Emergency curation as anticipatory maintenance: Lessons from the 2016/2017 data rescue movement

Abstract

This article examines how volunteers involved in the Data Rescue movement navigated concerns about the stability and stewardship of federal environmental data following the 2016 U.S. presidential election. Drawing on 16 in-depth interviews, the study explores how participants interpreted infrastructural fragility not simply as a technical risk, but as a relational and political concern rooted in shifting institutional commitments. The analysis introduces the concept of anticipatory maintenance to describe how volunteers responded to perceived threats by developing redundant, decentralized strategies for data preservation. Anticipatory maintenance is conceptualized as preventive, future-oriented infrastructural care that translates anticipated disruption into present interventions under uncertainty, in order to explain how volunteers acted as if future loss had already begun, building redundant and decentralized preservation arrangements. Findings highlight the role of particularized trust and systemic distrust in shaping grassroots responses, as well as the limitations encountered in volunteer-driven infrastructures, including challenges related to sustainability, governance, and the affective demands of ongoing maintenance. By foregrounding the interplay of technical, social, and emotional factors, this study offers a critical perspective on data stewardship in times of political uncertainty and underscores the need for collaborative approaches to infrastructural resilience.

Keywords

Anticipatory maintenance data rescue movement infrastructural fragility systemic distrust environmental data civic data stewardship affective infrastructures

Introduction

In late 2016, during the immediate aftermath of the U.S. presidential election, a wave of political anxiety swept through the scientific, archival, and civic technology communities. Among concerns that the incoming Trump administration would undermine environmental governance and public data infrastructures (Koebler, 2016), a loosely organized movement of librarians, technologists, researchers, and activists launched a series of grassroots data preservation events. Known collectively as the Data Rescue movement, these efforts aimed to identify, download, and archive publicly available federal datasets deemed vulnerable to loss of access or even deletion. The election of Donald Trump and his appointment of Scott Pruitt as head of the Environmental Protection Agency (EPA), both of whom had publicly expressed climate change denial, intensified concerns regarding the future of environmental governance and data transparency. Historically, similar political transitions had led to the undermining of environmental protections through the manipulation or suppression of scientific data, for instance, in Canada during the Harper administration between 2011 and 2016 (Kingston, 2015).

Built through academic, civic, and activist networks, and facilitated, i.e., through organizations like the Environmental Data & Governance Initiative (EDGI) and the library-driven community DataRefuge, the movement drew international media attention and widespread institutional support. EDGI is a network of researchers, technologists, and environmental advocates formed in late 2016 to monitor changes to U.S. federal environmental websites, document shifts in environmental governance, and promote public access to environmental data (Dillon et al., 2019). DataRefuge was a volunteer project initiated at the University of Pennsylvania in 2016 by librarians and academics (Janz, 2017). The Data Rescue events took place in libraries, universities, and civic tech spaces across North America. Participants articulated motivations that included civic responsibility, professional norms, and concerns regarding the continued accessibility of environmental data (Ray, 2018). These efforts can be characterized as intersecting practices of digital preservation, public protest, and infrastructural adaptation (Currie et al., 2018).

Almost a decade later, this paper presents a critical reappraisal of the first wave of Data Rescue, drawing on retrospective accounts from organizers and participants. It examines how these individuals evaluate the motivations, methods, and outcomes of their interventions in hindsight. The guiding research question is: How do participants in the Data Rescue movement reflect on their early contributions, and what do their accounts reveal about the intersections of trust, activism, and infrastructural fragility in politically contested contexts?

Central to this analysis is the concept of emergency curation, defined here as a provisional, collective practice of digital preservation mobilized rapidly in response to anticipated threats, which contrasts with established standards in archival and digital curation. Unlike professional frameworks that emphasize provenance, sustainability, and long-term interpretability, emergency curation functions primarily as anticipatory first aid: volunteers rapidly duplicate and redistribute data as a safeguard against perceived imminent infrastructural failure.

Work on anticipation highlights how institutions and communities prefigure futures through routines of preparedness and imagination (Adams et al., 2009; Clarke et al., 2015). In data-intensive settings, ethnographies trace anticipatory modes of living with digital uncertainty, anxiety, mess, and the search for trust amid that mess (Pink et al., 2018). Related scholarship shows anticipatory data practices within social movements under threat (Kazansky, 2021). This paper connects these threads to maintenance: volunteers’ speculative orientations took the form of maintenance work, thus converting affect such as worry or duty into material redundancy and governance experiments.

Building on sociological theories of trust and uncertainty, I approach systemic distrust as a productive disposition that arises when the background expectation of institutional reliability becomes unsettled but not abandoned. Following Luhmann (1979) and Giddens (1990), trust enables action under conditions of indeterminacy by bracketing the unknown, whereas systemic distrust arises when those uncertainties become specific; that is, when potential failures are named, and responsible agents identified (Sumpf, 2019). In such moments, distrust re-orients care: it transforms latent anxiety into anticipatory practices that seek to preserve continuity under threat. This dynamic is directly related to the concept of anticipatory maintenance, in which systemic distrust and imagined futures converge to produce preventive and future-preserving forms of infrastructural work.

This paper contributes to ongoing conversations on digital curation, archival studies, and information infrastructure studies through the lens of anticipatory maintenance, a proactive mode of infrastructure care that addresses not yet realized failures. Drawing on Fernando Domínguez Rubio's concept of mimeographic labor, which emphasizes the repetitive practices necessary to sustain fragile environments (Rubio, 2020, 2025) and Steven Jackson's broken-world thinking (Jackson, 2014), it highlights infrastructures as inherently precarious and continually in need of care, maintenance, and anticipatory interventions. It draws attention to the interplay of systemic distrust, ethical commitment, and affective labor that characterized the grassroots response of Data Rescue. It also raises questions about the limitations of urgency-driven interventions and the need to rethink what it means to preserve data in ways that are sustainable, usable, and politically conscious, especially with regard to recent authoritarian tendencies in the United States or during times of political crisis.

I argue that (1) Systemic distrust can be generative: it mobilizes redundancy and redistributes stewardship across civil society rather than only eroding confidence; (2) Anticipatory maintenance clarifies how volunteers acted before harm, extending maintenance beyond reactivity to include speculative, future-preserving care; and (3) urgency-driven duplication without context exposes limits of emergency curation, underscoring the need to pair availability with interpretability and governance.

Background and theoretical framework

The Data Rescue movement began on 17 December 2016 with a “Guerrilla Archiving” marathon at the University of Toronto. Approximately the same time, DataRefuge, a University of Pennsylvania collective, was hosting the so-called “Data Rescue” hackathons in the United States (Janz, 2019; Walker et al., 2018). Over the next 12 months, the two groups coordinated more than 50 community events that drew nearly 1500 volunteers. During these hackathon-style events, one of the main activities was nominating URLs to the End-of-Term Web Archive or downloading environmental datasets they feared might be deleted or altered under the new administration (Dillon et al., 2017). However, the availability of skilled metadata curators did not keep pace with demand. Crucially, the fact that most federal datasets ultimately remained accessible underscored the limitations of a copy-everything approach (Allen et al., 2017a; Lamdan, 2017).

Digital curation and preservation standards such as the Open Archival Information System (OAIS) cast duplication as only the first step in a life-cycle that also requires provenance, format migration, and community engagement to keep data intelligible (ISO, 14721:2025, 2025). While the movement did not fully meet these professional standards, it nevertheless achieved several important outcomes, including increased media attention, the exposure of structural weaknesses in federal stewardship, and the development of alliances between activists, librarians, and critical data scholars (Allen et al., 2017b; Currie et al., 2018; Currie and Paris, 2018; Vera et al., 2019). Subsequent work by EDGI and allied projects has emphasized that while redundancy is valuable, it is not a substitute for long-term digital curation.

Data Rescue can be framed as a form of data activism that aimed to protect threatened knowledge and reconfigure control over information infrastructures (Currie et al., 2018; Dillon et al., 2017; Paris, 2017; Paris et al., 2017). Currie et al. (2019) conceptualize the practices of the movement as relational infrastructures shaped by urgency, geography, and sociotechnical politics, pointing out that data and archiving are never politically neutral. Data Rescue's open-source, non-hierarchical organization reflected its commitment to participatory stewardship and decentralized governance of public data.

