Cancer Risk Is Still There, Even If the Data Isn’t
Can you estimate someone’s cancer risk from the air they breathe? Not exactly. Every person has different genetics, lifestyles, occupations, and environmental exposures that change over time. Air pollution varies from day to day, season to season, and even block to block. But scientists can make remarkably reliable estimates of a neighborhood’s cancer risk based on the type and amount of toxic air pollution.
After a year-long delay, this April, EPA released the latest air toxics data, which only included raw air data downloads. This year, for the first time in nearly 25 years, the air toxics data did not include cancer risk estimates.
Without a public explanation or opportunity for input, one of the nation’s most important environmental health datasets has quietly gone dark.
The disappearance of EPA’s cancer risk data continues a broader trend under this administration of environmental, public health, and other government datasets becoming less available, less complete, or more difficult to access.
What’s Changed?
For the 2021 data (this most recent release), only raw air emissions and concentration data are available for download. Previously, the EPA also produced detailed explanations and a mapping tool that made the data easier to discover, understand, and use. Additionally, they included cancer risk estimates to help translate data and complex scientific models into information the public could understand.
Now, if users want to understand cancer risk from air pollutants, they will need to download and analyze detailed air concentrations of nearly 200 toxic air pollutants for more than 8 million individual locations.
Researchers and geospatial data professionals can navigate government databases, process large datasets, and build custom analyses or maps. Everyone else relies on easy-to-use websites, interactive maps, and plain-language explanations to understand the environmental conditions affecting their communities.
Data hidden behind obscure downloads may satisfy a technical definition of transparency, but they do little to promote public understanding, meaningful community engagement, or actions to save lives.
Why Cancer Data Matter
Cancer affects nearly every family in the country, and the United States has one of the highest cancer incidence rates in the world. Understanding where people face higher exposure to cancer-causing pollutants is a key step toward reducing preventable risks.
Most people have no easy way of knowing what hazardous pollutants are being released into the air around their homes, schools, or workplaces. Air toxics are often invisible, odorless, and linked to health effects that can take years or even decades to develop. EPA’s cancer risk estimates gave communities, local governments, researchers, and others a way to identify areas where cancer risks from air pollution may be elevated and where action or additional monitoring might be needed.
History of Cancer Risk Data
The Clean Air Act lists nearly 200 toxic air pollutants known to cause cancer or other serious health effects, and requires reporting of these emissions by facilities. Amendments in 1990 directed the EPA to establish risk standards for any source emitting a cancer-causing pollutant that poses a lifetime risk of cancer of more than one-in-a-million.
EPA began publishing nationwide cancer risk estimates from hazardous air pollutants in 2002 with the first release of the National Air Toxics Assessment (NATA). These cancer risk estimates took hazardous air pollutants information from across the country (using facility emissions, vehicles, and other pollution sources), and combined them together with atmospheric modeling and toxicological research, making the data easier for the public to understand and use. This dataset became a core component of EPA’s mapping applications and was incorporated into numerous federal, state, academic, and community tools. Like any national-scale model, the NATA cancer risk data had limitations. It relied on emissions inventories, many of which were self-reported by industry, and modeled pollution levels for some locations, rather than direct air monitoring. It could not fully capture localized conditions or cumulative exposures, and the results often lagged several years behind current conditions. But public feedback on those limitations also inspired continuous improvements.
Over time, EPA scientists improved the quality of the pollution data, updated what scientists know about the health effects of toxic chemicals, and made the models more accurate. In 2022, the agency rebranded the program as the Air Toxics Screening Assessment or AirToxScreen, making the data more accessible and actionable through a dedicated mapping platform. With EPA having done the heavy lift of creating cancer risk estimates, other organizations can incorporate them into local processes. For example, New Jersey’s Environmental Justice, Mapping, Assessment, and Protection Tool (EJMAP) uses cancer risk data to help evaluate the cumulative impacts of facilities before approving or renewing permits.
Communities Impacted by Air Pollution-Related Cancer
Perhaps nowhere are both the strengths and limitations of these EPA data more apparent than in Louisiana’s Cancer Alley. This 85-mile industrial corridor along the Mississippi River contains one of the highest concentrations of petrochemical facilities in the United States, accounting for roughly one-quarter of the nation’s production. Residents have spent decades raising concerns about pollution and unusually high cancer rates. Here, some communities experience estimated lifetime cancer risks from toxic air pollution up to 50 times higher than the national average, and more than 47 times higher than EPA’s acceptable cancer risk benchmark. These findings have helped draw national attention to environmental justice concerns and have even prompted international human rights scrutiny.
At the same time, Cancer Alley demonstrates why screening and mapping tools alone are not enough. AirToxScreen is based largely on emissions inventories and modeling assumptions. It cannot fully account for cumulative exposures across multiple pathways, capture every emission event, or reflect neighborhood-scale differences that can vary dramatically within a city. Similar concerns have been raised by communities along the Houston Ship Channel, Chicago’s industrial corridors, and other heavily burdened areas across the country.
For these communities, data like AirToxScreen’s cancer risk estimates are an essential first step, but need to be accompanied by direct air monitoring, epidemiological research, or community-based science. And data need to be tied to action, such as strong environmental enforcement, in order to reduce Americans’ exposure to cancer-causing chemicals.
Additional Issues and Concerns
The disappearance of EPA’s cancer risk estimates comes at a time when protections against hazardous air pollutants are also under increasing pressure. The current administration has proposed rolling back or reconsidering several air toxics regulations, including standards for pollutants such as ethylene oxide, one of the largest contributors to cancer risk from industrial emissions.
At the heart of many of these proposals is a broader debate over what level of cancer risk should be considered acceptable and how that risk should be calculated. Loosening these standards does not reduce the amount of pollution in the air or the health risks facing nearby communities. Instead, it changes how those risks are defined and regulated. Combined with the loss of publicly available cancer risk estimates, these policy changes make it more difficult for communities to understand, communicate, and respond to the environmental health risks they face.
What’s the Data Policy Angle?
For more than two decades, EPA scientists steadily improved these estimates despite chronic underfunding, changing regulations, and shifting political priorities. Those challenges have only intensified with the dismantling of EPA’s Office of Research and Development and the erosion of scientific capacity across the agency.
EPA’s cancer risk data, though imperfect, has been an indispensable resource for researchers studying environmental disparities, journalists uncovering pollution hotspots, regulators prioritizing inspections, and communities advocating for cleaner air.
Now, the cancer risk estimate data so many depend on has disappeared. There was no public notice, no opportunity for public comment, and no consultation with one of several terminated scientific federal advisory committees.
The lack of public engagement is a policy problem.
The Open Government Data Act (OGDA), signed by President Trump in his first term, and subsequent OMB Guidance in M-25-05 include explicit requirements that agency data stewards engage with public stakeholders to understand the value of federal data, and ways it could be improved to better serve the needs of the American people.
OGDA and M-25-05 also unambiguously state that agencies:
- shall assist “the public in expanding the use of public data assets”; and
- “must provide adequate notice when initiating, substantially modifying, or terminating significant information dissemination products.”
The disappearance of EPA’s cancer risk data is just the most recent signal that the current patchwork of federal data policies is increasingly unprepared to address the challenges and opportunities facing our nation.
These are exactly the types of issues that the FAS Data Policy Institute is taking on, because cancer risk does not go away when the data disappear.
Debris-Creating Anti-Satellite Tests Score Low on Strategic and Military Benefits
There are more than 1.2 million pieces of space debris larger than one centimeter in orbit today; under the right circumstances, any of them could cause catastrophic damage to a nation’s space economy, its communications technology, and even its nuclear command and control infrastructure. And yet, the amount of debris in space is increasing rapidly, while countries continue to scale up their space programs.
An outsized amount of trackable space debris can be attributed to a relatively small number of low-frequency, yet extremely high-consequence events: instances of states shooting down their own satellites. These direct-ascent anti-satellite (DA-ASAT) tests, as they are known, have been conducted by the United States, Russia, China, and India, and are designed to test these countries’ abilities to destroy their adversaries’ satellites during wartime. Past tests have created space debris, increasing risks of collision and potential damage to other states’ satellites. There is shared recognition of the risks posed by space debris and such debris-generating tests. Debates on debris-creating DA-ASAT tests were prominent in the most recent session of UN space security talks in July. While exchanges were relatively more constructive than previous sessions, long-standing divisions reigned over what types of capabilities to regulate, how to verify them, and where to start. Some continued to advocate for a ban on debris-creating (‘destructive’) DA-ASAT tests, even suggesting a legally-binding ban, while others argued that focus on testing was too narrow and did not cover development, deployment, and use, or other capabilities of concern such as space-based weapons.
States’ views on missile defense have largely contributed to these debates, posing significant challenges to concrete de-escalation measures on multilateral or bilateral bases. Given the strong degree of technological overlap between missile defenses and DA-ASAT, states may perceive any suggested regulation of the latter as inherently limiting the former.
These debates have not captured the limited military utility of such tests. This essay argues that despite technological overlap, there is scope to decouple missile defense from DA-ASAT technologies on the basis of the limited military utility of debris-creating tests. Accordingly, states should consider commitments to ban debris-creating tests.
Overlap between missile defense and DA-ASAT technologies
There is an immensely strong degree of overlap between the technologies underlying missile defense interceptors and DA-ASAT weapons. Both systems require the same elements–rocket motors, guidance, navigation, control systems, and a payload. In addition, the technical knowledge and industrial bases needed to build them are largely identical. As a result, a capability to develop one system constitutes a latent capability to develop the other.
Indeed, repurposing a missile defense system for a DA-ASAT strike against a satellite can be less technically challenging than a missile intercept, depending on the state’s capabilities. Two components are essential: first, that the missile defense system can deliver a sophisticated kill mechanism to a high enough altitude to reach its target. A missile capable of launching a payload to its maximum range can typically launch that same payload vertically to an altitude of about half that range. For instance, a missile carrying a kill vehicle to a maximum range of 2,000 km (1,243 miles) could potentially carry that same kill vehicle to an altitude of approximately 1,000 km (620 miles). Second, the state would require advanced space situational awareness capabilities to determine the precise orbital parameters of its target. Satellites travel on predictable, observable trajectories and have limited maneuverability, thereby making them relatively easier targets for DA-ASAT systems. In contrast, a missile intercept mission could take place at any time, and requires both a sufficient alert posture and a degree of nimbleness that are not required for a DA-ASAT mission.
Importantly, DA-ASAT missions also do not encounter discrimination challenges to nearly the same degree as missile defenses. Modern ICBMs may contain combinations of decoys (such as mylar balloons) or penetration aids (such as chaff) that are meant to confuse radars or help evade incoming interceptors. While there are certain combinations of active and passive defenses that could theoretically be built into a satellite architecture to enhance resilience against attacks, many of these concepts remain aspirational and have not yet been deployed at scale.
As a result, a state that already possesses sophisticated missile defense capabilities would be able to repurpose those same capabilities for DA-ASAT missions without significant difficulty. For states reliant on space for military and civilian missions, the rise in such capabilities and their testing poses major concerns.
Different systems for different targets
Whether a particular missile defense system can be adapted for a DA-ASAT mission depends to a large extent on the orbital band where the targeted satellites are located.
Most strategic early warning satellites are located in geostationary orbit (GEO)–at an altitude of 35,786 kilometres–and highly elliptical orbit (HEO), in which satellites spend the majority of their orbit at an apogee of similar altitude. In order to kinetically target satellites orbiting at that altitude, an interceptor would need to achieve a degree of delta-v that is simply not feasible for nearly all missile defense systems in existence today, although this will likely change as technologies evolve. Some states allege that others are pursuing DA-ASAT capabilities to target satellites in GEO, but no state has publicly acknowledged such capabilities, and there is no evidence of their operational status.
Instead, the majority of operational missile defense systems are most suited for targeting satellites in low Earth orbit (LEO), and to a lesser extent, medium Earth orbit (MEO). These orbits typically include satellites with applications for telecommunications, remote sensing, military intelligence, surveillance, and reconnaissance (ISR), and navigation (including GPS, GLONASS, Galileo, and BeiDou), among other capabilities. In most cases, a missile defense system that is designed to intercept in midcourse phase will be most appropriate for a DA-ASAT mission against satellites in these orbits, although some terminal systems may also be relevant depending on the target and context.
These systems carry kinetic kill vehicles (KKVs) that utilize hit-to-kill technology. After the missile booster reaches a high enough altitude, it releases a KKV equipped with sensors that home in on the target and destroy it using kinetic force. If the guidance systems are accurate enough, the high closing speeds of the KKVs can project significantly higher force than a blast-fragmentation warhead, rendering them unnecessary.
There are several states that possess DA-ASAT capabilities, with four having conducted destructive tests: China, India, Russia and the USA. Among the five permanent members of the UN Security Council (the five recognized nuclear-armed states under the NPT), Chinese, Russian and US missile defense systems with ASAT potential is evident.
China has multiple systems capable of targeting satellites in LEO. These include the mobile HQ-29 system, which according to a Chinese military magazine, can carry up to two interceptors and engage targets at an altitude of 2,000 kilometres–as well as the DN-3, the HQ-19, and the SC-19. The SC-19 is believed to be derived from the DF-21 MRBM, and has previously achieved intercepts at altitudes of approximately 250 kilometres (155 miles) and 865 kilometres (537 miles), the latter resulting in the destruction of an orbital target in 2007 that reportedly created over 3,500 pieces of trackable debris (see page 05-01).
Russia similarly operates several systems that could function as both upper-tier missile defense systems and anti-satellite weapons. These most notably include the Nudol, which the US Defense Intelligence Agency assesses was used to conduct a test against a defunct Russian satellite in November 2021. The US Space Command estimated the test created over 1,500 pieces of trackable space debris, while the Russian Ministry of Foreign Affairs declared that debris from the test did not create any threat.
The United States operates both the Ground-Based Midcourse Defense (GMD) system, which is designed to intercept strategic ballistic missiles while they are traveling in their midcourse phase of flight, as well as the SM-3––which was used to destroy a non-functioning reconnaissance satellite in 2008, that created approximately 175 pieces of trackable space debris.
