Between January 2025 and the time of writing (July 2026), hundreds of organizations, from grassroots volunteer networks to major research institutions, have mobilized to rescue, preserve, and provide access to at-risk federal data in a rapid response to a sustained and accelerating assault on these knowledge assets. While this wave of work began with many ad hoc and siloed efforts, it has developed into a movement, and invested stakeholders have stepped forward to take on “hub” roles, helping to coordinate and guide the rescue efforts. A sense of movement generosity has also been building among the players, perhaps most evident in hard-hit fields such as environmental and climate science. Those with resources and know-how are sharing wherever they can, particularly via their time, energy, and networks.
Funders have responded to the crisis, with the Robert Wood Johnson Foundation, Hewlett Foundation, David and Lucile Packard Foundation, John D. and Catherine T. MacArthur Foundation, Alfred P. Sloan Foundation, Mellon Foundation, and others directing significant resources toward data rescue and resilience efforts. Some funders are working together in formal coordination groups (e.g., Funders for the Future of Public Data (3FPD) and the Portfolio to Protect Science) in hopes of moving resources in ways that reduce duplication, “noise,” while increasing investment alignment.
But a close look at where that money is flowing reveals concerning patterns that could subvert or undercut this work, limiting its ability to outlast the crisis that prompted it. Concretely, this includes dangerously low investments made in open infrastructures – including the technical platforms, standards, and software required to produce, share, discover, access, and preserve knowledge.
Crisis reveals what was always true
Funding is moving in familiar, system-driven ways that we have witnessed many times before. As my colleagues and I have noted elsewhere (e.g., Red Queen’s Race, Emperor’s New Clothes), stakeholders in the knowledge and information ecosystem have for years been rewarded for ways of working that provide short-term gains but that tend to fail to add up to long-term success.
Incentive structures, including funding and career advancement, encourage stakeholders to pursue innovation over maintenance, to build competing siloes rather than grow interoperability and coordination, and to misinterpret bare survival of open infrastructure communities and services as though that survival somehow signifies success.
The resulting fragility of open infrastructure components has been persistent, and the current political crisis of 2025-2026 hasn’t created precarity so much as it has exposed it. Data is disappearing because the infrastructure in which it is nested was already brittle. This moment is an opportunity, not just to rescue data, but to correct a long-standing dependency on temporary and inadequate funding for infrastructure-level work.
Tracking investment flows
IOI has been mapping the emerging data resilience ecosystem, tracking more than 85 “data rescue” and “data resilience” projects, initiatives, forums, and funding efforts that have launched or significantly evolved since 2025. The landscape clusters into recognizable functional layers: active rescue and collection efforts; tools, repositories, and discovery systems; monitoring and tracking; coordination and strategy; and advocacy and fundraising. Across these clusters, coordination-style work — convenings, roadmaps, governance documentation, advocacy campaigns — dominates the funded project list. Rescue and collection efforts, including some remarkable grassroots mobilizations, are also well represented in the list.
What is almost entirely absent in the projects and initiatives we are tracking is investment in the technical infrastructure layer: the tools, pipelines, standards, systems, and people that make any of the other work durable. Of the 87 projects and initiatives we are tracking, 34 involve coordination activities, 28 involve advocacy, and 33 involve data collection, while only 16 are actively building and maintaining technical tools, and more than half of those are building for their own project's needs rather than building shared, reusable infrastructure.
Those who are conducting “data rescue” or aiming for “data resilience,” depend on tools, pipelines, and standards to do their work, but those elements often are not being resourced through these funding streams. Rescued data has to be collected, and then it has to go somewhere; it has to be validated, described, made findable, and kept accessible. The systems that do that work, from web archiving tools to identifier infrastructure, from metadata standards to format normalization, and from discovery layers to storage and preservation activities, are either assumed to already exist and be adequately supported, or they are simply not visible in the investment picture.
The danger is hidden: much of the content that is being rescued is flowing onto known and trusted infrastructures, e.g., Internet Archive, ICPSR and Data LUMOS, and Texas Advanced Computing Center (TACC). The longevity of these environments is unknown and they are not being funded to take on a permanent or future-facing role for this content. What does that mean for the maintenance and reliability of these resources over time?
The infrastructure layer is chronically underfunded relative to the work it enables, with costs hidden through volunteerism and institutional subsidy. The data resilience funding landscape is replicating that failure under crisis conditions.
Learning from previous experiences
This is, of course, far from the first time activist communities have mobilized to rescue endangered digital collections at scale. For example, the 2016 data rescue movement, responding to the first round of federal data threats, activated DataRefuge, Preserving Electronic Government Information (PEGI), and the Environmental Data and Governance Initiative (EDGI), among other efforts. Libraries, cultural heritage professionals, and researchers have also mobilized around a myriad of events over time, from the Estonian cyberattacks of 2007 to the conflict zones in Sudan, Syria, Ukraine, and Afghanistan; and from responses to the British Library ransomware attack of 2023 to rescue efforts in the wake of wildfires, floods, and other major natural disasters. Each of these efforts generated hard-won knowledge about what works — distributed custody models, community triage processes, the limits of dark archiving sans deliberate attention to access, and the differences between rescue and preservation.
