In the digital humanities, understandings of what counts as “data” are often shaped by models borrowed from STEM fields: precisely structured, de-personalized datasets prepared for standardization, sharing, and further processing. Such models can be useful when standardization and reuse are the main goals. However, they also create a practical and conceptual gap between researchers working with situated ethnographic records and institutional actors expected to describe, store, and make such records available. Ethnographic materials—interviews, notebooks, recordings, photographs, sketches—do not always fit into this framework. They are open to different interpretations, shaped not only by what was recorded but also by fieldwork encounters, relationships of trust, and layers of meaning. For this reason, such materials pose particular challenges within frameworks such as FAIR, GDPR, and open science mandates.

In the focus of the paper is precisely this transition: ethnographic materials are increasingly understood as “data,” yet what makes them ethnographic—their connection to encounters, relationships, trust, silences, circumstances, and layers of meaning—resists any simple translation into the language of data management. Therefore, we approach ethnographic data as small, soft, and situated, and ask what happens when such materials enter legal, administrative, and infrastructural systems that were not designed with the intention to accommodate their complexity. In doing so, we distinguish between materials that are small because they emerge from the slow, situated, and relational logic of ethnographic research, and materials that become small because they remain insufficiently described, only partially processed, incompletely archived, or not fully visible within institutional systems.

We first present the policy environment in which humanities researchers are increasingly encouraged—or expected—to “manage” and “open” their data. Many policy documents acknowledge disciplinary differences but still assume a largely generic, decontextualized notion of data. FAIR has become a widely accepted standard, yet its application often presupposes stable data units and clear pathways for reuse (Wilkinson et al. 2016). Although social science data infrastructures have developed practices of restricted access and mediated reuse, such models cannot always be applied straightforwardly in small humanities archives. There, collections are often incomplete, built up through different institutional circumstances, while much about their creation, meaning, and possible use remains tied to knowledge that researchers and archivists carry with them, but could not always record systematically. Humanities initiatives, therefore argue for more contextual understandings of data and for workflows that are “as open as possible, as closed as necessary” (Tóth-Czifra 2019; Harrower et al. 2020), while the CARE Principles remind us that openness must be balanced with collective benefit, control, responsibility, and ethics (GIDA 2019; Carroll et al. 2020; Christen / Anderson 2019).

The paper is grounded in long-term work with the research archive of a humanities institute in Southeast Europe. Over several decades, ethnographic fieldwork carried out in and around the institute has been gathered into heterogeneous collections: field notebooks, interview transcripts, photographs, dance notations, analog and digital recordings, and their metadata. As analog methods have gradually been replaced, supplemented, or simply overlaid by digital tools, institutional work has also had to absorb a new vocabulary: data management plans, repositories, open science, access regimes, and long-term preservation. However, a considerable part of this material still circulates or quietly remains on personal computers, external drives, commercial cloud services, and in other spaces that cannot be described as either private or institutional. We do not understand this merely as a failure to follow rules, but as an important part of the actual conditions under which ethnographic materials are stored, described, remembered, and eventually passed on. Rather than framing the archive as a closed case study, we understand it as a situated institutional field in which policy language, infrastructural demands, limited resources, and the ordinary work of archival care intersect, not always smoothly or without contradiction.

Methodologically, the paper works with several layers of material: formal rules and guidelines, documentation forms, orally transmitted institutional knowledge, databases and their descriptive fields, interviews, and long-term conversations with researchers, archivists, and tool developers. This layered approach allows us to follow individual projects beyond the moment of fieldwork itself: from recordings made in the field and consent agreements, through anonymization, to later attempts to store, describe, preserve, and perhaps make the materials available. We are particularly interested in decisions that are never purely technical and cannot simply be settled in advance: what makes a description sufficient or meaningful, when access should be restricted, when it becomes necessary to return to consent, and who carries responsibility as materials pass from the researcher’s working space into an archive or repository.

In the first part, we explain what it means to call ethnographic records “small data.” We distinguish between records that are small by research design and those that become infrastructurally small: insufficiently documented, only partially processed, or never fully archived. This double smallness—epistemological density on the one hand and infrastructural fragility on the other—makes ethnographic materials a useful lens for examining how current data policies privilege certain forms of knowledge, while leaving others more difficult to recognize, describe, and support.

In the second part, we examine the ethical and practical challenges of consent, anonymization, and description. We consider legacy materials created before today’s consent standards and GDPR language, as well as newer projects that experiment with more dialogical and participatory approaches. We show how anonymization can protect participants, but can also weaken the contextual density on which ethnographic interpretation depends. Instead of treating anonymization as a technical step carried out once and for all, we understand it as an ongoing negotiation involving researchers, archivists, and sometimes community members, drawing on slow archives and related approaches that emphasize care, accountability, and relationship-building (Christen / Anderson 2019).

Finally, we return to the challenging question of access. How can ethnographic materials be stored and shared without losing the context and relationships that make them what they are? Instead of simply classifying them into those that can be shared and those that cannot, we consider the space between these extremes: metadata and descriptions can be publicly visible, part of the content can remain partially hidden, access can be mediated or time-limited, and agreements with interlocutors can make future use more responsible. In this sense, “as open as possible, as closed as necessary” becomes less a slogan and more a series of situated decisions about visibility, restriction, responsibility, and reuse. Drawing on guidance from the humanities (Buddenbohm et al. 2021; Tóth-Czifra 2019), we aim to discuss how such a model of openness can be built into existing infrastructures.

The paper ends by returning to the question of what happens to ethnographic materials once they enter the language and infrastructures of data. Ethnographic field records are not simply materials that need to be ordered, described, opened, or adjusted to existing standards. They carry the notions of the situations in which knowledge was made: encounters, relationships, silences, decisions, and responsibilities. This also means that we cannot understand openness solely in terms of access, scale, efficiency, or compliance. If these become the main measures of openness, much of what gives ethnographic materials their meaning may be weakened or lost. For this reason, we argue for a slower, more careful understanding of openness, one that keeps access, context, responsibility, and long-term care together