The USA in 2025 also announced a major shift in missile defense policy, with ‘Iron Dome for America’ (later renamed ‘Golden Dome’), to defend the entire US homeland from all types of missiles from all adversaries. While this initiative aims to implement a layer of space-based interceptors, which are distinguished from DA-ASATs (as they would be space-based), it will also likely include enhancements to the ground-based interceptor force, including the SM-3. Following the announcement of Golden Dome, fears about possible tests for such systems have also been rekindled.
Notably, no state has ever used a DA-ASAT to attack another state’s satellite; this is unprecedented, would be considered a clear use of force under international law, and invite escalatory responses.
Governance and initiatives
There are currently no restraints on missile defense under bilateral or multilateral instruments. The 1972 Anti-ballistic Missile Treaty introduced certain limitations between the US and the Soviet Union, later succeeded by Russia in the treaty. Parties agreed to maintain only one strategic missile defense site with up to 100 launchers and 100 interceptor missiles, and committed not to develop, test or deploy sea-based, air-based, space-based or mobile-landed ABM systems or components. However, the US withdrew from the agreement in 2002, a move that in part has driven the expansion of missile defenses and maneuverable missiles among competitors.
US withdrawal from the ABM Treaty had ramifications on space security talks, amplifying concerns about the US pursuing space-based missile defense systems that could undermine Chinese and Russian deterrents–concerns further fuelled by USA’s Golden Dome. Under the 1967 Outer Space Treaty, placement of WMD in orbit is expressly prohibited, with no consensus on the regulation of other types of weapons. The issue of DA-ASAT testing has seen renewed interest following the rise in destructive testing. In 2022, the USA pledged to refrain from destructive DA-ASAT tests, referring to previous Chinese and Russian tests. The US commitment did not however, mention its own previous tests or India’s test in 2019. Several states made similar national pledges in the months that followed, culminating in a US-led resolution adopted by a wide majority by the UN General Assembly in 2022 to refrain from debris-creating DA-ASAT tests. Nine states abstained from voting, including India, and nine voted against, including China and Russia. The issue of DA-ASAT testing has since been raised in subsequent UN space security processes, including the July talks. These exchanges would benefit from more comprehensive inquiry into the utility of debris-creating tests.
Low military utility of DA-ASAT tests
Much has been reported on harmful effects of debris from destructive DA-ASAT tests, which directly threatens space-based assets of all actors. Debris can even interfere with crewed missions, threatening human space exploration. Even tests conducted at lower altitudes still generate debris and increase risks of collision. LEO is considerably overcrowded with debris, as well as active ‘megaconstellations’ of satellites–such as SpaceX’s Starlink–indicating significant risk of collision already.
Yet, there is less emphasis on the low military utility of these tests, both in peacetime and in times of rising tensions preceding a conflict.
First, a destructive DA-ASAT test has permanent and indiscriminate effects. Debris from the test could undermine space-based assets that enable critical conventional weapons and operations for the executing state, which would be detrimental to its own interests.
Moreover, such a test could transform adversary dynamics by affecting other actors, as there is no specific effect against a singular adversary. Even if a state is more risk-seeking and less reliant on space, it could alienate allies or strategic partners reliant on space and isolate itself politically.
Second, the attribution of such tests is also undisputed, making it clear who the executing party is, and subjecting them to political isolation. This also highlights the ineffectiveness of strategic or coercive signaling for a destructive test, because it is unlikely to influence a specific adversary’s cost-benefit calculation without also impacting other stakeholders.
Third, conducting such a test would not signal a new capability to an adversary. States are well aware of the missile defense capabilities at their adversaries’ disposal; there is thus no need for a debris-creating test to convey their effectiveness. Open-source assessments, the use of these systems in ongoing conventional combat scenarios, and public parades of missile defense systems are already creating this effect.
Fourth, kinetic tests are not needed to validate a state’s operational DA-ASAT capabilities. As already discussed, these missions are significantly easier than conducting realistic missile defense intercepts, given the predictable trajectories of satellites. As a result, states can validate their capabilities through other means, including missile defense intercept tests and simulation technologies. To put it another way, if the United States was not confident in its ability to disrupt the satellite operations of its adversaries during wartime, it almost certainly would not have issued its 2022 moratorium on destructive DA-ASAT tests. Destructive testing of these systems is not necessary.
Steps forward: Seeking bilateral and multilateral test-ban commitments
Space security talks have historically sought common understandings on DA-ASAT systems focusing on debris aspects. The low military utility of such tests should also be a factor. This would allow states to consider widening the scope of previous efforts on destructive DA-ASAT tests–possibly extending these to all debris-creating tests, including those involving co-orbital systems. This is especially relevant given US pursuit of Golden Dome and its space-based interceptor layer, which has co-orbital ASAT potential. In addition, proposals should uniformly acknowledge states that have conducted past tests. Selectively naming and shaming actors that have conducted DA-ASAT tests only further drives political agendas and reinforces divides already tangible in UN forums. A balanced approach to regulate capabilities would in turn help secure buy-in from states that are more wary of the narrow scope of previous proposals for a DA-ASAT test ban.
Given the strong degree of technological overlap between defensive interceptors and offensive DA-ASAT weapons, regulation of such systems is no small feat, particularly with arms control at a standstill. Eventually, states could consider pairing test-ban commitments with restrictions on interceptors in strategic stability talks. This could spark consideration of measures that extend to development, deployment and use. However, in the current environment of declining trust and magnifying strategic competition, banning debris-creating tests is a concrete step towards stability that favors interests of all states seeking to conduct space activities. Emphasizing the low military utility of such tests is essential for more constructive exchange.
Note: Conversions from kilometers to miles are approximate.
How We Found Eight Data Center Community Benefit Agreements and Why It Matters
These days, data centers are the hot-button topic in policy circles everywhere. Between the explosion of AI and the sheer scale of some of these facilities, it seems like people cannot go a single day without hearing them come up in some capacity.
Communities weighing a data center project are considering real opportunities: construction jobs, new tax revenue, and the promise of economic development that comes with a major capital investment landing in town. But they are also considering real costs. Data centers bring elevated noise, especially for homes sitting close to a facility; they also can consume significant amounts of water and energy, affecting rates and the energy grid. And the tax breaks that some of the data centers receive could outweigh any financial benefits the community gets in return.
Local governments face a surmounting number of questions, with best practices not readily available. The policy decisions we make today will affect much of the work that comes tomorrow and the decades after that. Beginning in early 2026, FAS began interviewing experts in the field, as well as searching for executed agreements on data center developments that were difficult to find. This included tracking local government ordinances as well as finding Community Benefit Agreements (CBAs).
Many times, the negotiations between data center developers and officials from the municipality or county are conducted behind closed doors, protected by non-disclosure agreements (NDAs). As a result, community members have significantly fewer opportunities to influence the terms by which the data center should abide by as a fixture in the community.
In the spirit of increased transparency, FAS wants to share our experience of “flipping over rocks” and discovering the hard-to-find contract. This is how we discovered eight community benefit agreements and why it matters.
Local Policy Levers: The Ordinance and The Contract
Generally, local governments have two enforceable levers to create policies around data centers. First, an ordinance considers issues such as zoning, land use requirements, and other permitting features – ultimately providing a pathway to allowable activities (and specific limitations). The ordinance process includes a public comment period, and it applies to all projects that qualify under the definition of a “data center.”
A CBA involves the power of contract law, crafting and entering into binding bilateral (or sometimes multilateral) agreements that dictate the expectations of each party. CBAs are helpful in cases where existing zoning and permitting do not specifically identify “data centers” as a specific type of property to regulate (this is most frequent as the proliferation of data centers is so new). However, CBAs can be negotiated with fewer public comment opportunities, and they are required on a case-by-case basis.
CBAs are not a new legal concept, and they have been used for large scale projects before on major development projects, like airports or energy generation infrastructure. They very recently started to be used in the context of data centers, after the explosion in usage of AI. Thus, FAS went searching for every agreement that fit the CBA model.
As of the time of this writing, only one fully executed and publicly posted CBA exists that pertains to data centers: The Lancaster AI Hub agreement entered into by the City of Lancaster, Pennsylvania and the developers of two large data center campuses in the city. Lancaster’s CBA contains many terms and conditions that are beneficial to a data center agreement with the host community. It covers community transparency, water use, energy, noise restrictions, funding for public works, decommissioning, and many more. However, this agreement appears to leave much of the substance to be decided later on. Oftentimes, it requires the parties to submit separate plans to the city which detail the specifics of how the developers will conduct themselves.
Finding Agreements: Discovering CBAs
Finding executed CBAs took expertise, a concentrated effort, and follow-up. To expand beyond the Lancaster CBA, FAS first turned to blog posts that mentioned CBAs from sources like the Brookings Institution and Columbia University to see if there were any stories about localities that are negotiating terms with data center developers, or that have already done so. Initially, that is how FAS found Cedar Rapids, IA, St. Louis, MO, and El Paso, TX. Then, we looked for news articles referencing data center construction in these cities to see if they mentioned anything about an agreement between the city and the developers. If the terms of the agreement, or even discussion about its existence, were covered in these articles, that was where we knew we had to dig deeper.
While the majority of agreements found were publicly available, the only way to obtain some of them was through a public records request. Thankfully, most states did not have such broad prohibitions on what can be public record, allowing FAS to uncover more contracts to serve as references.
Over the summer of 2026, seven other agreements were uncovered besides the one in Lancaster between cities, counties, and data center developers that mention community benefits:
During the search for more agreements, FAS was making note of the trends in each one, and comparing trends among the different agreements. Many of these agreements contain a heavy focus on money and taxes. Tax abatements, exemptions, and Payment in Lieu of Taxes (PILOT) schemes served as incentives for the data centers to be built in various communities. In total, we found the following policy areas considered across multiple CBAs:
FAS’ Forthcoming Analysis + Next Steps
By putting a light on the language of multiple CBAs, we can start to understand what is commonplace, what sections need to be tailored to local needs, and where there is a need to consider more comprehensive policy requirements (e-waste, for example). In late Summer 2026, FAS will publish a report analyzing and comparing the various sections of each CBA. The report will also include a template CBA for the taking, and a summary of recent trends in local ordinances.
In the meantime, we hope that others will continue to look for these agreements, and apply pressure to private partners in providing them. How can we have an informed debate, when we don’t know what’s on the table?
If your community or local government has an executed CBA, please send it to metrolab@fas.org.
This will enable us to build best practices and inform communities nationwide on how local governments are crafting policies. Every community is different, and will therefore have different needs, underscoring the necessity for thoughtful decision making, not hasty and impulsive actions. It is also imperative to learn from these early agreements. Comparing them against each other allows us to understand early norms, “extreme” ranges, and expected outcomes from various projects. And by having these baseline terms, governments can enter into agreements with data centers with a clearer sense of what can be negotiated on a local level, what is reasonable to require, which policy areas need strengthening, and how to enforce these agreements over time.
How Do We Track Terminations of Federal Data?
Federal data benefit American lives and livelihoods in ways most people never see, touching every corner of our lives. This includes a farmer pricing a crop, a county planning a hospital, a business siting a warehouse — all of these decisions use federal data. Other data save patients money by identifying generic drugs that can replace more expensive brand names, help airplanes avoid deadly bird strikes, and warn consumers about recalls of dangerous products.
Because these data are mostly invisible, their disappearance is invisible too — that is, until we need the data and they aren’t there anymore.
As federal data policy nerds, the question we get asked all the time is How much data has the current administration terminated?
The answer is that it depends on what we count as “data” AND what counts as a termination. That’s not a dodge – working through those two critical nuances is the substance of this piece.
The answer also depends on what the information will be used for. Ours is a data policy question, so we looked for structured, numerical datasets that have been terminated – meaning there will be no collections of those data in the future. Defining terminations that way lets us ask how agencies consulted with the public about a dataset’s value before ending it, and what its loss means for the federal government’s ability to serve the American people.
The purpose of the Federal Data Terminations Tracker is to be the most policy-relevant, verified accounting of federal data terminations available.
To create this Tracker, the dataindex.us team identified dozens of federal datasets and hundreds of data elements that have been terminated – significantly fewer than other reports, but more tailored to informing future data policies needed to run a modern society. These are data that have long underpinned policymaking, journalism, advocacy and research that improve American lives and livelihoods. These figures will change as terminations continue, collections are merged, and as court orders restore data.
Want more details on why this question about federal data losses is so tricky? Read on. Want to see what data have been terminated? Visit the Federal Data Terminations Tracker at dataindex.us/terminations-tracker.
Why are there different numbers on federal data losses?
The number of datasets in data.gov is not a meaningful metric.
Studies and news reports over the last 18 months have varied widely, with the number “3,000 datasets removed” often being cited. That figure generally traces back to data.gov’s catalog counter, which dropped by roughly 3,000 entries in the administration’s first weeks. The counter is a poor measure of terminations, though.
In early spring 2025, Harvard’s Library Innovation Lab watched the data.gov collection swing up by 6,000 datasets one month, and then back down by 2,000 the following month; they note that fluctuations like these arise naturally as websites and data stores are reorganized. Also, the definition of a “dataset” is broad in data.gov, including datasets, parts of datasets, and other data assets, meaning it could easily count one American Community Survey as hundreds of “datasets” (if you count each of the many files that are released each year).
Some definitions of “data” are broader.
Counts from rescue efforts like the Data Rescue Project or EDGI measure a broader range of information losses, including web pages, files, reports, and historical data taken offline. Additionally, other definitions, taking a broad definition of what constitutes data, focus on content that has been removed from public access, rather than data collections which are terminated going forward.
What should be included when we count data terminations?
Easy choices
Let’s start with the easy part to define. When an agency ends a primary data collection – for instance, the survey stops going out, the forms stop being filed, or the sensor stops collecting data – that is an obvious data termination. The USDA’s CPS Food Security Supplement, SAHMSA’s Drug Abuse Warning Network, and the USDA’s Mink Survey all ended this way. This is clear cut and little judgment is required to declare them “terminated.”
Sometimes a collection is terminated, and technically other entities could or do collect that data, but without the gravitas of the U.S. Government. For example, until last year, many U.S. Embassies collected air quality data and reported those numbers publicly. Of course, others can still collect air quality data for these cities, but the official U.S. data provided an unbiased source of data to compare to numbers from local officials, which research showed resulted in “substantial reductions in fine particulate concentration levels.”