Public data infrastructures have long been considered essential components of democratic governance, scientific accountability, and civic participation (Edwards et al., 2007; Janssen et al., 2012). Infrastructures are not neutral backdrops but sociotechnical systems shaped by values, power, and history (SL and Ruhleder, 1996). As Star (1999) argues, the visibility of infrastructure often emerges at moments of breakdown, making them particularly legible during crisis events, such as the political rupture triggered by the 2016 U.S. presidential election. Recent work in digital sociology has also studied the circulation of emotions in and through data infrastructures, conceptualized as affective infrastructures (Bosworth, 2023; Piattoeva and Saari, 2022; Zembylas, 2023). These studies show that infrastructures are not emotionally neutral but channels that carry and amplify feelings such as anxiety, excitement, and fatigue. Integrating affect into theories of maintenance clarifies both the motivations underlying volunteer labor and the constraints that emerge when affective resources are exhausted.

Infrastructure studies have shifted analytical attention from the moments of invention to the ongoing work that keeps sociotechnical systems functioning. Early ethnographies highlighted the invisibility of maintenance within information infrastructures (Karasti et al., 2010; Star, 1999). Jackson (2014) formalized this concern under the rubric of broken-world thinking, arguing that repair is constitutive of infrastructural life. Mattern (2018) subsequently extended the lens to maintenance, conceptualizing it as a mode of care that foregrounds material vulnerability and temporal openness. This practice can be framed as maintenance, repair, or care that is not purely reactive to errors or damages, and focused on restoration, but that mends and remakes with a focus on continuity and change (Harb, 2025).

Against this backdrop, grassroots responses like the Data Rescue movement can be understood as infrastructural interventions. Their aim was to protect the continuity of epistemic systems perceived to be under threat and not just to preserve data. These practices have been analyzed within the literature on archival activism, which frames preservation as a political and ethical act (Caswell and Cifor, 2016; Cifor et al., 2018; Vukliš and Gilliland, 2016). Archival activism resists the assumption that archives are static repositories, highlighting instead their performative and future-oriented functions. This study also builds on recent critiques of open government data and connected assumptions about transparency. Scholars have increasingly questioned whether technical access alone can ensure meaningful reuse, especially under conditions of inequality and epistemic injustice (Jasanoff, 2017; Johnson, 2014). As participants in this study reflect, saving data is not sufficient if the relational, contextual, and instrumental dimensions of access are absent. These critiques inform the analytical lens through which the first iteration of Data Rescue in 2016/2017 is examined.

The dynamics of trust and distrust are central to understanding the motivations and outcomes of the Data Rescue movement. System trust, as distinct from interpersonal trust, refers to the confidence actors place in institutional procedures and public infrastructures (Luhmann, 1979; Sumpf, 2019). Classical sociological accounts distinguish interpersonal trust from trust placed in abstract systems and expert institutions (Giddens, 1990; Luhmann, 1979; Möllering, 2006). Abstract systems are embodied in modern institutions structured by routines and regulations, contributing to “ontological security” (Giddens, 1990: 92), according to Giddens, a precondition for trust. Routines and regulating mechanisms strengthen ontological security, establishing trust in institutions as abstract systems. For Giddens, trust is implicated by an individual's ontological security, the concept that Giddens uses to represent an individual's confidence in their social identity, their situation, and how to proceed with it (Giddens, 1990).

Trust at the system level stabilizes social action by rendering complex infrastructures intelligible and reliable. In contrast, systemic distrust emerges when actors perceive institutions as either unable or unwilling to honor their delegated responsibilities. This closely parallels the distinction between particularized and generalized trust as articulated by Freitag and Traunmüller (2009). In their view, particularized trust refers to confidence in known individuals or close networks, and generalized trust extends to anonymous others and the impersonal functioning of institutions. In this paper, I use systemic distrust as a form of generalized distrust, characterized by the erosion of trust in abstract systems such as government agencies or data infrastructures, while particularized trust in specific actors (e.g., career civil servants or colleagues) may persist. Importantly, systemic (generalized) distrust does not necessarily lead to disengagement. Instead, it can motivate collective action and the formation of alternative infrastructure arrangements, such as decentralized and redundant data stewardship. This perspective reframes distrust as an active, generative orientation toward infrastructure, anticipating possible breakdowns, and supporting preventive interventions.

Giddens sees ontological security as the taken-for-granted confidence that modern institutions, expert systems, and abstract tokens guarantee stability and trust. Rubio (2025) challenges this theory of modernity, arguing that the progressive narrative is both Western-centric and blind to the material earth that underpins it. “The earth happened,” Rubio writes (p. 5), exposing how presumed security can buckle under climate change and other planetary forces. In place of a linear tale of prosperity, he proposes thinking with fragility: not as a defect, but as a relational condition and a productive space that offers opportunities for alternative means of maintenance and repair. Stability and security, like fragility, are therefore contingent achievements; system-level trust must likewise be understood as provisional and continuously renegotiated, rather than as a stable backdrop to social life.

Critical data studies tend not to use the vocabulary of trust directly. Instead, it analyzes how data imaginaries, collectively produced visions of what data can achieve, shape the legitimacy of data infrastructures (Iliadis and Russo, 2016). Imaginaries operate through narrative, visualization, and material practice; they can sustain confidence or nurture skepticism. System trust through this lens can be described in terms of how confidence is assembled, contested, and potentially fractured around data infrastructures. I treat distrust not as an absence of trust, but as an imaginary of negation: a shared cognitive and affective orientation that anticipates failure, capture, or manipulation of data systems.

In this paper, I draw particularly on the concept of anticipatory maintenance, a form of care that proactively addresses expected future failures rather than reacting to existing breakdowns. Anticipatory maintenance differs from conventional maintenance in its speculative orientation toward potential, rather than actualized, infrastructural fragility. This concept is complemented by the Rubio (2020, 2025) notion of mimeographic labor, the repetitive practices through which actors continuously stabilize and preserve vulnerable infrastructures, preemptively ensuring continuity. While mimeographic labor broadly encompasses ongoing acts of stabilization, anticipatory maintenance specifically emphasizes actions taken in expectation of future threats, highlighting the forward-looking, affective, and speculative nature of emergency curation. Anticipatory maintenance, as used here, refers to preventive and future-oriented practices of infrastructural care through which actors translate anticipated disruption into present interventions that sustain continuity under conditions of uncertainty.

Anticipation, as theorized by Adams et al. (2009) and Clarke et al. (2015), describes the temporal and epistemic labor through which futures are made actionable in the present. Rather than prediction, it is a regime in which imagined futures exert a pull on the present, organizing obligations to prepare and mobilizing affects such as anxiety and hope (Adams et al., 2009). It proceeds through simplification, the selective narrowing of attention to specific imagined threats, and abduction, the inferential leap that links uncertain signs to plausible futures (Clarke et al., 2015). This resonates with what Kazansky (2021) terms anticipatory data practices as collective routines of pre-emptive action under perceived threat, which likewise translate diffuse uncertainty into structured preventive care. Reasoning under uncertainty mirrors the logic of maintenance: both are oriented toward potential breakdowns and mobilize care to sustain continuity. Anticipatory maintenance, therefore, marks the moment when these temporal and epistemic orientations materialize in practice. In the case of Data Rescue, imagined futures of data loss were abductively inferred from political cues, then simplified into actionable tasks such as scraping, mirroring, and packaging. Infrastructural care thus became a means of coping with systemic distrust by folding speculative futures into present labor. As Jackson (2014) and Mattern (2018) suggest, maintenance is always a future-oriented form of care; it “keeps the world running” through repetition, vigilance, and adjustment. But fragility itself can be a productive condition that invites sustained attention rather than denial (Rubio, 2025). In this way, anticipatory maintenance captures how actors translate the epistemic operations of anticipation and the affective energies of distrust into ongoing embodied practices of care that merge the future and the present.

