, and whether they identified competing or contradictory uses.²⁴

In between “natural” and experimental settings, some interventions in the socio-technical development and maintenance of the infrastructure may parallelly serve as ethnographic probes—that is, moments of active participation in the life of the infrastructure in contexts the researcher is personally involved in.²⁵ These interventions serve as tests (épreuves), in the sense of French pragmatic sociology, to explore the room available for critical thinking about open science within the development of infrastructures, or to assess to what extent it is possible to question the very values enacted by those infrastructures. Probing the infrastructure in this manner sheds light on the “value conflicts” (Cohoon 2024) that are muffled by its technical-rational functioning and shows how, in this context, what “open science” means is continually, yet fleetingly and discretely, negotiated among actors in dissymmetric situations.

c. In media res

Observing, interviewing, or probing raises important methodological questions both in epistemic terms (what phenomena and inscriptions thereof may qualify as valid research material) and in ethical terms (how to not betray the researchees/colleagues by informing and involving them rightly). These socio-epistemic issues immediately forbid two kinds of approaches: the extractivist and the “covert” approaches. First, the field and the people involved are not natural resources to be mined as articles, books, and blogs were during the first phase of OpenEdition Lab: “The social world is not constituted as a ‘deposit [gisement]’ from which the researcher draws objects for analysis, but as a space to which the researcher belongs and in which meaning and values are manifested, circulated, and modified for and by each of the protagonists involved. This approach has led me to accept proximity to the situations observed, and to live certain experiences with complete involvement. This is reflected in the difficulty of reporting on them for oneself or for others, or of ‘objectifying’ what was happening” (Labelle 2020b, 20). Second, while certain locked or shrouded fields may justify a “covert inquiry”—that is, without informing the researchers—in order to “unveil the objective structures” (Bourdieu, in Rowell 2022) of an organization, my fieldwork is premised on the pragmatist stance that the infrastructure’s agents act according to a plurality of “modes of engagement” that need be credited and understood as such rather than as the manifestation of some other hidden and biased rationale of which they would be unconscious.

Consequently, as both a colleague and an embedded researcher, in order to witness the collapse of boundary objects and take part in the life of the infrastructure, it is necessary for me to be in media res—that is, according to Labelle again, “Being embedded was not only to acknowledge that I was not master of the surveyed spaces and that I was not in a position of domination in relation to the people with whom I found myself, but also to inscribe myself in the discursive and pragmatic space of the survey sites, and thereby to signify the recognition of certain logics of action [signifier la reconnaissance de certaines logiques d’action]” (Labelle 2020b, 20). More importantly, this leads to acknowledgment that “the researcher is not the only person to produce knowledge, while nonetheless having the prospect of producing knowledge other than that recognized in the professional field under observation” (Labelle 2020a, 86).

Notably, in the case of OpenEdition and its Lab, it is of the utmost importance for the researcher to acknowledge and legitimize the specific expertise the infrastructure’s agents developed in the exercise of their profession. Often disqualified, delegitimized, and discredited as personal opinions or political endeavors—at the risk of turning open science into “institutional knowledge [savoir d’institution]” (Laboulais 2023)—these many expertises result from daily use, maintenance, and development of the infrastructure as a fully-fledged socio-technical device that is also in constant contact with the many different communities of users as well as with other national and international organizations. Eventually, researching in media res, sometimes to the point of “forgetting about the inquiry [l’oubli de l’enquête]” (Labelle 2020a, 80), implies co-constructing the knowledge about the infrastructure together with its agents.

The term “open science” thus appears to be a boundary object, or an expression used by a wide variety of actors, without any single, unambiguous definition being shared by them. It is this vagueness, which is sometimes deliberately maintained, that makes it possible to coordinate, in a particularly dynamic context, the collective action of extremely diverse and varied stakeholders—sometimes very difficult to grasp (think, for example, of non-academic users of open access scientific content). This vagueness can become frictional when these same stakeholders are called upon to define, defend, and update their own understanding of “open science” and the value system associated with it.

The role of the embedded researcher, in media res, is to co-construct knowledge about the infrastructure together with its agents. While Labelle calls this research posture “participant interpretation” (Labelle 2020b, 28), I would rather call it “observant participation”: “Where participant observation presents more opportunities for mobile positioning, outward gazing, and inscription, observant participation presents more opportunities for fixed positioning, inward gazing, and incarnation” (Seim 2024). Here indeed, rather than moving from one position to another, I am fixed as coordinator of the Lab, “embedded in labor, under and against management” (Seim 2024). I take an inward gaze as I confront my own socio-epistemic features with the infrastructure’s and am an embodied individual focusing on incarnated reactions alongside inscriptions.