More difficult choices
The next group – derivative products and modeled data – takes more thought. These include things like composites, indices, and model outputs that are built from data that still exist somewhere in government (or outside of it), so ending them doesn’t destroy an underlying collection of data.
We include these when they can’t be reproduced outside of government easily, or when they would stop serving their purpose if they were anything other than a government asset. An example is NOAA’s terminated Billion Dollar Weather and Climate Disasters data. Climate Central hired the NOAA researcher behind the data and reconstructed the model. Stewarding this dataset outside of government requires substantial resources and expertise. Being published by Climate Central means that the data may become more accessible and relevant as the nonprofit improves the product, but its effectiveness as a tool for policy change and action might be reduced without the imprimatur of the federal government.
We include data like the Billion Dollar Disasters data in our count, while recognizing that the line is blurry and reasonable people will draw it in different places.
The level of granularity that we count affects the total number of data terminations even more than what we count. For example USAID released multiple files for each country studied in the Demographic and Health Surveys, and each one could be called a “dataset,” but these surveys constitute one data product for our analysis. Counting files is a choice that, in our opinion, inflates the number. Instead our unit of analysis, what we actually count, are products.
BEA presents the same problem in reverse. Its collections continue, but the published tables and reports, which are sometimes the only place where the public has access to the data, have stopped. An example of this is metro-area GDP, the official measure of whether individual regional economies were growing or shrinking. Is each table or report a termination? Should we count these at all? We landed somewhere in between: we count them as a single discontinued data product for the whole agency and note that the collections behind them survive (as well as providing links to BEA’s own list of terminated reports).
Most difficult choices
Further, “terminated” is not necessarily a forever status. Most of the data taken down to comply with the day-one executive orders came back. The CDC’s Social Vulnerability Index returned by court order, but data are no longer updated. The Household Pulse Survey’s gender identity fields were removed, then back in the historical data by May 2026, but are no longer part of the survey moving forward. Other cases sit in a gray zone: CDC’s PRAMS is still collected but is no longer published at the federal level, and does not receive standardized national weights. The Federal Employee Viewpoint Survey was terminated and is “being reenvisioned.” The Violence Against Children and Youth Survey finished its pilot, lost its team to reductions in force, and ended without any formal notice or publication of results.
How do we measure “data termination”?
In order to be most relevant for data policy, we count a data termination as:
- the documented ending of a primary federal data collection;
- the discontinuation of a derivative or modeled data product that could not easily be reproduced outside government, or would not serve its intended purpose if it were not a government asset; or
- the removal of substantive elements from a continuing product (counting at the level of the data product).
Additionally, a data termination must be tied to primary evidence, for instance an agency notice, a press release, or a regulatory filing.
We track but do not count temporary takedowns that have been restored in our current total (though restored data doesn’t equal data that is safe forever). This means our number of terminated datasets is versioned, and updated at least quarterly.
So how much data has been terminated? (as of July 2026)
With the scope set defined above, we can get to the numbers. Terminated data number roughly two dozen (28 data products as of July 2026). Some of these terminations were relatively easy to find.
- USDA issued a press release when it terminated the Current Population Survey’s Food Security Supplement, calling the survey “redundant.”
- The National Agricultural Statistics Service posted a notice ending the Agricultural Labor Survey and the Mink Survey.
Others ended quietly, or were never announced at all.
- The National Law Enforcement Accountability Database was shut down and the record of its termination coming from the Congressional Research Service rather than the agency.
- The Violence Against Children and Youth Survey finished its pilot and then simply stopped, its team gone from reductions in force (aka, RIFs), with no formal notice anywhere.
- The Demographic and Health Surveys ended with the shuttering of USAID, which had an international impact by cutting off health and demographic data for dozens of countries that had relied on the program for decades. The government contractor formerly responsible for collecting these data has received outside funding to continue.
Removed data elements are a larger number. Hundreds of continuing datasets have lost variables, most commonly gender identity, sexual orientation, race, and ethnicity.
- The Bureau of Prisons no longer reports gender identity for inmates in federal prisons. This change happened overnight in a data report that was not archived, so we have no access to a before and after to understand how the agency reclassified the data.
- FedScope, OPM’s public portal for data on the federal workforce, dropped race and ethnicity in March 2025 when its diversity data was removed.
Some of these changes are visible in the paperwork if you know where to look.
- The reginfo.gov filings (information collection requests or ICRs) for the National Survey on Drug Use and Health and the Census of Fatal Occupational Injuries both document the removal of gender.
- For sexual orientation and gender identity measures alone, the inventory compiled by MAP and the Williams Institute runs to 400+ datasets.
Following our granularity rule, BEA’s discontinued tables and reports count as one product, with a link to the agency’s own inventory.
The 350 discontinued Producer Price Index series counts as one as well. The Billion-Dollar Weather and Climate Disasters dataset and the Future Risk Index each count on their own. These derivative data products aren’t easily recreated outside of government (though some like the Billion-Dollar Weather and Climate Disasters dataset, have been).
Two judgment calls
Everything above describes what we count. Two harder questions sit underneath the count, and we want to be clear about where our methodology becomes more subjective.
The baseline problem
The first is the baseline problem. Data terminations happen in every administration. Statistical agencies retire collections as industries shrink, methods improve, and budgets tighten. BLS maintains entire pages of discontinued databases and series, and most of its recent CPI discontinuations date to 2024. Judging any single termination means asking whether it is part of that ordinary churn or something above it.
BEA’s metro-area statistics show how ordinary a termination can be. The agency stopped publishing personal income and GDP for metropolitan areas, and at first glance that looks like exactly the kind of loss we track. But BEA documented its reasons: publishing metro aggregates forced more privacy suppressions on its county statistics, and shifting OMB geography definitions made the aggregates unreliable across the full time series. The county-level data continue, and the agency released a geographic aggregator tool so users can build their own metro estimates. We do not consider this termination above baseline, and therefore is not counted in the Terminations Tracker.
The Mink Survey is the harder call. National Agricultural Statistics Service (NASS) ended it in August 2025, saying it was no longer necessary, and the mink industry has in fact been shrinking for years. Read that way, it is routine modernization. But mink farms were a documented site of COVID transmission between animals and people, which makes the survey relevant to public health, not just to agricultural economics. Our assessment is that its public-health value argued for keeping it.
Uniqueness to government
The second judgment call is whether every termination is a loss that only the government can repair. Some data can be produced outside the government. Public Environmental Data Partners and Fulton Ring’s revival of the Homeland Infrastructure Foundation-Level Data (HIFLD) is one example. But that path exists only when the inputs are public and a well-resourced steward steps forward, and neither is guaranteed. No private group can publish misconduct records held inside federal law-enforcement agencies the way the National Law Enforcement Accountability Database did. When data like those end, they simply end.
These two judgment calls are why we publish our reasoning alongside our numbers, which brings us to how we actually did this research.
How we did the research
dataindex.us monitors America’s federal data infrastructure, from dataset availability and new releases to planned and unplanned changes to collections. Our team aggregates the OMB information collection request pipeline, which is otherwise difficult to find and harder to interpret, flag opportunities for public comment, and track selected datasets and their documentation so that when data go missing, get reposted, or fail to release on schedule, there is a record. We mix primary data collection, both manual and automated, with secondary sources to better understand the overall health of federal data.
Finding terminations is harder than it sounds, because there is no single place where they are announced. Federal data policy is fragmented, and some types of data are far more visible than others. Surveys and forms covered by the Paperwork Reduction Act must go through public notice and comment for any substantive change. Curiously, that requirement covers changes – but not terminations – unless the data collection is tied to a regulation. Even so, the paperwork trail on reginfo.gov is often our best evidence. Data collected through sensors, satellites, and other non-paperwork means typically have no statutory notification requirements at all. For those, we depend on agency websites, news reports, and experts who flag changes when they see them.
Some agencies do announce their terminations, and it makes a difference. NESDIS keeps a log of decommissioned products. BEA maintains its list of discontinued or delayed statistics, BLS its pages of discontinued databases and series, NASS its newsroom notices. We wish this were the rule rather than the exception.
For everything else, we rely on partnerships and monitoring: organizations like the Data Rescue Project, the researchers and journalists who publish inventories of lost data, social media and the news, and tips from data users who notice that something they depend on has gone dark. We do this in both manual and automated ways, and our process will continue to improve over time.
And then we verify. Every reported termination gets a line by line review before it enters our count. We deduplicate, because fifty state pages of one product are one product. We separate data products from web pages that describe data. We trace each item to primary evidence (an agency notice, a press release, an ICR, a CRS report, or data actually being no longer available on a website), and we record status changes over time, because as we said above, termination is not necessarily a forever status.
That standard is also an invitation to you. If a dataset you rely on disappears, loses variables, or quietly stops updating, tell us. The more eyes on federal data, the smaller the chance that something essential vanishes without anyone noticing.
We are here to verify data terminations at removals@dataindex.us. To learn more about the Federal Data Terminations Tracker, register for our webinar on August 19, 2026.
How Methane Emissions Events Impact Air Quality and Put Human Health at Risk
In October 2015, an underground gas storage well at the Aliso Canyon Natural Gas Storage Facility in southern California failed. Over the course of four months, it released more than 100,000 metric tons of methane into the atmosphere, setting the record for the nation’s largest known methane emissions event.
In the days following, people in Porter Ranch began to experience bloody noses, headaches, nausea, rashes, and respiratory problems. The leak forced the closure of two schools for the remainder of the school year, and the evacuation of 10,000 people living in Porter Ranch, a community one mile downwind.
Aliso Canyon put methane super-emitters, events with emissions of over 100 kilograms per hour of methane, in the spotlight and raised a simple but critical question: Since methane, the primary component of natural gas, is non-toxic, why were people getting sick? What else was in the gas?
Some state regulatory agencies require oil and gas producers to report what is in their product to help the agency assess the environmental and health risks of oil and gas emissions. These documents are technically discoverable by the public, but often hard to access.
PSE Healthy Energy, and authors here, combed through tens of thousands of records and found that 99% of the natural gas sampled near the point of extraction across the country contains hazardous air pollutants— chemicals that the EPA has identified for their potential to cause serious health problems. For example, benzene, a known human carcinogen, was in more than 97% of the gas samples.
Even as federal methane regulations are being delayed or rolled back, states and local governments can still have many options for protecting human health from methane super-emitters. Regulatory options include strengthening air permitting disclosure requirements, expanding oil and gas infrastructure setback distances, broadening the types of infrastructure covered by setbacks, requiring leak detection and monitoring, strengthening emissions control requirements, investing in risk assessment and modeling of super-emitter events, and adopting short-term health benchmarks that reflect the acute risks posed by methane emissions.
Operators can implement many of these protective measures voluntarily to protect human health and limit their own legal and regulatory risks.
Challenge and Opportunity
The Health Risks of Methane Emissions
Methane is a powerful greenhouse gas that contributes to climate change. At this moment, more than two dozen satellites orbit Earth to detect and measure methane emissions, offering governments and citizens the opportunity to monitor methane emissions, and responsible parties to mitigate them. Recent advances in satellite monitoring have revealed that methane super-emitter events occur frequently across the United States and other countries around the world. Reducing the frequency and duration of these events will not only help slow climate change but will also benefit air quality and health.
PSE applied the gas composition data we gathered from regulatory documents to methane emissions events recorded via satellite, and modeled the air quality impacts and health risks of 1,490+ methane emissions events across the United States. More than 98% of the events exceeded state-based short-term health risk benchmarks, posing human health risks. These results are documented on our publicly available Methane Risk Map (MRM). The air quality impacts and human health risks of methane super-emitters are not systematically captured in air permitting and regulatory processes. The Methane Risk Map provides regulators, lawyers, and policymakers with science-backed data to establish health-protective policies such as gas infrastructure setback distances, stronger monitoring requirements, and leak prevention, detection, and mitigation measures.
Our research found that different types of equipment have notable differences in the concentration of hazardous air pollutant emissions. For example, gas samples from liquid storage tanks—which hold crude oil, gas condensate, or produced water—contained benzene concentrations 59 times higher than those at wells or gathering pipelines, which collect and transport natural gas. This is because the stored liquid contains high levels of hazardous air pollutants, and emissions of working losses, breathing losses, and flash gas, which is formed due to changes in temperature or pressure, such as when a valve opens, take with them these hazardous air pollutants that were present in the liquid. Understanding how concentrations of hazardous air pollutants change and the causes of methane super-emitter events at different points in the gas supply chain can help regulators set policies that protect human health.
Policy State of Play
The U.S. EPA’s Methane Super Emitter Program (SEP) was established in the New Source Performance Standards and Emissions Guidelines in 2024 as part of a federal effort to improve detection, reporting, and mitigation of large, unscheduled, and intermittent methane emissions events. The SEP partnered with certified third-party remote sensing companies to identify and report methane super-emitter events. Oil and gas operators were required to investigate any detected methane super-emitter events within five days and report findings within 15 days. However, the SEP is constrained by 40 C.F.R. part 60 subparts OOOO, OOOOa, OOOOb, and OOOOC regarding which events are considered super-emitter events. Specifically, the code defines a super-emitter event as an emission event at or near an oil and natural gas facility (e.g., an individual well site, centralized production facility, natural gas processing plant, or compressor station), which excludes emissions from other sources (e.g., landfills, refineries, and other methane sources). In July 2025, future implementation of the SEP was extended until January 22, 2027. This delayed implementation affects enforceability at the federal level. While there may be state-level reporting requirements, there is little federal accountability for operators who are repeat offenders or fail to adequately mitigate events. Additionally, PSE’s research shows that these events almost always release hazardous air pollutants that pose acute health risks, indicating that mitigating methane emissions could provide health benefits.
The federal government is also taking steps to limit the ability of states to regulate the exposure of communities to harmful emissions. In January 2026, the federal government filed a lawsuit against the State of California related to SB1137, a state law passed in 2022 that requires new oil wells to be set back at least 3,200 feet from homes, schools, community centers, parks and playgrounds, healthcare facilities, or public buildings.
Despite these challenges, there are opportunities at the state and local levels to use the science-backed data on PSE’s Methane Risk Map to inform health-protective energy policies closer to home.