I refer to these practices as emergency curation, a situated manifestation of anticipatory maintenance that emerged under conditions of systemic distrust and political uncertainty. While professional digital curation models such as the OAIS Reference Model (ISO, 14721:2025, 2025) or the DCC Lifecycle Model (Higgins, 2008) emphasize long-term governance and procedural stability, emergency curation operates within a compressed temporal horizon. It translates anticipated disruption into immediate, distributed care, privileging responsiveness over durability. Compared to institutional standards, these volunteer-led practices prioritized redundancy and responsiveness over formal sustainability, functioning as temporary infrastructures designed to keep endangered data accessible until stability could be restored. Emergency curation thus provides a preliminary safeguard, deferring more comprehensive curatorial work to a future point when institutional stability may be restored. As such, emergency curation exemplifies how the imagined future becomes folded into the labor of the present: an affectively charged, collectively organized form of anticipatory maintenance through which systemic distrust is not just a symptom of crisis but a temporal practice of care, sustaining infrastructures by rehearsing their potential loss.

Overall, these dynamics reveal how infrastructures of data stewardship are shaped not only by technical protocols but by affective anticipations of failure and trust, illustrating how civic actors negotiate vulnerability in the politics of data infrastructures.

Methodology

This study adopts a qualitative case study design (Merriam, 1988; Stake, 2010), focusing on the first iteration of the Data Rescue movement in 2016/2017 and the formation of EDGI as a socio-technical response to infrastructural vulnerability. This methodological approach enables the examination of contemporary phenomena within real-life contexts, particularly where the boundaries between the phenomenon and its environment are fluid and intertwined.

The core empirical material consists of 16 semi-structured interviews with individuals directly involved in Data Rescue and/or EDGI, conducted between January and March 2025. Participants were identified through purposive sampling based on documented involvement in organizing, attending, or supporting Data Rescue events or participating in other EDGI projects. Attention was paid to disciplinary, institutional, and experiential diversity, resulting in a sample that includes librarians, environmental scientists, legal scholars, technologists, grassroots organizers, and civic data activists. 11 out of 16 interviewees were presenting as female and five as male. All of them had an academic background and were in different stages of their career. Two interviewees were Canadian, the others from the United States.¹

Interviews were conducted between January and March 2025. They lasted between 35 and 60 minutes and were held via a secure video conferencing platform. All interviews were recorded, transcribed, and anonymized. Random numbers were assigned to interviewees for all data processing steps during this process. Informed consent was obtained from all participants. Interviewees were asked to reflect on their involvement with EDGI and/or the Data Rescue movement, their motivations for participation, and their views on the outcomes and legacy of these efforts. Questions also explored broader themes such as perceptions of public data infrastructures, risk assessment, and trust and distrust in government data stewardship and federal data infrastructures.

The researcher's positionality as an information scientist with a background in digital curation was explicitly acknowledged and reflexively considered throughout the research process (Berger, 2015). Similarly, the peculiar timing of the interviews, conducted during the first weeks of the second Trump administration, was reflected during both the interviewing process and the analysis, especially taking into account the retrospective and reflexive nature of memory under conditions of recurring crisis (Berger, 2015; Gilliland and Caswell, 2016).

Data were analyzed using thematic analysis (Braun and Clarke, 2006) supported by MAXQDA. The analytical approach combined inductive coding with deductive insights informed by the theoretical frameworks outlined above. This hybrid approach facilitated both the identification of emergent themes and theory-driven interpretation. Codes were developed iteratively, beginning with line-by-line open coding, followed by axial coding to group themes, and finally by selective coding to identify overarching narratives. Core analytical categories presented in this paper included infrastructures, government data, affective reactions, personal motivations, challenges and impacts and systemic vulnerabilities as well as systemic trust and distrust.

Building on trust literature, particular attention was paid to how interviewees articulated trust and distrust in data infrastructures, not only in terms of perceived reliability of federal data stewardship, but also in relation to the grassroots infrastructures they themselves created. Based on systems theory (Luhmann, 1979; Sumpf, 2019), the analysis distinguishes between interpersonal trust (e.g., between collaborators), institutional trust (e.g., in libraries or universities), and systemic trust (e.g., in federal data infrastructures and programs).

This research also draws from feminist epistemologies and STS-informed approaches to infrastructure studies (Bowker et al., 2010; SL and Ruhleder, 1996), allowing for an interpretive understanding of infrastructure as both material and social. The inclusion of affective and ethical dimensions, such as fear, care, solidarity, and burnout, was informed by archival theory and literature on activist scholarship (Caswell, 2014; Gilliland and Caswell, 2016; Murphy, 2015), which emphasizes the politics and relational labor embedded in archival work.

Findings

Anticipating infrastructural loss

Data Rescue participants perceived federal data as precarious and began anticipating failure prior to any actual loss, acting as though critical datasets were already at risk of erasure. One college professor recalled walking home after the first post-election class and asking themselves, “I wonder if I’ll be able to do this class next year?” because Energy Information Administration tables might disappear under a Trump administration. Another librarian remembered a group of students coming to them immediately after the election, because they were worried that the kind of data they needed would soon not be available anymore. Another interviewee was a scholar who had just written a book on public access to science about climate change data and felt like all the work they had done on this topic was suddenly at risk.

This prospective anxiety was catalyzed by Canadian precedents. Interviewees recalled the immediate post-election period as filled with uncertainty, fear, and urgency. Several interviewees (I_79, I_56, I_13, I_08, I_43, I_69) explicitly drew comparisons to Canada's conservative Harper administration, which had systematically suppressed environmental data and shut down public scientific research. One interviewee (I_43) recounted how they had heard about decisions by the Harper government to “haul environmental records (…) into a dumpster and set [them] on fire,” an image that “really shook me up” and framed data loss as a politically plausible scenario in the United States. Another Canadian interviewee detailed how Harper “shut down polar research (…) destroy[ed] libraries and shred[ded] documents,” experiences that made Canadians “more alert to the possibility of what might happen under Donald Trump” (I_79). These experiences provided a historical reference point for anticipating data suppression or manipulation. This anticipation did not require proof; it was perceived as plausible and imminent. Librarians sent similar warnings to students, emphasizing that “government information disappears from the Internet every four years (…) that's totally a real concern,” (I_63) especially after the move from the physical depository-library system. Together, these memories and professional judgments created a shared perception of imminent loss that spanned disciplinary and geographical boundaries.

Practicing loss in advance reframed fragility as relational rather than intrinsic. The datasets themselves had not changed, but volunteers experienced them as precarious because political stewardship no longer appeared trustworthy. In this way, downloading, copying, and self-storage became forms of anticipatory maintenance, first-aid labor performed before any removal. In this way, systemic distrust acted as a productive force: directed at top-level political leadership rather than career civil servants, it mobilized collective action to redistribute confidence through redundancy rather than eroding it.

By making potential erasure thinkable, the election rupture inaugurated a practice-based politics in which fragility was diagnosed, trust recalibrated, and maintenance scripts rapidly invented.

Constructing redundant infrastructures: redistributing trust through multiplicity

In response to declining trust in institutional custodians, Data Rescue volunteers reconceptualized durability in terms of dispersion. Core technologists “built the plane while we were flying it (…) using the Interplanetary File System to move data, to stash data all over the place, with redundancy, so that it could never be deleted” (I_43). By routing copies through a content-addressable, peer-to-peer network instead of institutional mirrors, they made persistence a property of multiplicity rather than of any single archive. At the same time, newcomers without archival credentials adopted duplication as everyday insurance: they “just knew that stuff could disappear off the internet” and therefore scraped HTML tables, PDF reports, and FTP directories indiscriminately (I_63). Accessible web-scraping techniques facilitated the diffusion of redundancy practices, transforming them from specialized approaches into widespread civic strategies.

As the movement went on, redundancy evolved into continuous monitoring. EDGI's website monitoring team automated the detection of changes in the climate and energy pages, then worked with journalists to contextualize each change, developing scripts to detect and report changes that kept federal infrastructures publicly accountable (I_75). This coupling of machine vigilance and media translation converted static mirrors into an early-warning system attentive not only to disappearance but also to subtle discursive erasures. Parallel innovations in workflow design scaled these practices. The in-person Data Rescue events adopted assembly-line logics: organizers “assigned a portion of the web to a group,” sent crawlable content straight to the Internet Archive via browser plug-ins, and formed teams to hand download interactive, “uncrawlable” tables while recording enough provenance that “another person could look at it” (I_80). Metadata work, often sidelined in crisis settings, thus moved to the center of redundancy.