That knowledge is not being systematically drawn on in many of today’s projects, and the current US focus is producing efforts that are, in some cases, re-learning lessons that have already been learned elsewhere. We need to actively encourage and fund work that connects today's actors to this history.
For example, do we know what parts of the 2016-17 data rescue tooling and environment still exist, and what parts do not? Do we know what characteristics differentiate still-active groups like EDGI, ICPSR, and Internet Archive from other groups and toolsets that have sunset? Knowing more about the past can help us to invest more wisely and sunset more adeptly, establishing stronger scaffolding for the field. That requires directing some portion of the funding towards practical analysis of these past efforts.
Duplication of backbone (human) infrastructure
Our analysis shows a third gap that is perhaps the most costly today: instead of reinforcing the successful networks that already exist and distilling models from those that we can activate in other fields and disciplines, much of the funding available today is flowing to external contractors and consultants to build landscape analyses in order to better understand where the connectors and touchpoints between current efforts are.
Unintentionally, these external groups are both duplicating and undermining long-standing, often mature or maturing networks on the ground. While the contractors and consultants entering this space for the first time gain remarkable resourcing, existing network players, including strong hubs of activation, struggle for survival, with staff living from contract to contract, and no organizational runway available to them to enable the strategic planning or expansion work they may be uniquely suited to do. In other words, rather than directing funds to strong cornerstone players (e.g., EDGI, PEDP) in support of the (volunteer-driven) bridge building they already have underway, resources are going to external and/or new groups.
This problem is structural, not intentional, but it still deletes resources from the grassroots actors and specialists in the field in at least three ways. It happens first as dollars flow out of the information management ecosystem to fuel the work of often expensive, deliberately temporary players that promise to provide a system-level view and unbiased recommendations. It happens again, as those consultants build their own maps and recommendations only by drawing heavily on the time and knowledge of the very specialists they are shadowing — asking them to explain what they do, who the key voices are, who is and isn’t connected, and where funding is most needed. And then it happens again as the specialists don’t spare attention for funding calls and proposal deadlines because they are spending their time doing the work.
An unresolved strategic question
Underneath the funding gaps lies a more fundamental question that the field has not yet answered: what is today’s data rescue and resilience effort actually building toward? There is a significant difference between building survival infrastructure to keep specific repositories and datasets alive through the current period of threat, versus building perpetuity infrastructure – systems and services that repositories and datasets can plug into for long-term, stable access regardless of political conditions or other threats. Both are needed. Neither is being resourced with that distinction in mind.
Key, instructive models and seasoned players, ones that have actively solved this problem for adjacent challenges, seem to be missing from the current networks that are assembling: for example, CLOCKSS and Portico for journal content, DataCite for research data identifiers, Crossref for publication metadata, OpenAlex as an open catalog of scholarly works, and Digital Preservation Coalition and National Digital Stewardship Alliance for community standards. Each of these longstanding and well-embedded infrastructure elements has succeeded by defining a function, building infrastructure to serve it collectively, and creating a governance and funding model that distributes the cost across many stakeholders. None of them emerged from crisis response; all of them were built in periods of relative stability. Many of them could provide strong, grounded perspective and leadership to help tie today’s activity to years of preservation practice.
The current moment is not stable, but it is not too early to ask what long-term preservation and access infrastructure for federal data looks like, and to begin directing some portion of crisis-response investment toward building it.
What this means
The data resilience field is doing to data rescue what scholarly communications has done to open infrastructure for thirty years: funding the visible and easy-to-narrate activities while underinvesting in the operational and technical layer that makes those activities durable. Coordination without tools is a map without roads. Rescue without perpetuity infrastructure is a temporary holding action, not a long-term solution.
Three specific redirections could materially improve the current investment picture.
- First, dedicated funding for the tools layer, including web archiving software, ingest pipelines, metadata infrastructure, and identifier systems needs to appear in funding portfolios alongside rescue and coordination projects. Both types of funding are needed, and neither can be substituted for the other.
- Second, structured knowledge exchange with communities that have done this work before, e.g., the digital preservation community, the 2016-2017 data rescue efforts, SUCHO and other more recent engagement, should itself be an active, funded transfer-of-knowledge activity. Where are there tools, models, and examples that we might build on, and who can quickly help us identify their value? Funding these groups to do this research could vastly improve our readiness to act in the near future.
- Third, instead of funding external groups to tell us what our experts know, let’s fund the lynchpins that are already in place and doing the work, and let’s free up some of their time to help surface and advance the models they are already successfully using. Too many of the organizations best positioned to lead this work, including the established data stewards and the coordination hubs, are operating on project-to-project funding with no runway for strategic planning. Shoring up those existing human and organizational network agents so that they can shift from crisis response to longer-term strategizing is likely to be the best investment we can make.
The urgency of this moment is real, and so is the opportunity it offers. The infrastructure we build in response to this crisis will shape what is possible for data access and preservation for a generation. Building it on the same underfunded, fragmented, misaligned model that has characterized open infrastructure for decades would be a preventable mistake.
IOI is actively mapping the data resilience landscape and tracking investment trends across the ecosystem, and as we do so, we are grateful for the help and perspective of many other groups, including American Geophysical Union, Center for Open Science, Data Rescue Network, Internet Archive, Public Environmental Data Partners, and The Data Foundation. Please contact katherine@investinopen.org to contribute to or build on this analysis.