Oil and gas companies seeking to build new infrastructure must evaluate the air quality impacts that their proposed operations might have on the local community before they are given permission to expand. This makes air permitting processes for proposed natural gas infrastructure a key opportunity to advance health-protective energy policies. Oil and gas companies control the start and end of any short-term planned maintenance events—even ones that meet or exceed super-emitter emissions rates—and in doing so can account for any related blowdown events in annual emissions inventory reporting. However, non-routine methane super-emitter events—the kind that PSE has shown can have far-reaching air quality impacts—are not systematically accounted for in these permitting processes. Because super-emitters have historically been difficult to predict and detect, states do not include these events in their reporting requirements for oil and gas companies, leaving the health risks of such events unaccounted for.
Regulatory requirements and enforcement vary widely across states. Some states have regulatory requirements that help mitigate health risks from gas infrastructure. For example, California established setback distances for wells or gas production facilities and established the California Satellite Methane Project (CalSMP). CalSMP expands on the initial federal SEP work by mapping more methane sources, including landfills. Additionally, some states have health protections that are focused on chronic exposure to hazardous air pollutants, but ignore acute or short-term risks. For example, Colorado’s Air Quality Control Commission held hearings in September 2025 to set new chronic health-based standards for five-priority air toxics, including benzene. PSE’s Methane Risk Map demonstrates that hazardous air pollutants emitted with methane during super-emitter events can produce ambient air concentrations high enough to pose health risks for people living near these emissions. Rules that only consider chronic exposures do not account for these risks.
Plan of Action
Human health risks of methane super-emitter events can be mitigated through evidence-based interventions. Even as federal air quality and human health protections are being constrained, both voluntary actions by operators and regulatory actions at the state, local, and tribal level can provide solutions.
Recommendation 1. Increase gas composition transparency through air permitting.
Generally, state regulators require companies to estimate emissions from routine gas leaks (also known as fugitive emissions) using gas composition data from the site, from a representative site, regional data, or a composition profile that is not site-specific. But these data are often unavailable to the public, may require a Freedom of Information Act (FOIA) request to access, or are in formats that are difficult to utilize. Furthermore, how regulators and operators define a representative site, or sample, is not always disclosed.
In the process of building its national gas composition database, PSE learned that disclosure requirements and operator practices vary widely from state to state. Some states, such as Texas, New Mexico, and North Dakota, require gas composition analyses from accredited independent labs as part of permit applications, and make the full applications available to the public via web portals. However, other states, like California, lack gas composition lab reports or detailed data in publicly available documents, even though, in some cases, companies are required to perform these analyses. In California, public access barriers to gas composition data often come in the form of documents with extensive redactions, long and burdensome California Public Records Act processes, and fees.
To help regulators, researchers, and impacted community members determine the health risks of hazardous air pollutants released during methane super-emitter events, states can require disclosure of gas composition data in publicly available databases. Since leaks occur throughout the supply chain, the best practice would be for every oil and gas facility operator to publicly disclose the composition of its gas and flash gas from multiple types of infrastructure, analyzed by accredited independent laboratories. Furthermore, when operators use representative samples (such as for facilities that are not yet built), best practice is to explain why the sample is representative of the facility. Currently, states take an à la carte approach to disclosure, with many only checking some of these boxes, creating an inadequate patchwork of data to support health risk assessment and policymaking.
Recommendation 2. Establish greater setback distances for oil and gas infrastructure.
Currently, only some states require setback distances for oil and gas infrastructure (Table 1). This setback means that wells or other equipment must be a minimum distance—usually on the order of hundreds to thousands of feet—from a home, school, or other sensitive facility. However, our findings show that typical setback distances are insufficient. First, setback regulations often overlook storage tanks, whose flash gas emissions can contain much higher benzene concentrations than well leaks. Furthermore, across the 1,490+ events currently on the Methane Risk Map, we found that modeled benzene concentrations typically exceeded short-term, state-based regulatory benchmarks on average out to two miles from the source of emissions. Differentiating setback requirements by equipment category and gas composition would improve safeguards against the most hazardous air pollutant emissions.
Recommendation 3. Prioritize flash gas emissions because of their health risks.
While some regulatory requirements for flash gas and storage tanks exist, PSE’s research shows that when flash gas is emitted from tanks, it can pose significant health risks due to the potential for high benzene concentrations in the released gas. Given these risks, regulators should prioritize flash gas sources for enhanced leak detection and repair requirements and compliance oversight, even though these types of events may not always have the highest methane emissions rates. When leaks occur, regulators can require operators to monitor, report, and address air quality impacts. These data would improve tracking of methane super-emitter events, inform maintenance practices, and help protect nearby communities and onsite workers.
Recommendation 4. Invest in modeling air quality impacts of gas leaks to inform risk management and emergency planning.
Regulatory approaches typically rely on routine emissions data, which do not capture the scale or reach of pollution from events such as tank failures, well leaks, or pipeline ruptures, which, as demonstrated by PSE’s Methane Risk Map, can generate short-term concentrations of health-harming pollutants that far exceed safe levels and extend far beyond facility boundaries. Without air quality modeling that incorporates gas composition and site-specific conditions, regulators and operators cannot assess the full scope of potential health risks or design effective mitigation and response strategies. Establishing transparent standardized air quality impact modeling, especially around residential areas and sensitive receptors like schools and childcare facilities, that accounts for methane super-emitter scenarios as part of permitting and risk management processes would provide data to support informed decisions by both operators and regulators.
Recommendation 5. Expand air quality monitoring in fenceline communities.
Equipping regulators with optical gas imaging (OGI) cameras and initiating targeted mobile or stationary air monitoring in fenceline communities and at facilities with repeated emissions events would enable independent detection of emission events and verification of mitigation attempts. These measures would focus oversight where risks are highest. Furthermore, site-specific information, including the cause and estimated duration of the leak, can improve health risk assessments and modeling. Indeed, PSE is already incorporating new, site-specific data provided by regulators into the Methane Risk Map to improve its modeling of tank emissions.
Recommendation 6. Adopt short-term health benchmarks for hazardous air pollutants.
Effective air quality modeling of methane super-emitters requires establishing short-term (hourly) health-based concentration thresholds for the hazardous air pollutants that are found in natural gas, such as benzene, toluene, ethylbenzene, xylenes, and hexane. While some states, such as California and Texas, establish such benchmarks, some do not. As shown by PSE’s Methane Risk Map, methane emissions events at natural gas facilities can lead to short-term spikes in hazardous air pollutant concentrations. Adopting and requiring short-term health benchmarks, combined with air quality modeling of non-routine event scenarios, to inform permitting, monitoring, and emergency response decisions ensures that health protections reflect real-world risks and prioritize fenceline communities.
Conclusion
Our research reveals that methane super-emitter events not only impact climate, but also impact air quality and pose human health risks. The Methane Risk Map shows that hazardous air pollutants are routinely released alongside methane during upstream natural gas leaks, often at concentrations that degrade air quality and pose acute health risks to nearby communities. Risks from these types of events remain largely invisible under current regulatory frameworks.
The Methane Risk Map can help both operators and regulators characterize health risks from methane emission events and target voluntary measures and regulatory oversight where they are most needed. By strengthening control requirements, data coverage, availability and transparency, updating and effectively implementing health benchmarks, improving siting, and expanding monitoring and risk modeling, policymakers can close regulatory gaps that leave people’s health vulnerable, and develop energy policy that protects human health. Through voluntary actions and transparent cooperation with regulators, operators can proactively mitigate health risks and manage the potential risks posed by independent detection and monitoring technologies.
PSE collected and analyzed gas composition samples from federal and state agencies across major oil- and gas-producing basins. Using this information, we assembled the most comprehensive speciated database of gas composition covering a substantial share of reported upstream methane emissions nationwide. This database enables PSE scientists to estimate site-specific hazardous air pollutant concentrations in surrounding communities for each methane emissions event on the MRM.
The MRM only includes methane emissions events for which we have sufficient information to estimate source-specific gas composition; however, this only accounts for a small fraction of all the methane emissions events in the United States. The MRM is regularly being updated with new events as methane emissions and gas composition data become available. In the future, we plan to model methane emissions events from other parts of the gas supply chain, as well as from other U.S. states and countries.
The emissions rate for each event is derived from available satellite and aircraft measurements. These observations typically only provide emissions rates at a snapshot in time. Therefore, to be conservative, we compare one-hour average concentrations to acute or short-term health benchmarks.
If the MRM shows exceedances of health or safety benchmarks near your location, it suggests that there was a human health risk during those events. Prolonged or repeated exceedances—especially from the same site—can point to a concern. If you’re seeing risk indicators, it may be worth reaching out to local environmental groups or contacting your local health or environmental agencies for more information.
Ongoing DOE Staff Shortages Limit Trump Energy Priorities
America’s energy innovation ambitions are only as strong as the federal workforce responsible for delivering them. Yet, over the past year, the U.S. Department of Energy (DOE) has suffered massive losses of staff capacity and skills, largely due to the Trump administration’s Deferred Resignation Program (DRP), which slashed headcounts across the federal government. Occupations with skills essential to making and managing awards were among the hardest hit. As a result, many energy innovation programs ground to a halt in 2025. Although DOE activity began to rebound in 2026 without an increased and targeted hiring strategy, the department risks falling short just as the demand for energy innovation is accelerating.
The EFI Foundation (EFIF) and the Federation of American Scientists (FAS) analyzed changes in DOE’s workforce and spending trends using EFIF’s updated Energy Innovation Project database. The following analysis identifies the workforce needed to restore DOE’s capacity to deliver on its energy innovation mission.
In 2025, DOE Lost Staff Critical to Program Execution
Under the Trump administration, DOE has lost over 2700 federal staff, a majority of whom departed through the agencywide DRP in 2025. Nearly any federal employee could opt-in to the DRP, meaning DOE had very little control over what expertise left the department.
Of all staff that have departed DOE since February 2025, EFIF estimates over one third were in occupations with essential skills for making and managing award programs. Individuals in these roles are hired for experience managing budgets, personnel, and resources critical to executing the goals of specific programs. The occupations most impacted include management and program analysis (40% reduction), grants management (47% reduction), and loan specialists (42% reduction). Nearly every occupation type with skills essential to award management experienced a higher-than-average proportional decrease in headcount at DOE over the past year and a half.
Notes: Data are from the U.S. Office of Personnel Management (OPM) and represent the percent difference in total headcount by occupation type across all DOE offices except the National Nuclear Security Administration and Federal Energy Regulatory Commission. Occupations are listed as occupational series, which are job categories defined by OPM as the description that an employee’s position falls under based on the duties and required qualifications of the position. EFIF identified 18 occupational series critical to managing DOE’s awards which saw significant changes across 2025 and 2026. Some categories are combined here for simplicity. See our approach section for details.
Without rebuilding the loss of essential skill sets, DOE will struggle to translate congressionally appropriated funding into real-word energy innovation outcomes.
The massive reduction in federal staff has a compounding impact on DOE capacity due to the loss of institutional knowledge. In total, the thousands of federal workers who have departed DOE under the second Trump administration took with them a combined 29,792 years of service. Nearly 90% of departing staff were career employees, over half of whom had more than five years of service, with tenure likely spanning multiple administrations. In contrast, over one third of hires over the same period were political staff, who serve limited terms and are often new to the federal workforce.
Notes: Data are from U.S. Office of Personnel Management and represent hires and departures across all DOE offices except the National Nuclear Security Administration and Federal Energy Regulatory Commission. Appointment types are defined by OPM as the type of appointment an employee is serving under in terms of permanence and competitiveness. EFIF separated OPM appointment types into career and political roles based on review of OPM appointment type descriptions. See the approach section for a detailed methodology.
Federal career staff are critical to ensuring agency function and smooth transition across presidential administrations. Career staff carry an understanding of department-specific administrative processes, lessons learned from prior programs, and relationships with private-sector partners—knowledge critical to executing and maintaining awards. While political staff often bring much-needed new perspectives and ideas into the fold, they are, by definition, more focused on overall policy direction than day-to-day functions. Increased political staff capacity cannot replace the role of seasoned career staff at the department. Energy innovation at DOE, which often operates on multi-administration timelines, depends on the health of its entire workforce.
DOE is Ramping Up Activity, But Execution Requires the Right Staff Levels and Expertise
In the first quarter (Q1) of 2026, DOE outpaced all of 2025 in funding announcements and project selections across its science and energy innovation offices. The department announced the same number of awards (14) in Q1 2026 as it did in all of 2025, but the dollar value of the 2026 awards was four times greater. In addition to an increase in new program funding, on April 15, 2026, Energy Secretary Chris Wright announced the unpausing of nearly 2,000 awards.
However, new spending activities in the second quarter of 2026 have dropped drastically. Is the agency experiencing the implications of an ongoing staffing shortage?
Notes: Data are from Grants.gov, DOE’s press releases, and USAspending.gov from February 1, 2025, to June 30, 2026 and were last updated on July 6, 2026. Solicitations without funding, such as requests for information (RFIs), are excluded. Data reflect DOE science and energy innovation offices, excluding the Office of Energy Dominance Financing (EDF), which was formerly known as the Loan Programs Office. See the approach section for a full list of offices tracked.
In addition to issuing new funding opportunities, DOE science and energy innovation offices actively manage over 5,000 awards worth $41.1 billion, according to EFIF’s Energy Innovation Project database. Yet, these offices were the hardest hit by the DRP. According to data obtained by FAS, an estimated 1,190 of DOE’s DRP participants were from its science and energy innovation offices. As of October 2025, these offices had lost an estimated 43% of their federal employees.
To rebuild a workforce that can drive forward American energy innovation, DOE must not only make up for the skills and expertise it lost in 2025, but also identify the workforce capacity and capabilities needed to achieve its priorities.
Although the department can shift staff capacity and programs around to increase headcount in priority offices, this does not replace the program-level expertise lost with the departure of seasoned staff. For example, the Office of Clean Energy Demonstrations (OCED) lost 85% of its staff by June 2025, leaving an estimated 40 employees to manage about 100 projects and a $27 billion portfolio. While the department can move OCED programs and limited remaining staff to other offices with higher headcounts, this does not replace the awardee relationships and commercial-scale project expertise former staff built while managing OCED’s portfolio.