Interviewees explicitly linked these experiments to a longer lineage of archival activism in which communities “crawl existing data and mirror it (…) to draw awareness to issues around public access” (I_77). Redundancy also served a pedagogical function, introducing newcomers to the concept of stewardship as a collective obligation, often anchored in a more generalized distrust of state custodians. Taken together, technological improvisation, vernacular scraping, automated monitoring, modular workflows, and activist memory reframed maintenance as a distributed process of safeguarding. Instead of strengthening a central core to combat fragility, it was addressed by creating multiple redundancies, transforming systemic distrust into a politics centered around replication.

Emotional energies and their limits: affective arcs of emergency curation

The moral shock that fueled Data Rescue also exposed the fragility of a movement fueled by volunteer affect: adrenaline, solidarity, and care initially sustained anticipatory maintenance, yet the same energies later turned into frustration and burn-out.

Affective responses to the political climate consistently emerged as key motivators among interviewees, with many describing the 2016 election as a shock that generated a sense of urgency and a desire for agency. One interviewee recalled: “But fundamentally it was like, this is a way, you know, in feeling, and sort of channeling the shock, frustration, worry, that a lot of us were facing at that moment. This is a way to do something; that felt like we were actually doing something” (I_33). Events thus functioned as spaces of “collective therapy” (I_79) or “psychological hygiene” (I_56), offering structure in a time of political disorientation. As one interviewee put it: “I think people showed up and were super earnest and they wanted to do something, and didn’t know the channels to do it. And I think sometimes, honestly, the events were a bit more like an affective space, just a place for people to come with those emotions.” (I_13). Many interviewees described the events as improvisational and chaotic, driven by frenzy and urgency, but also as deeply affective, characterized by a mix of solidarity, activism, mutual aid, and building towards a shared goal (I_77, I_49, I_08, I_80, I_13, I_43, I_33, I_63, I_56).

However, many of the interviewees also expressed how, after the initial start of the movement, a sense of frustration and exhaustion, followed by burnout, set in. Although many datasets were successfully mirrored and technically available, most of the interviewees questioned whether the resulting data collections were in fact meaningful, last but not least because during the first Trump administration, no data actually vanished. In fact, interviewees recalled a sense of disillusionment with the movement pretty early on and while it was still going on (I_79, I_13, I_08, I_80, I_77). One interviewee used the wording of “essentially taking the family values and putting them into jars and burying them in the backyard” (I_79) when critically reflecting on the Data Rescue activities. This metaphoric description points to the general challenge of how impactful downloading static snapshots of datasets, while many of the data were actually not static but regularly updated, which constituted part of their informational value.

Burnout was a recurring theme in interviewees’ retrospective accounts. For some, the end of their involvement was not only a relief, but also came with a deep sense of exhaustion and ambiguity about the movement's impact. One interviewee (I_80) resumed, “I basically didn’t talk about that 2017 experience from about 2018 (…) it ended pretty sour. It was pretty horrible for me.” Another described carrying on “past the moment where at least I believed in it” (I_79). This sense of emotional letdown was reinforced by a recognition that, despite the urgency of their efforts, the outcomes remained uncertain. As another interviewee reflected, “We got a lot of people to do all this archiving and the data sets didn’t disappear (…) we didn’t actually do anything useful with the data afterwards. There are a lot of reflections of like, well, how useful ultimately was that?” (I_72). One interviewee reflected that the solution to download as much data as possible in retrospect felt short-sighted.

On the side of the events themselves, I think there was sometimes the feeling of like, okay, if you catalogue all the sites of the EPA that you can easily get to now…. You’re done now. The EPA is protected now. We have all of the data and we’re going to share. (…) I think that can feel good in the moment. But sometimes, I worried that that made it feel like that was enough or that was sufficient. (…) I think it started slightly as this, like, “We can code our way out of this societal problem”. And I think that's just never true. And I wish we had addressed that earlier, honestly. (I_49)

The affective trajectory described by interviewees, moving from urgent excitement to eventual burnout, resonates strongly with existing literature on activist burnout (Brown and Pickerill, 2009; Gorski and Chen, 2015). Like other forms of digital activism, the Data Rescue movement depended significantly on emotionally driven volunteer labor, underpinned by a sense of moral urgency and immediacy. However, reliance on affective energy without institutional support structures frequently leads to fatigue and eventual disengagement (Gorski and Chen, 2015). These findings highlight a critical infrastructural vulnerability: the limited sustainability of affectively driven volunteer movements in the absence of systematic support and recognition.

Generalized distrust and particularized trust: conditional confidence in data stewardship

Interviewees held complex views on the value and limits of government data. They articulated a distinction between particularized trust as the confidence placed in specific individuals or groups, such as civil servants, and generalized trust, referring to broader confidence in government institutions or the political system. This differentiation shaped both the legitimacy of their rescue work and their emphasis on redundancy and monitoring.

Since the Data Rescue movement in 2016/2017 mainly focused on federal environmental data, a focus was on data that would be useful for research around environmental studies, especially with regard to environmental justice and climate change. Participants expressed a deep belief that federal data is a public good and important to civic life. This value was often framed in terms of civic entitlement and democratic accountability (I_01, I_63, I_49). But many also expressed long-standing skepticism toward the political neutrality and epistemic completeness of these datasets (I_43, I_70, I_08, I_79, I_01, I_45). Some interviewees experienced tension between their academic training and their new activist identities. One interviewee reflected on the irony of STS scholars, often critical of government and state knowledge, now mobilizing to save government data. “We’re wrecking our brains, because we’re doing all this work, that's critiquing the government and critiquing policy. And here I am, an anarchist, trying to save government data?!” (I_45).

These moments of epistemic dissonance were common, especially among STS researchers, who on the one hand critically reflected the “performative” (I_70, I_08) and often fatally de-contextualized (I_01, I_77) nature of public data to construct a “facade of objectivity” (I_45), and on the other hand got very involved in saving or “defending” (I_43, I_79) the data they criticized. However, the object of distrust was not the datasets themselves, but the political system charged with the maintenance of the infrastructures. Some interviewees talked about personal development from anti-statist skepticism to pragmatic confidence in bureaucratic routines: the civil service was seen as “a very powerful mechanism for certain kinds of reliability … very difficult to imagine a replacement for that sort of hierarchically organized, top-down approach to data collection” (I_79). Interviewees repeatedly drew a line between who produces data and who controls budgets. “I actually do trust the civil servants that maintain a lot of the public government information (…) but what I don’t trust is the resources available to some of these federal agencies” (I_75). Here, technical fragility is traced to underfunding and not to scientific malpractice, anchoring trust in professional norms while casting doubt on organizational capacity.

These reflections show that trust in federal data infrastructures is rarely absolute or straightforward. Many interviewees described robust particularized trust in the everyday stewardship of civil servants, but profound skepticism toward the system as a whole, especially under conditions of political volatility. Others drew a distinction between trusting data to be reliable and distrusting its capacity to capture what matters for environmental justice, arguing that “the data should be both trusted and distrusted” (I_79). This dynamic, shaped in part by prior national experiences of data loss or erasure, shows that distrust can serve as a catalyst for community mobilization and intervention when systemic vulnerabilities are perceived. In addition, the interviewees critically reflected on the infrastructural fragility inherent in digital data systems, emphasizing how “loss of understanding and the kind of commitment to the kind of permanence that we should expect from information” (I_80) is itself an object of both trust and distrust. In this way, interviewees moved beyond binary views of trust, instead foregrounding its relational, situated, and contingent character, especially with regard to trust in complex, abstract infrastructures.

Governance frictions and data-justice futures: negotiating stewardship beyond emergency curation

Non-hierarchical organizational structures inside the movement facilitated rapid mobilization but also exposed persistent tensions related to expertise, authority, and justice. These tensions prompted the movement to consider stewardship models that balance institutional durability with community accountability.

Interviewees remembered the first hackathons as “chaotic, but beautiful and galvanizing,” then immediately noted that “there weren’t a lot of people who were experts at what to do with all the data we were scraping (…) and there were disagreements about how that should be done” (I_63). Librarians worried about provenance and metadata; technologists focused on speed. The clash signaled that duplication alone could not guarantee meaningful public access. Leadership in the group was “not about having more authority, but about devoting more time,” as one member put it (I_45). Although this ethos cultivated solidarity, it also blurred decision rights, especially around the attribution of invisible labor and the adjudication of archival standards.

Interviewees critically looked back at ideas resolving around tools, distributed web solutions and blockchain, which in 2016/2017 were seen as technology-centered umbrella solutions for various problems. Community-based archiving was in hindsight described by some as inclusive, but also as unorganized, panic-driven and generally idealistic, taking on “herculean, poorly framed tasks” (I_13) in a “sort of undifferentiated panic (…) where it was very difficult to distinguish immediate, true, but genuine urgency from, you know, paranoia” (I_79). I_56 described these tensions as a “clash of values” and the distinction extended to epistemological values: some interviewees favored community-accountability and distributed data collections, while other participants, often with a background in library or information sciences, criticized overly technocratic approaches without taking into account archival practices, such as authenticity, integrity, and adequate metadata creation for the downloaded data.

One interviewee (I_70) vividly remembered an episode, where several EDGI members met up in person for a workshop to discuss what a more accountable version of public data and public data infrastructures might look like. The interviewee described this experience as “striking” because it revealed a shared struggle to articulate a positive, forward-looking vision for public data infrastructure. While it was easy for the group to critique existing systems and name what they opposed, they found it deeply challenging to imagine concrete alternatives. The moment stood out because it highlighted how difficult it was to move from critique to construction. Another interviewee (I_80) recalled a meeting with representatives from large institutions in order to form a plan in which these institutions would coordinate to keep copies of federal information. This initiative similarly failed, leaving the interviewee utterly frustrated toward the unwillingness or inability of established infrastructures to take responsibility. These reflections underscore a justice-oriented perspective, emphasizing that safeguarding data is necessary but insufficient in the absence of more accountable, inclusive, and resilient infrastructural systems.

Reflecting on the following phase of EDGI's work, some interviewees insisted that future stewardship must engage the people most affected by environmental harm: “What communities need to be engaged? (…) How can we use that same process to restore some hope that knowledge can guide policy and improve people's lives?” (I_79). In this way, data preservation becomes inseparable from questions of equity and civic empowerment.

The governance frictions highlighted by interviewees, particularly regarding metadata quality, archival standards, and decision-making authority, indicate broader tensions inherent in volunteer-based activist infrastructures. These tensions align with critiques from archival scholarship on the complexities of “participatory archiving” and “community-based stewardship,” where decentralized democratic ideals often collide with the practical requirements of long-term preservation and interpretability (Caswell, 2014; Huvila, 2008). It also resonates with work on citizen sensing and civic technoscience, where volunteer infrastructures both extend and strain expert practice (Murillo, 2016; Wylie et al., 2014). Interviewees’ frustrations with institutional reluctance to engage reflect a broader challenge: bridging grassroots activism with institutionalized archival practices requires clearer policies and structured dialogue between civic initiatives and formal stewardship institutions (Caswell, 2014).

Discussion and conclusion

The empirical findings presented above offer insight into how anticipatory maintenance and emergency curation unfolded in practice during the 2016/2017 Data Rescue movement. The central themes of infrastructure fragility, systemic distrust, affective mobilization, and governance tensions highlight critical dynamics shaping grassroots data preservation efforts. These findings underscore theoretical insights from infrastructure studies, particularly around mimeographic labor and anticipatory maintenance.

From emergency curation to anticipatory maintenance

Retrospective accounts of Data Rescue participants reflect urgency, emotion, and critique. In the immediate aftermath of the 2016 U.S. presidential election, the Data Rescue movement was perceived as an urgently needed intervention and a way to materially respond to the risk of epistemic erasure under an administration widely seen as hostile to science and transparency. Interviewees described this period as marked by affective intensity and a strong sense of moral and/or professional purpose. However, with the benefit of hindsight, interviewees also articulated a critical distance from their previous actions. The tactical focus on data downloading and mirroring, while symbolically significant, was often seen as having limited practical impact over the long term.

Interviewees voiced skepticism about the long-term efficacy and epistemological soundness of emergency curation. Some questioned whether the focus on rapid data download and redundancy, without systematic attention to context, metadata, or long-term stewardship, risked rendering data less meaningful or usable over time. This echoes concerns in archival and infrastructure studies that decontextualized data can lose much of their evidentiary and interpretive value, especially when provenance and relational metadata are lacking (Borgman, 2017; Bowker et al., 2010). The interpretive value of data depends not just on availability, but on the governance processes that document their provenance and recontextualize them across sites of reuse, without which data risk losing much of their evidentiary power (Leonelli, 2019). Others critiqued the movement's limited institutional alliances, noting that an adversarial stance toward government agencies and formal repositories sometimes hampered collaboration and integration with existing archival infrastructures. Some interviewees also observed that the prevailing media narrative of “rescue” tended to oversimplify the persistent, gradual forms of data loss endemic to digital infrastructures, potentially diverting attention from the need for sustained and less visible forms of care and maintenance.

This duality between the initial affective intensity and subsequent critical reassessment highlights a broader theme of reflexive activism. Similar to what Caswell (2021) describes in archival activism, interviewees negotiated the tension between action as moral imperative and action as strategic uncertainty. These reflections align with the literature on crisis informatics, which demonstrates that moments of rupture can prompt ad hoc innovation, but may also lack sustainability or coherence (Kaufhold, 2024; Soden and Palen, 2018). For Data Rescue, the question became not only what was preserved, but how preservation itself was conceptualized, whether as a short-term safeguard or as part of a broader vision of infrastructural transformation.

The Data Rescue movement exemplifies what Jackson (2014) calls broken-world thinking, an orientation that treats infrastructures as perpetually vulnerable and in need of continuous care. In particular, this orientation was extended in practice to encompass proactive future-focused strategies. Volunteers worked to maintain data before a breakdown occurred, rehearsing the loss so that it would remain hypothetical. Their precautionary workflows and the “backyard burial” metaphor voiced by one interviewee make explicit what Mattern (2018) describes as maintenance as speculative care: the conviction that the present action preserves the possibility of certain futures.

Extending this argument requires foregrounding the temporal dynamics that make anticipatory maintenance distinct. Anticipation, as Adams et al. (2009) and Clarke et al. (2015) emphasize, organizes present conduct through imagined temporal horizons instead of just projecting futures. Within Data Rescue, the affective charge of distrust transformed an indeterminate future risk into a present actionable one: volunteers behaved as if the future loss had already begun. In this sense, care was not simply reactive but pre-emptive, folding speculative futures into the routines of downloading, monitoring, and verification (cf. Pink et al. (2018) on anticipatory modes and “just-enough” trust amid everyday data anxieties). This anticipatory pull of the future on the present (Adams et al., 2009) situates maintenance as an act of keeping time as much as keeping data: it synchronizes political anxiety, infrastructural temporality, and moral responsibility. Similarly, trust becomes temporal as a wager that current care will sustain a future of continuity even when institutional reliability feels suspended. Anticipatory maintenance therefore operates in a chronopolitics of care, where the future is constantly rehearsed, enacted, and sustained in the present.

In this chronopolitical sense, treating emergency curation as anticipatory maintenance clarifies two questions that run throughout participants’ reflections: what forms of care become possible under systemic distrust, and what forms of stability they can meaningfully achieve. Data Rescue was not just about protecting data, but about practicing how to respond together if something went wrong. Second, this view highlights how acting early changed the meaning of trust and stewardship. By taking action before a crisis, volunteers shifted trust away from formal institutions and toward community groups and networks. This suggests that the legitimacy of data preservation does not depend only on official authorities, but can also be built through collective, civic efforts. In this way, anticipatory maintenance is not only technical work, but also a way for people to show who they trust to take care of public data in times of uncertainty.