Notes: “Total budgetary resources” include DOE’s full obligational authority in a given fiscal year, including appropriations, unobligated carryover balances, and authority from offsetting collections. For the purposes of this analysis, off-budget financing accounts are excluded. Data are from USAspending.gov and reflect DOE science and energy innovation offices, excluding EDF. “FTE” is used throughout to mean full-time equivalents for federal employees, as measured by OMB. FTE data are from OMB’s Presidential Budget Request Technical Supplement for fiscal year 2027.
In addition to achieving the capacity needed to manage existing programs, the department should also acquire the workforce skills required to realize its new ambitions. For example, DOE’s new funding opportunities include increases in nontraditional other transaction agreements (OTAs). To execute OTAs, DOE needs expertise in both federal cost accounting—a narrow specialization—and the Generally Accepted Accounting Principles used by the private sector. The demand for new skills extends to DOE’s contracting capabilities as well. Due to their lack of a standard structure, OTAs require highly specialized contracts. As stated by a Bipartisan Infrastructure Law (BIL) program awardee, there exists “a disconnect between developers who know how to deliver projects and bureaucrats trying to write a contract.” DOE must carefully evaluate whether its remaining contracting workforce can accommodate burgeoning demand for OTAs and commercial-style agreements.
Without the right expertise, DOE will be unable to spend federal dollars responsibly, manage existing awards, and strategically impact the energy sector. Every award requires staff to evaluate applications, negotiate terms, obligate funds, and monitor performance over the life of a project. The pace of new and restarted awards appears to be building faster than DOE’s current workforce can absorb.
DOE Increased Hiring But Still Saw a Net Loss of Employees in 2026
In 2025, the Trump administration assumed that it could achieve its priorities with substantially fewer career staff by relying more on political staff. Yet, a year after the DRP, it is increasingly clear that a new workforce strategy is needed. Mass departures under the DRP have crippled federal agencies. The Office of Personnel Management (OPM) has recognized the negative impacts of staff losses and is looking to reverse cuts to critical positions. As stated by OPM director Scott Kupor, “we probably have some skills that we now need to hire back, quite frankly.”
In April 2026, OPM launched a governmentwide hiring initiative focused on project management and data science, aiming to hire 250 professionals across the federal workforce. While a start, this is far from sufficient to fill existing gaps. DOE alone has experienced a net decrease of 140 program managers under the Trump administration. Larger hiring efforts will be needed to recover workforce capacity.
Hiring has increased slightly at DOE in 2026. From January to May 2026, DOE hired 193 federal employees, more than in nearly all of 2025 (129, excluding January). Despite the uptick, however, DOE continues to experience a net loss of federal staff in 2026.
Notes: Data are from U.S. Office of Personnel Management and represent hirings and departures across all DOE offices except the National Nuclear Security Administration and Federal Energy Regulatory Commission. Average monthly separations and hires in 2025 exclude data from January 2025.
The 2027 president’s budget request (PBR) calls for increasing staff in select science and energy innovation offices, specifically the Office of Nuclear Energy (NE), Office of Hydrocarbons and Geothermal Energy (HGEO), and the Office of Energy Dominance Financing (EDF). However, in several science and energy innovation offices, requested staff size in the PBR still fall below pre-2025 levels. It is unclear if increased staff capacity in favored offices will come from new hires or a transfer of existing staff from other offices. Given current—and growing—workloads among science and energy innovation offices, it is unlikely that many offices have excess capacity to redistribute.
Notes: “FTE” is used throughout to mean full-time equivalents for federal employees, as measured by OMB. Data are from OMB’s Presidential Budget Request Technical Supplement for FY 2027. “Office of Energy Dominance Financing” represents all Loan Programs Office accounts; “Hydrocarbons and Geothermal Energy Office (HEGO)” represents both Fossil Energy and HGEO accounts.
It is unclear if and when DOE intends to hire new staff in key offices and replace the loss of essential award management skills. The pipeline to hire federal staff is long and narrow. As of July 22, 2026, there are 25 publicly open positions at DOE listed on USAJobs.gov but only two were in science and energy innovation offices. To achieve the administration’s energy innovation goals, DOE needs to up its hiring.
In the past, DOE has filled gaps in federal workforce capacity through contracting. However, combined staff pay and contractor spending in fiscal year 2026 is currently at pre-BIL levels, when the department had less than one third of its current science and energy innovation budget.
Notes: Data are from USAspending.gov, accessed July 2026. Obligations to contractors are calculated using the “advisory and assistance services” object class. Staff pay obligations are calculated by summing four object classes: civilian personnel benefits, full-time permanent, other personnel compensation, and other than full-time permanent.
Moreover, increasing contractor capacity does not fully replace the function of federal employees. Contractors are not allowed to fulfill some critical award-making functions reserved for federal employees including awarding grants, obligating funds, and executing loan commitments. These occupations are fundamental to project deployment, and increasing contractors is a temporary, far-from-optimal solution for a strained federal workforce.
Staffing DOE to Execute Its Mission
Rebuilding DOE’s workforce goes beyond simply replacing the employees the agency has lost. DOE first needs a holistic view of its goals as an agency to identify the workforce capacity needed to carry out that mission.
DOE has expanded in recent years into commercial-scale project deployment and has undergone changes to its priorities and internal operations during the second Trump administration. This means that the department’s workforce needs are likely different today than they have been in the past. Because of this, DOE must look to align hiring with current and expected future workload demands and organizational needs rather than simply restoring previous staffing levels one-to-one.
DOE faces an uphill battle. Executing ambitious energy priorities requires an equally ambitious workforce strategy. Yet, hiring trends in 2026 thus far are still insufficient to maintain—let alone build—workforce capacity. While targeted OPM hiring efforts across federal agencies for project management skill sets are a start, they must be expanded to meaningfully restore the loss of key workforce capabilities. Further, the Trump administration’s tumultuous start to its relationship with the federal workforce will likely have long-term effects on its ability to attract new or returning talent to federal agencies. Nevertheless, DOE needs to rebuild a workforce that can deliver on its energy priorities because any energy strategy is only as strong as the workforce responsible for implementing it.
Our Approach
All data used in report figures, with the exception of project selections, are available in EFIF’s Energy Innovation Project public database. For a full methodology on data gathering and cleaning, as well as offices included in EFIF’s science and energy innovation offices, please refer to the website’s methodology page.
DOE-wide staffing data were downloaded from OPM on July 6, 2026. Employment (headcount) figures are benchmarked to the end of the fiscal year on September 30th of each year with the exception of fiscal year 2026 which is benchmarked to May 30th 2026 due to data availability. Hires and separations are cumulative totals across all months within each fiscal year. The data set includes all DOE employees, limited to full-time employees by work schedule, employees whose sub-agency element is Department of Energy, and excludes employees under National Nuclear Security Administration pay plans. Office-specific staffing data are from the 2027 president’s budget request and DOE data obtained by FAS through the Freedom of Information Act which included staffing numbers for OCED; Office of Manufacturing and Energy Supply Chains; Grid Deployment Office; EDF (formerly known as LPO); Office of Federal Energy Management Programs; Office of Energy Efficiency and Renewable Energy; Advanced Research Projects Agency-Energy (ARPA-E); Office of Technology Commercialization; Office of Indian Energy Policy and Programs; Office of Cybersecurity, Energy Security, and Energy Responsibility; Office of Electricity; Office of Science; Office of Fossil Energy and Carbon Management; and Office of Nuclear Energy. Funding and awards data are from USAspending.gov, grants.gov, and EFIF review of DOE press releases.
For simplicity, EFIF combined multiple OPM-defined occupational series into single categories. “Management and program analysis/assistance” represents series 0344 and 0343, “miscellaneous administration and program” represents series 0301, “contracting and procurement” represents series 1102 and 1106, “program management” represents series 0340, “financial support and accounting” represents series 0501, 0505, and 0510, “grants management” represents series 1109, “loan specialists” represents series 1165, and “legal and patents” represents series 0901, 0905, 0950, 0963, 0986, 1221, and 1222. “Technical experts” were identified by their classification as Science, Technology, Engineering, and Mathematics (STEM) occupations, omitting those occupational series related to technology support or operations.
To distinguish career from political employees, EFIF combined OPM appointment types in the following way. “Career” includes career (competitive service permanent), career (senior executive service permanent), career-conditional (competitive service permanent), executive (excepted service permanent), nonpermanent (competitive service nonpermanent), and Schedule B (excepted service permanent). “Political” includes executive (excepted service nonpermanent), limited term (senior executive service nonpermanent), noncareer (senior executive service permanent),, and Schedule C (excepted service nonpermanent). “Other” includes other (excepted service nonpermanent), other (excepted service permanent), Schedule D (excepted service nonpermanent), Schedule A (excepted service nonpermanent), Schedule A (excepted service permanent, Schedule D (excepted service permanent), and invalid.
International Collaboration to Strengthen Domestic Critical Minerals Efforts
Despite most people not being able to name a single one, critical minerals are fundamental to the advanced technologies that underpin U.S. economic competitiveness and national security. Think microchips, electric vehicles, and life-saving medical devices. However, as well-documented by FAS, the United States remains dependent on exports from a limited number of countries that mine and process them. This is despite efforts to onshore and diversify supply chains through major initiatives like the Bipartisan Infrastructure Law, Inflation Reduction Act, and targeted trade policies. The consequences of failing to meet this domestic demand, and meet it promptly, will be passed onto American consumers: inflationary risks, supply chain disruptions, and the possibility of the U.S. being “exposed to the leverage of resource-rich countries or market incumbents abroad.” That’s flowery speak for the U.S. being held hostage by our reliance on rocks.
We’re not the only country dealing with this. And given the scale of this challenge, no single nation will be able to address it solely through domestic policy. Rather, it will require significant collaboration and cooperation amongst allied nations through vehicles of international science and technology engagement.
Since late last year, we have seen the Trump administration’s willingness to address export promotion, supply chain security, and technology leadership for other national priorities through a flurry of bilateral technology prosperity deals:
- In October 2025, the U.S. signed a technology prosperity deal with Japan focused on AI, quantum, 6G telecommunications, fusion energy, and space exploration.
- That same month, the U.S. formalized a technology prosperity deal with Korea, focused on AI, 65 telecommunications, quantum, pharmaceutical supply chains, space exploration, and basic research and STEM exchange.
- In September 2025, the U.S. announced a technology prosperity deal with the United Kingdom focused on AI, civil nuclear energy, quantum, 6G telecommunications, and research security; however, just a short time later, in December, implementation was suspended over frustration over a lack of progress, but restarted back in February 2026 with a more narrow focus on nuclear energy.
- Most recently, in May 2026, the U.S. signed a technology prosperity deal with Sweden focused on AI, 5G/6G connectivity, biomedical research, advanced manufacturing, nuclear energy, space exploration, quantum, and research security.
These technology prosperity deals reflect the administration’s approach to science diplomacy as focused on economic and security returns for the United States, first and foremost. But more importantly, while signaling intent to cooperate on shared interests, all these agreements explicitly state that they do not create legally binding obligations, and that nothing commits the participants to the expenditure of funds, thus rendering them effectively toothless. Rather than attempting to tackle critical minerals through bilateral technology prosperity deals that commit nothing and bind no one, the U.S. needs a dedicated, properly resourced device for international science and technology cooperation with real authority, funding, and accountability.
The Strategic Technology and Resilient Alliances Act of 2026 (STRATA Act) would do exactly that. This new bipartisan bill by Representatives Young Kim (R-CA-40) and Ami Bera (D-CA-06) would establish a Critical Minerals Innovation Partnership program within the Department of State, to be led by a Senate-confirmed director reporting to the Under Secretary of Economic Affairs. Unlike the toothless technology prosperity deals, the STRATA Act would create real mechanisms for cooperation: binding partnership agreements with specific objectives, quantitative benchmarks, multiyear funding plans, and intellectual property protections. The Director would be empowered to issue joint solicitations with partner countries, coordinate demand aggregation across allied governments and the private sector, and establish International Centers of Excellence for innovative extraction and processing technologies in partner nations. A public-facing digital platform would connect U.S. startups, universities, and research institutions to funding opportunities and collaborative projects. Crucially, the bill also amends the Foreign Assistance Act to explicitly authorize financing for critical minerals S&T cooperation, giving the program the legal and budgetary backbone that bilateral deals have lacked. The program would sunset after ten years, providing a built-in accountability horizon while allowing existing agreements to run their course. There are historical and contemporary analogous efforts that can serve as models of success.
The Soviet Nuclear Threat Reduction Act of 1991 established and authorized the Cooperative Threat Reduction program to facilitate “cooperation between the United States, the Soviet Union, its republics, and any successor entities to (1) destroy nuclear weapons, chemical weapons and other weapons, (2) transport, store, disable, and safeguard weapons in connection with their destruction, and (3) establish verifiable safeguards against the proliferation of such weapons.” 25 years into the program, the Defense Threat Reduction Agency (DTRA) documented the program’s significant accomplishments, including the destruction of 2,532 missiles, the decommissioning of 1,300 delivery systems, the destruction of over 4,700 tons of chemical weapons agents, and the employment of 22,000 former WMD scientists.
More recently, the CHIPS and Science Act of 2022 authorized the International Technology Security and Innovation Fund (ITSI). It provided the Department of State $100 million per year over five years “to promote the development and adoption of secure and trustworthy telecommunications networks and ensure semiconductor supply chain security and diversification.” ISTI awards have included a $13.8 million cooperative agreement in 2024 with Arizona State University to expand semiconductor workforce capacity and strengthen supply chain infrastructure across partner countries in the Americas and Indo-Pacific. That same year, through the ITSI Fund, the State Department collaborated with the Inter-American Development Bank to launch the CHIPS ISTI Western Hemisphere Semiconductor Initiative, to enhance “semiconductor assembly, testing, and packaging (ATP) capabilities in key partner countries, beginning with Mexico, Panama, and Costa Rica.” Additionally, the fund enabled a partnership between the State Department and the India Semiconductor Mission, Ministry of Electronics and IT, Government of India, focused on assessing India’s semiconductor ecosystem, including its regulatory framework, workforce, and infrastructure needs, to inform potential future joint initiatives.