Both trust and anticipation operate as responses to uncertainty. As Luhmann (1979) and Giddens (1990) argue, trust enables action by reducing complexity and separating uncertainties to achieve a tolerable condition for decision-making. Similarly, in anticipation, uncertainty is neither resolved nor denied; it becomes actionable through simplification and abduction (Clarke et al., 2015). In this sense, both trust and anticipation are techniques of managing the unknowable, translating indeterminacy into plausible futures and practicable action. Hope, as Clarke et al. (2015) add, becomes the affective corollary of this epistemic work: it sustains engagement despite the absence of guarantees. Within the Data Rescue movement, systemic distrust functioned as a mode of specified uncertainty (Sumpf, 2019), in which a possible failure was named and addressed preemptively. Anticipatory maintenance, in this sense, brings together these two logics of managing uncertainty. Like trust, it allows actors to act despite not knowing; like anticipation, it translates uncertainty into concrete routines of care.

A very similar pattern can be observed starting in late 2024, after the second election of Trump as president and concerning a second wave of data availability loss and Data Rescue initiatives (Kellam, 2025; Lucas, 2025).² However, three contrasts stand out. First, while the first wave of Data Rescue in 2016/2017 focused mainly on risk in the storage layer of federal data, fragility in the collection layer is now much more prominent (Calma, 2025; Chandler, 2025; Holthaus, 2025). This refines Rubio's thesis of relational fragility by showing that the locus of vulnerability migrates with political strategy. Second, redundancy infrastructures built in 2016/2017 provide a platform for rapid response, indicating the theoretical assumption that maintenance can be prophylactic. However, they also exemplify Jackson's broken-world thinking: every fix reveals new challenges, and a focus on availability will not ensure accessibility. Third, systemic distrust now extends beyond political appointees to include fears of wholesale abdication of data-production duties and corporate enclosures of public science.

EDGI's 2025 strategy deepens anticipatory maintenance by shifting from “copy-everything” redundancy to situated relational stewardship: rather than mass-mirroring federal repositories, a coalition called Public Environmental Data Partners³ now triages vulnerable, irreplaceable datasets and archives them together with the social perspectives, metadata, and interpretatory tools that give the data meaning. This move both relocates fragility from access loss to contextual interpretability of data, and affirms that maintenance must enable interpretability as well as availability, preserving the relationships among data, infrastructures, and affected communities (Nost et al., 2025).

The reflections of the interviewees consistently demonstrated a heightened awareness of infrastructural fragility. Rather than presuming the durability of federal data infrastructures, interviewees identified them as vulnerable to political volatility, administrative neglect, and technical decay. The ease with which data could be rendered inaccessible, whether through budget cuts, interface changes, or the quiet deprecation of tools, underscored the precariousness of what by many had previously been considered stable public resources. This recognition led many interviewees to reassess the adequacy of emergency preservation strategies. Although the focus on data availability was urgent in 2016, it was increasingly viewed as insufficient in the absence of parallel investments in accessibility, metadata, interpretability, and long-term curation. Participants recognized that technical duplication alone does not ensure public utility, a point echoed in digital preservation scholarship, where usability, context, and continuity are central to meaningful access (Leonelli, 2019).

Fragility is not an intrinsic property but “a relational and contingent condition” that is rendered within particular configurations of power and care (Rubio, 2025). In the context of U.S. open-data systems, budget cuts, politicized leadership changes, and obsolete web stacks turn data collected for climate models and pollutant registries fragile. Maintenance interventions are therefore not just reactive fixes, but they represent moments in which actors actively recalibrate which forms of fragility, and whose interests, are prioritized. This relational lens explains why the Data Rescue coalition invested in redundancy and distributed hosting: if fragility is produced by central-state volatility, resilience must be produced through decentralized stewardship. However, interviewees identified a structural gap between their efforts and institutional capacity. Despite public visibility and media attention, the Data Rescue movement lacked systemic backup from durable, well-funded infrastructures. This lack of institutional follow-through amplified a sense of impermanence and underscored the broader challenge of building resilient information infrastructures in the absence of state commitment. As Star (1999) suggests, infrastructures are most visible when they break down. For many in the Data Rescue movement, this breakdown was both a technical concern and a political reality.

From generalized distrust to particularized trust

Trust and distrust played a central role in shaping both the epistemic and political logic of the Data Rescue movement. Echoing Luhmann’s (1979) proposal that trust and distrust are both forms of risk management in modern society, interviewees described how their involvement was often triggered by a breakdown of systemic trust in federal stewardship of environmental and scientific data. Importantly, this distrust did not foster disengagement; rather, it catalyzed new forms of collaborative action and alternative infrastructure-building. Thus, distrust was not so much an obstructive force as it was a productive one.

Interviewees repeatedly distinguished between trust placed in particular individuals or groups, such as career civil servants responsible for data stewardship (particularized trust), and trust in the broader government infrastructure or political leadership (generalized trust). Freitag and Traunmüller (2009) argue, that particularized trust can remain robust even as generalized trust in institutions declines, a pattern echoed in interviewees’ accounts of the Data Rescue movement. This shift from generalized trust in government institutions to particularized trust in specific individuals or networks more than a defensive response to systemic uncertainty. The breakdown of generalized trust can spur actors to experiment with alternative forms of organization, mobilizing new resources and leading to creative alliances outside established structures. In the case of Data Rescue, systemic distrust generated action by making infrastructural fragility thinkable and by mobilizing communities to act preemptively.

The findings also indicate that trust is influenced by both temporal and institutional factors. Data Rescue volunteers anticipated potential future failures, including the possibility of management decline, while suspending critiques of government data that many of them had held previously. This dynamic created a particular kind of vulnerability that did not focus on data corruption but on the predicted future disruption of knowledge continuity. By framing deletion, metadata decay, and data collection gaps as imminent rather than speculative, participants transformed theoretical risks into concrete maintenance strategies.

The entanglement of trust and activism within Data Rescue reflects a broader shift in how infrastructures are imagined in the wake of political rupture. Participants in Data Rescue acted out of a breakdown in system trust, particularly out of distrust in the Trump administration's intentions toward environmental data, informed in part by similar actions under Canada's Harper government (Nost et al., 2021). Rather than viewing data stewardship as solely an institutional obligation, participants assumed the role of temporary stewards, reframing trust as an emergent property of community-driven infrastructures. Interviewees’ reflections also challenge the assumption that trust in data is primarily rooted in technical form or institutional origin. Instead, trust was conceptualized as relational, contingent, and frequently grounded in social accountability and governance. Infrastructures were deemed trustworthy not just because of their design, but because of the people and ethical commitments behind them. This echoes a notion of trust as a process of suspension, in which actors engage with systems not because they are certain, but because they are willing to act as if trust is justified, often in spite of, rather than in the absence of, systemic uncertainty (Möllering, 2006, 2013).

Conclusion

Empirically, the experiences of Data Rescue volunteers illustrate Rubio’s (2025) concept of fragility as a relational and contingent state rather than an inherent attribute of infrastructures. Rather than just reacting to technical failures, volunteers proactively anticipated infrastructural disruptions by establishing redundancy and decentralized stewardship. This aligns with Jackson's (2014) broken-world thinking, where infrastructures are inherently precarious, continuously requiring preventive care. However, the findings also expand on this literature by highlighting how anticipatory practices are deeply embedded in affective labor, which simultaneously fuels and limits volunteer participation. The observed burnout trajectory emphasizes the importance of integrating structured emotional support and sustainable mobilization practices into grassroots infrastructural activism (Caswell, 2014; Gorski and Chen, 2015). In addition, systemic distrust acts as a generative force that reshaped data stewardship. By drawing distinctions between politically appointed custodians and career civil servants, volunteers recalibrated trust relationships, strategically placing confidence in redundant and distributed infrastructures rather than in centralized governmental authority. This reconfiguration of trust underscores that data infrastructures are not solely technical entities but fundamentally social and political constructs (Star, 1999).

It is also important to recognize the central ambivalent tactic of “copying” by the Data Rescue movement (cf. Leonelli (2019) on governance and interpretation). Although mass duplication provided a sense of agency and material insurance against possible data loss, it sometimes favored volume over value, sidelining deeper questions about context, provenance, and long-term usability. Interviewees’ retrospective doubts about the meaningfulness and impact of these duplicated collections reflect a broader tension in digital preservation: emergency copying can offer temporary reassurance, but may inadvertently contribute to contextual loss, potentially overwhelming future data curators and users with decontextualized or unmanageable datasets. Data Rescue thus illustrates both the immediate necessity of preservation during perceived crises and the ongoing challenge of defining responsible curation practices under conditions of uncertainty.

The experiences and reflections of Data Rescue participants analyzed here show connections between infrastructure fragility, anticipatory maintenance, and systemic distrust. By conceptualizing emergency curation as anticipatory maintenance, this study extends Jackson’s (2014) broken-world thinking to proactive, speculative, and anticipatory maintenance. In addition, it enriches Rubio’s (2020, 2025) concept of mimeographic labor by emphasizing how repetitive maintenance practices serve not only to repair existing infrastructures, but also to prevent unforeseen failures. Crucially, by foregrounding systemic distrust as a generative rather than debilitating force, the analysis reveals how distrust catalyzed grassroots innovation, fostering redundant and resilient data stewardship infrastructures.

Anticipatory maintenance exposes how uncertainty is not only a condition to be mitigated but a medium through which care unfolds. Both trust and anticipation are temporal responses to the unknown: as Luhmann (1979) and Giddens (1990) argue, trust allows action by bracketing uncertainty, while anticipation, as described by Clarke et al. (2015), renders uncertainty actionable through simplification, abduction, and hope. In Data Rescue, these orientations converged. Participants acted as if the feared future had already arrived, translating systemic distrust into preventive action. Through such practices, uncertainty became operational rather than paralyzing: it was transformed into a temporal infrastructure of care. Anticipatory maintenance thus names not only a form of preventive work but a temporal mode of stewardship, one that holds space for indeterminacy while enacting continuity under conditions of fragility.

Overall, these findings suggest that the Data Rescue movement represents more than a response to imminent data loss; it is also a site for testing and reimagining approaches to digital stewardship. The retrospective accounts of the interviewees highlight the importance of recognizing infrastructures as ongoing social and political processes, shaped by uncertainty, collaboration, and care. By approaching data preservation as anticipatory maintenance and acknowledging the productive role of systemic distrust, this study contributes to a more nuanced understanding of how communities respond to infrastructural fragility in times of political change. Rather than treating data as static assets, the experiences of Data Rescue volunteers emphasize the value of building resilient, adaptable infrastructures that can accommodate both technical and social forms of uncertainty. In this light, infrastructural resilience appears less as technical stability than as an ongoing temporal politics of care in which systemic distrust is converted into anticipatory maintenance: acting as if the future had already arrived to keep continuity possible.

Footnotes

Acknowledgements

This paper is part of my PhD research. Most of all, I would like to thank all interview participants for sharing their time, experiences, and reflections for this study. I am grateful for their openness and generosity, and for the careful way they discussed both the opportunities and the limits of Data Rescue work. Their contributions were essential to this research.

ORCID iD

Laura Rothfritz

Ethical considerations and informed consent

This study is based on qualitative interviews conducted as part of doctoral research at the Humboldt-Universität zu Berlin. According to the ethics board at the humanities department of the Humboldt-Universität zu Berlin, formal approval was not required for this type of research. All interviewees provided written informed consent prior to participation in the study. Participants were informed about the study's aims, the voluntary nature of their participation, data storage, and their rights regarding confidentiality and withdrawal according to GDPR regulations. Interviews were anonymized and treated confidentially to protect the privacy of participants. All data are stored privately with no access by third parties.

Funding

The author received no financial support for the research and authorship of this article. The article proccessing charge was funded by the Open Access Publication Fund of Humboldt-Universität zu Berlin.

Declaration of conflicting interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

The disaggregated data underlying this study are not available to protect the privacy and confidentiality of the participants interviewed.

Notes

Author biography

Laura Rothfritz is a research associate and doctoral candidate at the School for Library and Information Science at Humboldt-Universität zu Berlin, Germany. Her research focuses on trust and distrust in public data infrastructures, with a particular emphasis on environmental and climate data, data activism, and socio-technical maintenance practices.

References

Adams

Murphy

Clarke

(2009) Anticipation: Technoscience, life, affect, temporality. Subjectivity 28(1): 246–265.

Allen

Eke

Foster

, et al. (2017a) Building data refuge: From bucket brigade to sustainable action. CNI Spring 2017 Meeting. Available at: https://www.cni.org/topics/digital-curation/building-data-refuge-from-bucket-brigade-to-sustainable-action (accessed 9 February 2026).

Allen

Stewart

Wright

(2017b) Strategic open data preservation: Roles and opportunities for broader engagement by librarians and the public. College & Research Libraries News [Online] 78: 9.

Berger

(2015) Now I see it, now I don’t: Researcher’s position and reflexivity in qualitative research. Qualitative Research 15(2): 219–234.

Borgman

(2017) Big Data, Little Data, No Data: Scholarship in the Networked World. Cambridge, MA: MIT Press.

Bosworth

(2023) What is ‘affective infrastructure’? Dialogues in Human Geography 13(1): 54–72.

Bowker

Baker

Millerand

, et al. (2010) Toward information infrastructure studies: Ways of knowing in a networked environment. In: Hunsinger

Klastrup

Allen

(eds) International Handbook of Internet Research. Dordrecht: Springer Netherlands, 97–117.

Braun

Clarke

(2006) Using thematic analysis in psychology. Qualitative Research in Psychology 3(2): 77–101.

Brown

Pickerill

(2009) Space for emotion in the spaces of activism. Emotion, Space and Society. Activism and Emotional Sustainability 2(1): 24–35.

10.

Calma

(2025) Donald Trump’s data purge has begun. Available at: https://www.theverge.com/news/604484/donald-trumps-data-purge-has-begun (accessed 3 February 2025).

11.

Caswell

(2014) Community-centered collecting: Finding out what communities want from community archives. Proceedings of the American Society for Information Science and Technology 51(1): 1–9.

12.

Caswell

(2021) Urgent Archives: Enacting Liberatory Memory Work. London: Routledge.

13.

Caswell

Cifor

(2016) From human rights to feminist ethics: Radical empathy in the archives. Archivaria 81: 23–43.

14.

Chandler

(2025) The data is falling. Bulletin—Association of Canadian Map Libraries and Archives (ACMLA) 175: 22–24.

15.

Cifor

Caswell

Migoni

, et al. (2018) “What we do crosses over to activism”: The politics and practice of community archives. The Public Historian 40(2): 69–95.

16.

Clarke

, et al. (2015) Anticipation work: Abduction, simplification, hope. In: Bowker

Timmermans

Clarke

(eds) Boundary Objects and Beyond: Working with Leigh Star. Cambridge, MA: MIT Press, 85–119.

17.

Currie

Donovan

Paris

(2018) Preserving for a more just future: Tactics of activist data archiving. In: Munshi

Verma

(eds) Data Science Landscape: Towards Research Standards and Protocols. Studies in Big Data. Berlin: Springer Nature, 67–78.

18.

Currie

Donovan

(2019) What difference do data make? Data management and social change. Online Information Review 43(6): 971–985.

19.

Currie

Paris

(2018) Back-ups for the future: Archival practices for data activism. Archives and Manuscripts 46(2): 124–142.

20.

Dillon

Lave

Mansfield

, et al. (2019) Situating data in a Trumpian era: The environmental data and governance initiative. Annals of the American Association of Geographers 109(2): 545–555.

21.

Dillon

Walker

Shapiro

, et al. (2017) Environmental data justice and the Trump administration: Reflections from the environmental data and governance initiative. Environmental Justice 10(6): 186–192.

22.

Edwards

Jackson

Bowker

, et al. (2007) Understanding Infrastructure: Dynamics, Tensions, and Design. Working Paper, January. Available at: https://hdl.handle.net/2027.42/49353 (accessed 9 February 2026).

23.

Freitag

Traunmüller

(2009) Spheres of trust: An empirical analysis of the foundations of particularised and generalised trust. European Journal of Political Research 48(6): 782–803.

24.

Giddens

(1990) The Consequences of Modernity. Cambridge: Polity Press.

25.

Gilliland

Caswell

(2016) Records and their imaginaries: Imagining the impossible, making possible the imagined. Archival Science 16(1): 53–75.

26.

Gorski

Chen

(2015) “Frayed all over:” The causes and consequences of activist burnout among social justice education activists. Educational Studies 51(5): 385–405.

27.