When Congress provides dedicated authority and real funding, the State Department can serve as an effective vehicle for technology-focused international partnerships. The same model should be applied to critical minerals. As Congress considers broader packages to advance critical minerals production and supply chain resilience, science diplomacy vehicles must be part of that conversation. It cannot be an afterthought; rather, it must be an intentional pillar of any strategy.
The Science of Moonshots: Using Evidence to Design Transformative Initiatives
“We choose to go to the moon. We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard, because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one which we intend to win, and the others, too.” – President John F. Kennedy, announcing the original moonshot in 1962
This piece developed in concert with the Moonshots and Metascience event made possible by the Alfred P. Sloan Foundation.
In 1945, the Federation of American Scientists was founded by a group of physicists that helped to develop the atomic bomb under the Manhattan Project. FAS was born at a critical juncture – the creation of the atomic bomb was the first time the world had seen a coordinated, interdisciplinary effort to harness science for an ambitious goal. FAS’s founders had just witnessed the power and potential impact of transformative science initiatives. Deeply concerned about the potential for harnessing this impact for malice, they came together to ensure that science and technology would promote societal benefit rather than harm.
Eighty years later, we find ourselves at yet another pivotal moment for the scientific enterprise. Americans are asking critical questions about whether publicly funded research is delivering for them, their families, and their communities. To respond to that challenge, we can learn from FAS’s roots by asking: what does it take to design initiatives centered on big, challenging goals that lead us to better science and a better world? After all, whenever the government spearheads such ambitious initiatives, it can lead to tremendous public benefit in terms of completely new technologies and capabilities that serve critical public needs.
Critical to launching such initiatives is understanding the components that make them so successful. To this end, FAS organized a recent event “Moonshots for Metascience and Metascience for Moonshots” to ask hard questions about what has made prior ambitious efforts work, and why some of those efforts failed – building toward a conversation about the types of moonshot-style initiatives needed today. You can get an inside look at one of those conversations, featuring renowned scientist Lee Hood, here.
From those conversations, we’ve identified the key ingredients of successful moonshots that meet the moment, and developed recommendations about what future efforts can and should look like.
Defining the term “moonshot”
The word moonshot was described by a number of attendees as tired, overused, and overhyped. It has evolved into a marketing term for ambitious science, rather than a specific term used to describe a program with discrete structure and objective.
Other, similar terms might get closer to more generalized ambitious science: entrepreneur and founder of the X-Prize Peter Diamandis uses the term “moonshot mindset” for what he calls “10x thinking”; in other words, the type of approach that treats every problem as solvable and aims for an improvement of 1000% over the current state. Economist Mariana Mazzucato centers her work around “mission-oriented innovation”, an approach that leverages frontier knowledge to help solve big challenges. And philanthropy increasingly uses the term “big bet” to describe these types of projects.
Ultimately, whatever you want to call it, our goal is to better understand and promulgate the essential elements of both moonshots and their close cousins: initiatives that set audacious and transformative goals, how those initiatives chart a clear path toward achieving those goals, and the roadmap for stretching beyond the limits and mechanisms of the traditional scientific R&D enterprise.
It’s not enough to know what your “moon” is – you need to know the trajectory of the “shot”, too.
Successful moonshots involve both a clear what and a clear how. If you think about prior moonshot efforts, there are probably a select few that jump to mind: the Apollo Program, the Manhattan Project, and the Human Genome Project are the most notable efforts. What’s something these all have in common?: A “moon”, a goal that anyone can clearly define. This is a critical element of successful moonshots.
But other similar efforts have been tried before to less avail. As an example, consider the Human Brain Project. This project had a clear goal: to simulate the entire human brain on a computer. It concluded in 2023 after 10 years and €1 billion spent. The effort did lead to significant advances in neuroscience, but it failed to accomplish its stated goal. From the jump, scientists were highly critical of this initiative because they did not believe the requisite technology existed, or was even close to existing. In other words, the how was unclear.
This is common for many such initiatives. For example, when the Human Genome Project was launched, we did not have the technology capable of meeting the mission. The project was highly criticized by many in the research community because the traditional method of DNA sequencing could not possibly be scaled to meet the goal. They needed a new way to sequence DNA faster, with longer reads, and for less money. Therefore, the first goal was to drive investment in the development of technology that could answer the how question.
A successful moonshot is one where it’s clear that we are on the precipice of achieving a once-unimaginable goal, if only we make a significant investment toward that end. Think of the original moonshot: when JFK announced this initiative, we had seen a man orbit the Earth. And we knew that putting a man on the moon was challenging, but far from impossible – we just needed to devote the appropriate time, resources, and organization around that goal.
Organizational structure and governance matter
The history of major scientific initiatives suggests that organizational structure and governance are not merely administrative details. They are often the deciding factor between transformational success and expensive stagnation.
The Human Genome Project illustrates this directly. Years of planning preceded the project’s formal launch in 1990. The project was organized using a joint agency structure, partnering the National Institutes of Health (NIH) and the Department of Energy (DOE), with the two agencies signing a memorandum of understanding to coordinate their efforts. NIH instituted a university grant-based program funding multiple genome centers for five-year periods, a deliberate choice to distribute the work across institutions while maintaining central coordination. In this example and others, structure was not the container for science – it shaped what science is possible in the first place and funneled efforts towards a central goal.
Governance matters for a related but distinct reason: it determines who has the authority to make bets and whether an initiative can maintain its ambition as it encounters institutional and technological barriers. The Human Genome Project’s governance was tested early on. Its inaugural director resigned in 1992. Its new director took over nearly a year later, bringing a revised five-year plan. The ability to absorb that leadership transition without losing momentum was itself a governance achievement; it depended on the project having clear enough goals and distributed enough institutional ownership that no single departure could derail it.
Evidence from other successful models like DARPA similarly suggests that concentrated decision-making authority combined with clear accountability to a defined endpoint is strongly associated with the risk tolerance that transformational work demands. Governance structures send signals that shape who joins and how they behave: a structure that punishes failure will produce risk-averse science regardless of how bold the original mandate was. In this sense the organizational choices made at founding are not just operational. They are a statement of values that propagate through every subsequent decision made to drive the initiative forward.
Today’s moonshots can, and should be, informed by metascience
Sometimes it’s more obvious that we are ready for a moonshot. Again, thinking back to the Apollo program: if we can send someone into orbit around the Earth, we can almost certainly find a way to get to the moon. But in many scientific fields, it’s less clear that we are on the verge of a breakthrough. However, new technological advances and a wealth of available data are starting to make it possible to identify signals that we might be on that verge. In a pre-print published in 2025, Davis et al describe an approach using the PubMed database to detect when breakthroughs are on the horizon in biomedical research; Kristine Willis describes how we can apply that framework to science funding in general. This is an example of how we can directly leverage evidence to inform future ambitious efforts, and will be especially useful for moonshots where understanding where science will likely be in 4-5 years can help direct ambitious investments.
The field of academic study that Davis, Willis, and many others contribute to is called metascience, also referred to as the science of science or research on research, which uses the scientific method to study science itself; it seeks to build an understanding of how research is funded, conducted, and published.
Metascience can help us understand not just what topics might be ripe for a moonshot, but also how to best run those initiatives. There is a limited amount of existing evidence on moonshots specifically, but metascientists can offer useful insights. For example, Dashun Wang’s work on the different types of innovations produced by different size teams or his finding that interdisciplinary teams produce research that is more widely-cited can inform the structure of projects funded under a moonshot effort. Or we can look to Pierre Azoulay’s work on funding people vs. projects, or where grant funding is most useful to spur scientific progress to inform when and how government takes the lead in developing moonshots. These insights, all generated in the past 15 years, help us build new moonshot initiatives that are more efficient, allowing us to make the case that we are better positioned to launch ambitious moonshots because we have a strong evidence base for what works.
Metascience expertise should also be embedded within moonshot efforts to help us understand if we’re making the progress that we hope for and to set achievable timelines. Learning from similar efforts in the UK, there have been recent calls to embed metascience teams within U.S. federal agencies to evaluate the effectiveness of their funding approaches. A metascience team, funded as part of a moonshot initiative, can build evidence along the way so we can learn from any successes and failures and adjust strategy to improve output and efficiency.
In short, metascience can help us understand when we’re ready for a moonshot, how to structure it, and evaluate its effectiveness as it proceeds.
What are the next moonshots?
We have the ingredients for successful moonshots and other transformative scientific efforts, a mandate from the public to engage in scientific discovery that will have a positive impact on their lives, and ambitious national priorities that require significant investment in R&D.
At our event, participants imagined a set of possible moonshot efforts that might meet the moment:
These are not just thought experiments. Each proposal represents a domain where we have clear signals that we are on the verge of something transformational, where the public interest is clear, and where the missing ingredient is a coordinated, well-designed effort to close the gap between what we know and what we could achieve. In other words, each has a clear “moon”. The work ahead is designing the “shot.”
Eighty years ago, FAS’s founders emerged from the Manhattan Project having witnessed what science could accomplish when organized around a bold goal, and committed themselves to ensuring that power was directed toward human benefit. Today, we face a different but parallel challenge: not just identifying the right goals for science, but building the institutional capacity to pursue them effectively. The Human Phenome Project, which our Q&A with Lee Hood describes in detail, is one example of what that could look like – an initiative designed from the start with the organizational structure, governance model, and embedded metascience capacity that transformative efforts require.
FAS was founded on the conviction that scientists have a responsibility not just to do science, but to think carefully about how science is organized and to whose benefit. That conviction is as relevant now as it was in 1945. The moonshots we need are within reach. The question is whether we are willing to be as rigorous about how we pursue them as we are about the science itself.
Algae as Analogy in an Age of Disruption and Transition Complexity
This piece developed in concert with the Moonshots and Metascience event made possible by the Alfred P. Sloan Foundation.
Algae is making plenty of headlines lately with the recent drama at the Lincoln Memorial reflecting pool. But this is far from the first time that algae has seized the spotlight. When I was a graduate student at the University of Washington (UW) in the late 1990s, I biked daily along the Burke-Gilman trail from my house in Seattle’s Wallingford neighborhood to campus. As a recent transplant from New York City, the blue waters of Lake Washington on sunny days never failed to amaze me. But the stories of Lake Washington’s not-so-distant “Lake Stinko” past were the stuff of legend. A few decades prior Lake Washington had a giant pollution problem — cloudy, dirty, and smelly, fouled with 20 million gallons of wastewater pumped into it each day, which prompted algae outbreaks and then algae die-offs.
At UW I was inspired by professors Tommy and Yvette Edmondson, then emeriti faculty and local heroes whose work on ecological tipping point theory and Lake Washington offered living proof that systems could rapidly undergo dramatic transitions between different ecological states. And sometimes, they could snap back to a pre-transition state. Their research led directly to the creation of Metro, the Seattle agency that diverted sewage effluent away from the lake, enabling Lake Washington’s dramatic recovery.
This history is top of mind because complex systems – from ecological to political to socio-technical – rarely change the way we expect. They can absorb pressure for years, even decades, then bifurcate, reorganizing rapidly around a new equilibrium that can look nothing like the old one. What happens after the transition has a lot to do with how resilient the system is to begin with.
It is perhaps unsurprising then, that this current moment of disruption and transition in the U.S. scientific enterprise is making me think deeply again about tipping points and state changes, beginning with the question of how resilient was this system to begin with? Even prior to this current moment of acute disruption, the U.S. scientific enterprise has been showing early warning signs of instability for years — research becoming measurably less disruptive, grant success rates falling, administrative burden consuming nearly half of researcher time, alternative funding models proliferating at the margins (e.g. FROs, Fast Grants, etc), because the core systems were not sufficiently responsive.
Many of our research institutions were built for a scientific, technological, and geopolitical environment that no longer exists. That environment was the one Vannevar Bush designed for in 1945: a single Cold War rival, science conducted by individual investigators in small academic labs, and a public trust in science near its historical peak. Eighty years later, the competition spans AI, biotech, and advanced manufacturing at once; research is large-scale and computational; and that trust is being actively renegotiated. The architecture hasn’t kept pace.They were optimized for incentives and stakeholders that have changed and have lacked the flexibility to adapt.
The acute perturbation over the past year, massive cuts to federal grants and large-scale erosion of the federal workforce, has rapidly exacerbated an instability that was already there, and is forcing the system into a rapid and uncertain transition. Though Congress ultimately rejected the administration’s deepest proposed cuts to NIH and NSF, the agencies tasked with running the system never recovered the staff needed to operate it: as of April 2026, NSF had awarded only about a fifth of its typical grant volume for the fiscal year, and NIH about half.
Metascience Separates Progress from Noise
Meanwhile, the growth of artificial intelligence (AI) applied to science is layering rapid, untested acceleration on top of institutional turbulence, with promises of massive scientific breakthroughs with fewer guardrails. For example, in October 2025, OpenAI CEO Sam Altman announced the company is internally tracking toward an intern-level AI research assistant by September 2026 and a fully autonomous “legitimate AI researcher” by 2028, on the same day OpenAI completed its shift away from nonprofit governance constraints. The institutional capacity to evaluate, govern, and correct what such systems produce is not advancing on anything close to that timeline. Acceleration without the institutional capacity to evaluate what is working, course-correct when necessary, and build knowledge that outlasts any single initiative will create noise, not progress. What we do next needs to be grounded in evidence. This is where metascience (the science of science) is poised to play an outsized role – by actively treating this moment of disruption as a large-scale systems-level experiment and learning as we go. This will only work though, if we can grow the field of metascience and position it for success.
In our recent FAS workshop, “Metascience for Moonshots and Moonshots for Metascience” we grappled with two distinct questions. The first proved more tractable than the second. It proved relatively straightforward to contend with the idea that metascientific approaches should be applied to help drive and evaluate progress toward ambitious goals with clear finish lines; for example, a climate-related goal to fly a zero-emission passenger plane from Washington to London within 10 years and health-related goal to build a national initiative to increase healthspan by 5% over 10 years – these are goals that can be formally constructed by working backward from a defined outcome. Metascience expertise embedded within moonshot efforts can help us assess the progress and to set achievable timelines, as well as build evidence along the way to enable learning and adjust strategy to improve output and efficiency.