Harb

(2025) Repair work, resistance and the invisible data labour of Lebanon’s digital humanitarian infrastructures. Big Data & Society 12(1). DOI: 10.1177/20539517251318268.

28.

Higgins

(2008) The DCC curation lifecycle model. International Journal of Digital Curation 3(1): 134–140.

29.

Holthaus

(2025) Major US climate website likely to be shut down after almost all staff fired. The Guardian. Available at https://www.theguardian.com/us-news/2025/jun/11/climate-website-shut-down-noaa (accessed 9 February 2026).

30.

Huvila

(2008) Participatory archive: Towards decentralised curation, radical user orientation, and broader contextualisation of records management. Archival Science 8(1): 15–36.

31.

Iliadis

Russo

(2016) Critical data studies: An introduction. Big Data & Society 3(2). DOI: 10.1177/2053951716674238.

32.

ISO 14721:2025 (2025) Space Data System Practices—Reference model for an Open archival Information System (OAIS).

33.

Jackson

(2014) Rethinking repair. In: Gillespie

Boczkowski

Foot

(eds) Media Technologies: Essays on Communication, Materiality, and Society. Cambridge, MA: MIT Press, 221–239.

34.

Janssen

Charalabidis

Zuiderwijk

(2012) Benefits, adoption barriers and myths of open data and open government. Information Systems Management 29(4): 258–268.

35.

Janz

(2017) Maintaining access to public data: Lessons from data refuge. Against the Grain 29(6): 30–33.

36.

Janz

(2019) The Data Refuge Project for protecting federal data in the United States. In: Duclos M and Fjeld A (eds) Liberte de La Recherche: Conflits, Pratiques, Horizons, p. 7. Available at: https://repository.upenn.edu/handle/20.500.14332/38406 (accessed 9 February 2026).

37.

Jasanoff

(2017) Virtual, visible, and actionable: Data assemblages and the sightlines of justice. Big Data & Society 4(2): 2053951717724477.

38.

Johnson

(2014) From open data to information justice. Ethics and Information Technology 16(4): 263–274.

39.

Karasti

Baker

Millerand

(2010) Infrastructure time: Long-term matters in collaborative development. Computer Supported Cooperative Work (CSCW) 19(3): 377–415.

40.

Kaufhold

M-A

(2024) Exploring the evolving landscape of human-centred crisis informatics: Current challenges and future trends. i-com 23(2): 155–163.

41.

Kazansky

(2021) ‘It depends on your threat model’: The anticipatory dimensions of resistance to data-driven surveillance. Big Data & Society 8(1). DOI: 10.1177/2053951720985557.

42.

Kellam

(2025) The Data Rescue Project and wider initiatives. IASSIST 2025, Bristol, UK. DOI: 10.5281/zenodo.15598141.

43.

Kingston

(2015) Vanishing Canada: Why we’re all losers in Ottawa’s war on data. MacLeans. Available at https://www.macleans.ca/news/canada/vanishing-canada-why-were-all-losers-in-ottawas-war-on-data/ (accessed 9 February 2026).

44.

Koebler

(2016) Researchers are preparing for Trump to delete government science from the web. VICE. Available at https://www.vice.com/en/article/researchers-are-preparing-for-trump-to-delete-government-science-from-the-web/ (accessed 9 February 2026).

45.

Lamdan

(2017) Lessons from datarescue: The limitations of grassroots climate change data preservation and the need for federal records law reform. University of Pennsylvania Law Review Online 166: 231–248.

46.

Leonelli

(2019) Data governance is key to interpretation: Reconceptualizing data in data science. Harvard Data Science Review 1(1). DOI: 10.1162/99608f92.17405bb6.

47.

Lucas

(2025) The volunteer data hoarders resisting Trump’s purge. The New Yorker. Available at https://www.newyorker.com/news/the-lede/the-data-hoarders-resisting-trumps-purge (accessed 9 February 2026).

48.

Luhmann

(1979) Trust and Power: Two Works by Niklas Luhmann. Chichester: Wiley.

49.

Mattern

(2018) Maintenance and care. Places Journal. DOI: 10.22269/181120.

50.

Merriam

(1988) Case Study Research in Education. A Qualitative Approach. San Francisco: Jossey-Bass.

51.

Möllering

(2006) Das Aufheben von Ungewissheit als Kern des Vertrauens: Just do it? MPIfG Working Paper. Köln: Max-Planck-Institut für Gesellschaftsforschung.

52.

Möllering

(2013) Trust without knowledge? Comment on Hardin, ‘government without trust’. Journal of Trust Research 3(1): 53–58.

53.

Murillo

LFR

(2016) New expert eyes over Fukushima: Open source responses to the 3/11 disaster in Japan. Anthropological Quarterly 89(2): 399–429.

54.

Murphy

(2015) Unsettling care: Troubling transnational itineraries of care in feminist health practices. Social Studies of Science 45(5): 717–737.

55.

Nost

Gehrke

Poudrier

, et al. (2021) Visualizing changes to US federal environmental agency websites, 2016–2020. PLOS ONE 16(2): e0246450.

56.

Nost

Gehrke

Vera

, et al. (2025) Why the environmental data & governance initiative is archiving public environmental data. Patterns 6(1): 101151.

57.

Paris

(2017) Data-mirroring as archival activism. In: Data Power 2017: An international conference on (big) data & power, Carlton, CA, 22–23 June 2017. DOI: 10.22215/datapower.v2017i0.22.

58.

Paris

Currie

Donovan

(2017) Grand Intentions, Small Interventions: Climate Data Rescue as Counter-Data Action. AoIR Selected Papers of Internet Research. Available at https://spir.aoir.org/ojs/index.php/spir/article/view/10187 (accessed 9 February 2026).

59.

Piattoeva

Saari

(2022) Rubbing against data infrastructure(s): Methodological explorations on working with(in) the impossibility of exteriority. Journal of Education Policy 37(2): 165–185.

60.

Pink

Lanzeni

Horst

(2018) Data anxieties: Finding trust in everyday digital mess. Big Data & Society 5(1). DOI: 10.1177/2053951718756685.

61.

Ray

(2018) A motivating climate: An exploratory case study of the DataRescue movement. Master’s Thesis, University of North Carolina, US. DOI: 10.17615/s089-rb86.

62.

Rubio

(2020) Still Life: Ecologies of the Modern Imagination at the Art Museum. Chicago, IL: University of Chicago Press.

63.

Rubio

(2025) Introduction: Avowing fragility. In: Rubio

Denis

Pontille

(eds) Fragilities: Essays on the Politics, Ethics, and Aesthetics of Maintenance and Repair. Cambridge, MA: MIT Press, 1–31.

64.

Ruhleder

(1996) Steps toward an ecology of infrastructure: Design and access for large information spaces. Information Systems Research 7(1): 111–134.

65.

Soden

Palen

(2018) Informating crisis: Expanding critical perspectives in crisis informatics. Proceedings of the ACM on Human-Computer Interaction 2(CSCW): 1–22. DOI: 10.1145/3274431.

66.

Stake

(2010) The Art of Case Study Research. Thousand Oaks, CA: Sage Publications.

67.

Star SL (1999) The ethnography of infrastructure. American Behavioral Scientist 43(3): 377–391.

68.

Sumpf

(2019) System Trust: Researching the Architecture of Trust in Systems. Wiesbaden: Springer Fachmedien.

69.

Vera

Walker

Murphy

, et al. (2019) When data justice and environmental justice meet: Formulating a response to extractive logic through environmental data justice. Information, Communication and Society 22(7): 1012–1028.

70.

Vukliš

Gilliland

(2016) Archival activism: Emerging forms, local applications. In: Filej

(ed) Archives in the Service of People—People in the Service of Archives. Maribor, Slovenia: Alma Mater Europea, 14–25.

71.

Walker

Nost

Lemelin

, et al. (2018) Practicing environmental data justice: From DataRescue to Data Together. Geo: Geography and Environment 5(2): e00061.

72.

Wylie

Jalbert

Dosemagen

, et al. (2014) Institutions for civic technoscience: How critical making is transforming environmental research. The Information Society 30(2): 116–126.

73.

Zembylas

(2023) Theorizing ‘affective infrastructure’ in education policy: Articulating new political imaginaries for a more equitable future. Journal of Education Policy 38(6): 944–962.