But unlike curing cancer, sequencing the human genome, or landing on the moon, defining a moonshot for metascience resists a clearly defined finish line. Through the course of the workshop, the group came to a consensus that the specific moonshot framing did not hold well. There was no argument that the field of metascience needs to grow and should play a role in meeting this moment of transition in the scientific enterprise, but it proved difficult to define a “moonshot for metascience.” Rather, metascience lends itself to a different kind of ambitious mission.
Accelerator versus Transformer Missions
Economist Mariana Mazucatto distinguishes between “accelerator missions” that are hard, expensive, and coordinated, but can ultimately be defined by a single verifiable achievement. These are traditional moonshots. In contrast, she outlines a different class of “transformer missions,” which lack a specific finish line, but succeed by changing what the system itself is capable of doing. Her prime example of a transformer mission is Germany’s Energiewende, the country’s decades-long commitment to shifting from fossil fuels to renewable energy. Framed around a clear political direction but designed to stimulate bottom-up innovation across multiple sectors, it had no single endpoint, no moment at which Germany could declare the transition complete. Instead, the investment built the regulatory, market, and institutional infrastructure that made the direction of travel self-sustaining across governments, industries, and generations. The goal was transformation, not achievement. By this articulation, metascience is a transformer mission. Its aim is not to solve a single problem but to build the capacity that makes the entire science system better at learning from evidence — and to embed that capacity inside the accelerator missions already underway.
Over the waning weeks of 2025, the White House Office of Science and Technology Policy (OSTP) issued an ambitious Request for Information (RFI), asking for public input on how to “Accelerate the American Scientific Enterprise.” The RFI called for a comprehensive assessment of how the federal government prioritizes and structures scientific research, including a specific question on the role of metascience. FAS submitted a comprehensive response. Many organizations called for evidence-based reform of federal grantmaking, with near-universal agreement on the need to reduce administrative burden, reform peer review, and build metascience capacity inside federal agencies. But notably absent is any proposal to build metascience as a discipline in its own right and equip it to measure and evaluate this scientific enterprise in transition.
Metascience Will Be Central to Transformative Change
Metascience cannot stop the rapid disruption and changes in the scientific enterprise underway, but it can help decide where it ends up. That means doing a few concrete things now, while the system is still in motion, before it settles into a new normal.
- Treat what’s already happening now as THE experiment. The Genesis Mission, NSF’s new X-Labs, and other major initiatives launching right now are, in effect, live tests of new ways to fund and organize science, whether or not anyone designed them that way. Metascience doesn’t need to wait for a dedicated initiative of its own; it can study what’s already underway and feed lessons back before these efforts harden into permanent practice. But that only works with real rigor and intention, building in the comparisons and honest evaluation needed to learn and generate evidence.
- Build the basic data and record-keeping first. Before you can run real experiments on what works in funding science, agencies need to publish their data in usable, machine-readable form, with consistent standards so grants can be compared across agencies, and pre-registration so results can’t be quietly reshaped after the fact.This will be the only honest way to tell whether any reform actually worked.
- Build the field, starting with the people. Right now, a growing number of people want metascience’s findings, but almost nobody is funding what it takes for the field to actually exist over time — training programs, places to publish, a professional community. Training a researcher takes years, so anyone who wants to be doing serious metascience work by 2035 needs to start now, in the next year or two, whether through a Ph.D. or another path into the field.
- Give the people doing this work real say, not just a seat at the table. If metascience teams inside agencies like NSF, NIH, and DOE can only offer advice, their work tends to get quietly absorbed into business as usual. They need a direct line to the people who decide what gets funded, real ability to change how grants are evaluated, protection from being slowly watered down over time, and enough independence to report findings honestly even when those findings are unflattering to the agency that funds them.
Complex systems in transition do not suddenly pause. And unlike Lake Washington, the transition of the U.S. scientific enterprise will almost certainly not restore to a prior equilibrium once acute pressures are removed. We need the field of metascience to be ready to guide the system into a state that is more responsive, more trustworthy, and fit for purpose in a world that looks nothing like the one the current system was built for. That is metascience’s “moonshot.”
Revealing the Hidden Data Supporting Alzheimer’s Patients with the Federal Data Field Guide
Federal data play an essential – and mostly invisible – role in supporting the more than 7 million Americans with Alzheimer’s and their nearly 12 million unpaid caregivers.
Too often, we think of federal data as limited to high-profile datasets like jobs, weather, and the census. But beneath that surface is a diverse ecosystem with well over 500,000 datasets – including those tackling Alzheimer’s disease and related dementias (ADRD).
We recently published the Federal Data Field Guide to highlight the different species of federal data that benefit everyday Americans. In this post, we use the Field Guide’s framework of eight categories (Statistical, Administrative, Geospatial, Scientific, Accountability, Evaluation, Navigation, and Reference data) to scout for federal datasets that are improving the lives of Americans affected by ADRD.
Federal Data Improving Our Understanding of ADRD
Here’s a quick look at the valuable federal data underpinning our understanding Alzheimer’s disease and related dementias (ADRD):
How prevalent is Alzheimer’s Disease in the U.S.?
Statistical Data measure population level characteristics. The National Center for Health Statistics’s National Health Interview Survey estimates that about 4% of the non-institutionalized population over 65 has been diagnosed with dementia, while mortality data from the National Vital Statistics System (NVSS) track deaths from Alzheimer’s. NVSS also fits in the Administrative data category from the Field Guide, because it draws information from death certificates processed by state governments.
What are the genetic, social, and environmental determinants of Alzheimer’s disease?
Geospatial Data describe location and environmental information about the world. Evidence links air pollution to increased Alzheimer’s risk. The EPA’s Particulate Matter Pollution data–collected through a network of monitors operated by state, local, and tribal agencies–are vital for enforcing clean air regulations and, by extension, reducing risk for ADRD.
Scientific Data advance knowledge through federal or federally-funded research. The Veterans Administration’s Million Veteran Program (MVP) has identified key variables associated with ADRD in veterans, such as traumatic brain injury, depression, and military environmental exposures, offering critical insights for both prevention and intervention. And, the NIH’s GenBank hosts a trove of genetic data that researchers use to develop screening tools, treatments, and medications for ADRD.
How do federal data help caregivers and patients?
Navigation Data help citizens find and access services. The Centers for Medicare and Medicaid Services’ Nursing Home Care Compare Database helps caregivers assess facilities for their loved ones. This dataset also fits into the Accountability category, as it includes critical quality metrics. NIH’s ClinicalTrials.gov dataset helps families identify the clinical trials that might be a match for their loved one. Another important navigation dataset from NIH is PubMed, which houses over 40 million citations to the biomedical literature, enabling scientists, clinicians, and families to stay up on the latest Alzheimer’s research.
Reference Data provide standardization across systems. How can this type of data help people with ADRD? The Social Security Administration maintains a reference dataset on medical conditions qualifying for Compassionate Allowances. This in turn enables a 45-year-old diagnosed with early-onset Alzheimer’s to fast-track her disability benefits because the condition is officially recognized in the dataset.
Evaluation Data assess how effective different programs are. For example, HHS has funded data collection on millions of telehealth appointments to evaluate access, quality, and utilization. These data are used to improve telehealth services, which can be a game changer for ADRD caregivers – especially those with mobility impairments that make it hard to make in-person medical appointments.
These are just 10 of the many federal datasets related to ADRD that span the eight categories of federal data. Without understanding the different categories of federal data, it is easy to overlook datasets like these that are essential for improving the lives of people impacted by Alzheimer’s disease.
It is more urgent to appreciate the interplays of data brought to bear on just one health issue at a time when administrative actions, budget cuts, and destaffing, threaten the capacity of agencies to collect, maintain, and publish them.
Mapping out a broad range of federal datasets in a specific domain like ADRD is a useful exercise for two reasons. One, most federal data are underutilized or taken for granted. We’ve already paid for these data as taxpayers and we should make good use of them; every reuse of a federal dataset is value-added to its return on investment. And two, the federal data ecosystem will be more resilient when more people know about, care about, and advocate for the datasets on which they depend. Identifying and talking about the value of federal data is the first step to protecting their continued flow.
Take action
Do you want to get a better understanding of the federal data that make your life, or your life’s work, better? And do you want to get the tools to help keep those essential data flowing?
Here are some concrete steps you can take:
- Use the data categories in the federaldatafieldguide.us to map out the federal datasets you might be taking for granted, whether you work on issues around housing, climate, veterans, children, small businesses, agriculture, or whatever.
- Sign up for the newsletter at dataindex.us to be up to speed on opportunities to give feedback to federal agencies on specific datasets or data policies.
- Visit the Data Checkup at dataindex.us to see which of your datasets have been assessed for risk – and let us know what datasets you value and want prioritized.
- Check out essentialdata.us to see if the datasets you’ve identified are represented in our collection of use cases about how federal data benefit everyday Americans. If not, submit a dataset, or reach out to schedule a workshop to create use cases that cover your domain.
The Federal Data Field Guide is a free, plain-language resource developed by Denice Ross and Christopher Marcum as part of an Executive Fellowship in Applied Technology Policy at UC Berkeley. Learn more at federaldatafieldguide.us.
The FAS Data Policy Institute is a catalyst for advancing the field of data policy to build better outcomes for the American people, by building the civic infrastructure to monitor changes to federal datasets; mobilize data stakeholders to engage with government officials; advance policies to protect and improve essential public data; and design America’s future data ecosystem.
Photo: Histopathogic image of senile plaques seen in the cerebral cortex in a patient with Alzheimer disease of presenile onset, CC-BY KGH.
Scaling Team Science is the Important Experiment We Need
A tumultuous period for federal research funding would seem like precisely the wrong moment to experiment with new funding models. But that’s wrong.
Federal research funding is a cornerstone of U.S. competitiveness and national security – and one that is facing an unprecedented moment of uncertainty. Navigating this uncertainty will require experimentation with new models that can accelerate scientific progress. The National Science Foundation’s (NSF) new X-Labs program, which encompasses roughly just two percent of the funding institution’s overall research budget, is exactly that.
X-Labs build on promising early results from a recent proliferation of philanthropically-supported independent lab models, and in particular, Focused Research Organizations (FROs) that catalyze capital and harness team science to unlock entirely new fields of scientific inquiry. These ambitious and high-leverage research infrastructure investments have already delivered tremendous breakthroughs: the largest map of drug-target interactions ever built, a global atlas of ocean alkalinity enhancement to aid in carbon dioxide removal, and reductions in the cost of proteomics twenty-seven-fold that can accelerate biomedical research.
As impressive as early FRO results have been, they just scratch the surface of the model’s potential. There are many more possible breakthroughs, significantly beyond the capacity of philanthropic resources to support. Yet team-based, goal-oriented science has always been a difficult fit for the dominant federal funding models organized around research project grants. That’s why what’s truly novel about X-Labs is that NSF has found a way to achieve scale: up to $1.5 billion in funding structured around milestones, giving teams an opportunity to work on problems that lend themselves to a different set of incentives than academia typically requires.
Not everyone is excited. In a recent essay in Science, Pierre Azoulay of MIT makes a careful case for skepticism that government funding is the right model for scaling independent labs. Others have highlighted the potential zero sum funding dynamics facing academic institutions, based on recent Science reporting, though the perception that NSF is diverting resources from academic research funding to X-Labs is inaccurate. Still others have expressed concerns about public accountability, given X-Labs’ substantial operational independence at a moment when the Trump administration seeks to assert greater political control of grantmaking.
The harsh reality is that academic institutions are at a crossroads; the modern research university runs on a fraying cross-subsidy model that has hidden the real price of doing research. With simultaneous pressure on federal grants, indirect cost recovery, tuition and endowment returns, the path forward will require experimentation with alternative models at the institutional level.
NSF has explicitly framed X-Labs as an opportunity for academic institutions, giving them the license to create a platform for their researchers to engage in goal-oriented team science with less internal administrative burden and absent the proposal and publishing treadmills. For some institutions, this opportunity will prove a golden one. And in the spirit of a true experiment, some of what works in this sandbox – intellectual property, team structure, novel metrics – may be worth broader adoption as institutions find their way past the crossroads.
I see X-Labs as the culmination of two significant developments in how research is conducted.
The first is the bipartisan mandate handed to NSF by Congress when it created the Technology, Innovation, and Partnerships (TIP) Directorate through the CHIPS and Science Act. Some architects of this legislation, and the Endless Frontier Act on which it was based, were explicit that formidable and growing competition from China meant that our research institutions would need to evolve, and their intent was to provide NSF the latitude to fund targeted priorities differently. As always, diagnosing unified Congressional intent is difficult, but what Congress codified was the creation of the first new NSF directorate in over 30 years, one with the flexibility to deploy flexible “Other Transactions Authority” for nontraditional grantmaking – not as a substitute, but as a supplement and translational accelerant to NSF’s core research funding mission.
This bet is already delivering unexpected results. In 2022, few thought NSF was well-suited or capable of successfully driving the regional impact of translational research. But four years later, patience is paying off: $135M of TIP investment in nine regional hubs through the Regional Innovation Engines program has catalyzed more than $2B in matching commitments – an impressive 15x multiplier on taxpayer dollars.
Like the Engines program, X-Labs is an experiment being undertaken with TIP’s own budget, not with funds siphoned from other directorates that would otherwise be used for research project grants, as recent reporting has suggested. And X-Labs’ initial focus on scientific instrumentation directly feeds discovery science across NSF’s core research domains. For the community that championed CHIPS and Science, these kinds of experiments are exactly what we had hoped to see.
Second, X-Labs follow strong, decades-long precedent for the federal research enterprise to embrace and scale models that philanthropy has de-risked. Examples abound.
For years, the Howard Hughes Medical Institute (HHMI) has provided stable, investigator-centered institutional funding as a means to enable transformative research by insulating extraordinarily promising PIs from the constraints of grant-based funding. Azoulay’s own pioneering research on the HHMI model — which funded ‘people not projects’ — found this model had a significant role in generating novel research. Inspired by the example, the National Institutes of Health has since added pathways to its funding portfolio – such as the Transformative Research Award – to bet on extraordinary people rather than their projects.
In 2003, philanthropist Paul Allen put $100 million into a new brain science institute built on the premise that a tightly coordinated team focused on an ambitious goal – creating an atlas of gene expression in the mouse brain – could deliver success where a series of traditional research grants wouldn’t. This team-based model succeeded, delivering the foundational infrastructure on which significant subsequent federal efforts such as the BRAIN Initiative ultimately relied to scale.
Recent years have lent the philanthropically-supported independent lab model more significant momentum. Since 2021, the Arc Institute in Palo Alto – a university-affiliated biomedical research organization – has organized researchers into teams in pursuit of breakthrough research outside of a traditional academic structure. And in 2020, I helped champion Adam Marblestone’s and Sam Rodriques’ original FRO concept for independent, team-based and time-bound nonprofits aimed at problems that sit in the gap between what academia rewards and what industry will fund. The original proposal targeted federal research agencies, but I saw firsthand that these institutions weren’t ready for the experiment without some fundamental questions answered through philanthropic de-risking. That day has come: Convergent Research, with hundreds of millions in philanthropic support, has since launched ten FROs in domains as distinct as brain mapping, ocean carbon modeling, and software for mathematics, with significant proof points.
Now, nudged by some sharp policy thinking from Caleb Watney at Institute for Progress on how the government can scale the approach, X-Labs seek to expand on what FROs have shown is possible: the generation of foundational infrastructure for entire new fields of research science. The vision for X-Labs is appropriately as a tool of portfolio diversification – as announced, just two percent of NSF budget – rather than a means to replace or indirectly reform the dominant funding model of research program grants. And as with any true experiment that pushes scientific boundaries, there will be learning and evolution along the way. X-Labs funding will be time-bound, and as institutions they should wind down when they’ve reached their goals. Some will surely fail altogether, and some will unlock whole new fields of scientific inquiry, and with that economic competitiveness and growth.
On this path forward, NSF will also need to strike a difficult balance between autonomy and accountability. Its decision to fund X-Labs through TIP’s Other Transaction Authority – a flexible contracting mechanism simply defined by what it’s not – should not mean an absence of accountability. But accountability mechanisms must be fit for purpose – lightweight, public-facing, and outcome-oriented (i.e., does the public know what was funded and what it produced?) -– rather than process-heavy and approval-based. The initial X-Labs solicitation offers encouraging indications of results-oriented accountability (milestone payments) and IP management plans, but the details are very much in execution to come.
I’m bullish. At a tumultuous moment for research, we need more experiments like X-Labs, not fewer.
To scale up climate solutions, local governments need to accelerate system changes
When I ran for city council in Boulder, Colorado in 2023, everyone talked about climate change. Forum after forum, all ten candidates spoke up for the climate.
And cities saying climate change matters is typical. The number of US cities with adopted climate action plans is in the hundreds.
That’s what we need, since cities drive the bulk of greenhouse gas emissions and are on the front lines of climate havoc.
More specifically, for large-scale climate solutions to work, cities have to really stretch. That’s according to the Intergovernmental Panel on Climate Change (IPCC), which says cities need to rapidly become compact, efficient, electrified, and nature‑rich urban ecosystems where we take better care of each other and avoid locking in more sprawl and fossil‑fuel dependence.
Yet, big-picture progress in the United States is critically insufficient. Those are the words of Climate Action Tracker, an independent scientific analysis evaluating climate commitments. The US has pledged to reduce 2030 GHG emissions levels by 50–52% below 2005, yet the latest projections show we are on track to achieve at best only 29–39%—assuming no further backsliding.
And earlier this month, the Trump administration withdrew our federal government from the international climate agreement process.
So when local governments say “we’re on it,” what is a concerned citizen to think?
What local government climate solutions look like
Climate advocates are used to talking about climate action. But for local governments, the measuring stick for climate progress isn’t simply action. What counts is measurable progress towards specific, substantive transitions.
Transitions to walkable, compact neighborhoods where abundant, space-efficient middle housing near jobs and services let most residents meet daily needs within a short walk or bike ride, reducing trip lengths and housing and transport costs.
To transit-rich, highly bikeable towns where frequent, accessible service and a connected, protected network allow seniors and youth travel independently and where per-capita car dependence falls.
To fully-electrified communities in which homes and transportation run on clean, distributed power, working efficiently, that delivers lower bills, healthier indoor air, and outage resilience, with benefits accruing equitably to residents.
To enhanced landscapes of bioswales, permeable streets, restored wetlands, and drought- and fire-resilient shade trees that cool neighborhoods, absorb stormwater, and buffer heat, flood, and smoke risks.
To resilient local food systems that blend urban agriculture with regional producers, food hubs, cold storage, and compost-to-soil loops to deliver reliable, affordable, nutritious food even during heat, drought, or supply disruptions.
There is good news: The transitions we need, and the solutions and capacity we need to implement them, are showing new signs of life. That’s evident in two trends.
One trend is local governments playing a bigger role in climate solutions. The number of U.S. cities reporting to the CDP, a global system for disclosing climate progress, has grown to over 150. Now more than 200 US cities have committed to 100 percent clean electricity. And cities’ climate action plans are showing a visible shift from a focus on municipal operations to community‑wide impacts of buildings, transportation, and waste, and more sophisticated thinking about resilience.
As the federal government has retreated, advocates are increasingly realizing cities and counties have tools to lead. Local governments manage streets, land use, buildings, public fleets, transit, and major service contracts. They can strongly influence state-level actors, like energy utilities and air quality programs, and be providers of those services directly.
There is proof of this awakening in the large numbers of people suddenly running for local office on climate. Political organizing coalitions such as Run on Climate and Climate Cabinet helped elect more than 50 local leaders running on climate in 2025. One of the year’s most high-profile candidates, Zohran Mamdani, won with “fast and free” buses–one of the measures IPCC has highlighted as a meaningful mitigation measure that saves more money than it costs–as a centerpiece of his campaign.
The other trend is a greater focus on wellbeing. Research included in the latest IPCC report shows demand-side measures can cut end-use emissions by roughly 40 to 70 percent by 2050 while improving daily life and making communities stronger. And wellbeing is the currency of local governments and local politics. Concrete quality of life issues dominate local elections and policymaking, which is where climate action takes root—or doesn’t.
Climate action prompted by a desire for healthier, happier, and less expensive lives is happening. People are adopting electric cars, e-bikes, heat pumps, and induction stoves because they work better, are cheaper to operate, and healthier. The intersection of climate solutions and wellbeing is central to a 2025 bestseller Abundance and to the national conversation it kicked off about defining and achieving “abundance.” The topic of wellbeing was a bright spot at the COP30 climate talks via the World Health Organization’s report, “Delivering the Belém Health Action Plan.”
These two trends reinforce each other. Local governments oversee the services where wellbeing, decarbonization, and resilience meet. When those services are designed as a system, investments can compound to create more value for more people, who then have a stake in continuing the transition. And the importance of rallying around local governments to carry climate solutions forward is becoming clearer as U.S. national policy looks structurally less reliable than most experts used to think.
Difficult conditions for change
But local governments face headwinds. Existing policies and markets, like those that have created widespread car dependence and extensive natural gas systems, create momentum that favors the status quo and encourages continued investments that lock us in further. Simply put, it’s easiest to keep doing it the way we’ve done it before, and then we dig ourselves in deeper.
Local governments purposefully design systems to keep things stable. Most likely, whatever your town or county is doing is based on the direction of long-term plans, from departmental plans to bigger comprehensive plans. Those plans often come up for renewal only every few years or longer, and if you miss that window or fail to follow procedures, making big change is nearly impossible. Related, local governments tend to have policies and practices for conducting community engagement that deliberately create a high bar for making major turns.
On top of all that, local governments in the U.S. are suffering a long-term decline in investment that leaves them with significant and growing cash flow constraints, heavy workloads, limited time to deliberate, and pressure to deliver. The pandemic and recent national political forces reduce their maneuverability even more.
Political will necessary but not sufficient—concrete transitions are needed
In order to drive climate transitions under such tough conditions, political will is necessary but it is not sufficient. For local governments to scale up climate solutions, they need to take tangible, visible steps to change systems, consistent with evidence-based recommendations, outlined by institutions like the IPCC.
Here is what that can look like – and what advocates can look to encourage:
1. Transition plans
Climate issues touch everything, so all local governments can point to doing climate things. But the difference between lists of activities and high-reward strategic commitments that make good use of time is everything. The latter requires a clear plan to make transitions happen, with defined outcomes and milestones, and dogged pursuit.
Ambitious climate action at the local government level means being clear about the transition(s) the community is focused on, which could include the previously mentioned examples, along with what successful completions looks like and by when. This involves working on at least two tracks concurrently—both integrating ambitious transformations into long-term planning exercises, for which adopting changes may or may not be available right away, and taking whatever more tactical action is possible now to support such planning and concrete action to the fullest extent possible.
- Transition plans might consider: What needs to be different? Who are the elected officials and partners needed to make the transition work? What barriers could inhibit progress, and what is the strategy for overcoming them? What are the key inflection points in behavior adoption, and what is the change model to reach them?
2. User experience
Cities often add a bike lane in one place or restore a bus line in another. What truly changes behavior is a complete experience that makes the pro-climate option the intuitive choice. Kids can bike around town without parents fearing they could be hit by a driver. You can count on bringing a large electric bike anywhere and park it safely. Buses are within a 10-minute walk of home and arrive every 10 minutes. Utility investments in electrification actually lower monthly bills. To make climate transitions attractive and sticky, we have to confront gaps that get in the way of people’s experience from their vantage point.
A practical opportunity for local governments is to use the tools of user experience (“UX”) and be responsible for how the ecosystem works and feels from the immersive standpoint of users. UX is an interdisciplinary field that uses research, psychology, and design to remove friction and ensure a seamless journey for users.
- Some questions for managing the user experience: Who is/are the transition(s) for—who are the “users” involved that will experience change? What do they need to do and stick with to make the transition work? What performance measures are needed to constitute progress, and how do we get elected officials and executives to care about them? Where are there persistent, hard-to-reach gaps that can be spotlighted as “known issues” for partners and other jurisdictions to see and possibly be helpful with? What are the gaps between our intent and reality, and how can we overcome them?
3. Public service delivery
One of the core jobs of local government is to provide public services like zoning, safe transportation, building standards, air quality protections, and emergency management. Providing services is also generally the justification for spending public money. And services are where the planning activities that local governments tend to be so careful about materialize in the real world. So if local governments are going to be engines of climate action, then day-to-day service delivery—their core product—is where most of that action will show up. Climate action will appear in what gets approved, funded, built, maintained, enforced, measured, and improved.
Local governments already deliver public services. So the opportunity is to evaluate how core local government services can or should be tuned and/or reorganized to drive climate and resilience outcomes. This includes formal adoption in comprehensive plans, capital improvement programs, and strategic plans, and clear alignment with budget priorities. When leaders routinely report on progress and adjust course publicly, it signals that climate transitions are a core organizational responsibility rather than a side project.
- An evaluation of services might consider: How is climate action aligned with where money is actually spent? What changes in standards, contracts, and operations are needed to lock in better outcomes? How can the thinking and tools of service development improve user experiences?
4. High-level ownership
Plans only come to life when people who have the right level of power and accountability own delivery. Inside local government, that means both the elected body (mayor, city council, and/or their equivalents) and executives (city manager, their deputies, and in the case of a “strong mayor” form of government, the mayor) adopt the initiative as their own. Roles and accountability are defined and gaps are addressed. Resources are allocated through direct investments and through partnerships that expand capacity.
High-level ownership of climate solutions in local government happens when transitions are included in the agency’s highest-level plans and strategies.This includes formal adoption in comprehensive plans, capital improvement programs, and strategic plans, and clear alignment with budget priorities. It also looks like leaders routinely communicating to the public about the transitions under way, the progress against them, and how community members can help support the journey.
- Some ways to gauge high level ownership: Do staffing, partnerships, procurement, budgets, and intergovernmental advocacy support transition intent? Is change management happening as needed, with a clear vision and ongoing communication expressed to the public? Is the work of transitions integrated into core functions and the day-to-day responsibilities of staff necessary for success? How are electeds supporting and leading?
5. Playbook of procedures
Local government commitments are heavily shaped and constrained by procedure, like protocols for what gets a hearing and when, annual or biennial work plans, and comprehensive plans that may come around only every few years or longer. Communications between elected officials and staff may be limited by city ordinance, and communications among elected officials may be very limited by state law. There are also often arcane, highly-localized meeting customs. Getting things done requires working through these procedures and often landing decisions in small windows that are easy to miss.
A playbook for how climate transitions are going to make their way into staff proposals, planning processes,and budgeting is fundamental to turning a good idea into something real. Such a playbook is needed to spell out who does what, when, and through which formal channels, so that key decisions do not depend on heroic one-off efforts. It also helps new staff and elected officials quickly understand how to use existing procedures to advance climate goals, rather than be derailed by them.
- Some questions to ask: What are the relevant scheduled planning processes and upcoming meetings, and who is required to fully elevate the work there and how? What learning systems or listening structures exist, and how are they being used to surface discrepancies and continuously drive improvement?
Conclusion
To scale up climate solutions through local government, we need at least two things. First, political will, which is familiar to most advocates. Looking into 2026 and beyond, climate advocates have great opportunities to continue increasing the proportions of elected local bodies who are led by politicians serious about climate solutions. Everyone has a role to play: run for local office, support local climate candidates, use whatever powers of creativity and persuasion you have–from writing to speaking to organizing and beyond–to help make climate action a core election issue in your community.
The second—and where we need greater shared focus—is to make local governments responsible for specific, strategic commitments to systems change. To do that, help build transition plans that commit to providing great user experiences, an approach to public service delivery that is aligned with those objectives, ownership by city council and the city manager or mayor, and a clear playbook for how strategic climate commitments are going to be adopted and rolled out.
Not everything is going right for the climate movement. But there are some fantastic bright spots, and one of those is big new local government innovations that are starting to unfold.
Looking into 2026, I’m excited to be a part of the movement to help local governments drive the next generation of climate progress. And a big hat tip to FAS with its regulatory rethink and government capacity work as well as ICLEI USA, both partnering with local officials like me to map out how cities can translate ambitious climate goals into durable systems change.
There are great things ahead, and so much room to work together.