More than 10,000 doctoral-trained experts left federal science roles in 2025. That loss will not stay in Washington. It will show up in labs, classrooms and hospitals, including our own at Dartmouth.
Late last month, the journal Science published a statistic that deserved far more attention than it received: 10,109 doctoral-trained experts in STEM and health fields left federal jobs in 2025, based on its analysis of employment data from the United States Office of Personnel Management.
A workforce statistic can feel abstract. However, federal scientists and Ph.D.-level reviewers are part of the country’s basic safety infrastructure; they evaluate evidence, set standards and provide the public data that everyone else relies on, including universities and hospitals.
Science’s analysis found that this wave of departures represented about 14% of the government’s STEM Ph.D. workforce at the end of 2024. That is not normal turnover; it represents a significant loss of institutional knowledge.
This was not primarily a story of mass firings. Reporting on the same dataset indicates that, at most agencies, the dominant categories were quitting and retirement, not reductions in force. That matters because it points to something more corrosive than a single round of layoffs: an environment that makes highly trained people decide their work is no longer possible, or no longer worth it.
OPM itself has described new separation mechanisms that accelerated departures in 2025, including a “Deferred Resignation Program.” Consequently, 317,000 federal employees left the government last year. You do not have to declare war on science to weaken it: reduce the size of government, and specialized expertise thins out.
Meanwhile, the administration’s original 2026 budget request sought steep reductions in federal science funding, including an NIH budget roughly 40% below its 2025 appropriation. Congress ultimately rejected the steepest cuts and preserved core NIH funding. But for months, research institutions operated under the threat of deep reductions. Stability is not instantly restored simply because the final number is higher than the proposal.
We can disagree about particular programs. But the overall direction is hard to ignore. If the federal government is losing large numbers of Ph.D.-level experts while also proposing major cuts to research funding, the result is straightforward: less research, slower progress and policy decisions made with less evidence behind them.
Dartmouth is not insulated from that volatility. Federal science policy runs directly through the College’s labs, its medical center and the research positions that shape student careers.
In February 2025, The Dartmouth reported that roughly 1,300 Dartmouth employees and 400 Dartmouth-Hitchcock Medical Center employees held positions supported by federal grants and could be at risk under proposed NIH funding changes.
New Hampshire Public Radio likewise reported last winter that proposed NIH changes could put approximately $24 million at risk for the College, which received about $97 million in NIH funding in fiscal year 2024. Those are not symbolic figures. They translate into payroll, lab maintenance, clinical trials and the undergraduate and graduate research roles that often serve as entry points into scientific careers.
Dartmouth’s own administrative response illustrates the point. The Office of Sponsored Projects posted ongoing federal-transition guidance, including updates on litigation over NIH indirect cost policy changes. Universities do not build such pages when conditions are calm.
So, what is the logic behind the administration’s moves? A charitable answer is “efficiency.” But the pattern here does not look like careful pruning. It looks like hostility toward the very idea of independent expertise.
If your goal is rapid deregulation or the ability to declare “problem solved” without measuring the results, independent scientists are an obstacle. Their work produces evidence. Evidence creates friction. It slows decisions and demands accountability.
In 2024, Politico’s E&E News reported on leaked Project 2025 — a right-wing presidential power consolidation initiative developed by the Heritage Foundation — training material in which a former Trump administration official urged appointees to “look for climate change language and get rid of it.” Whether or not every single element of Project 2025 becomes policy, the instinct is telling: When evidence is politically inconvenient, don’t rebut it. Delete it.
History offers sobering examples. In the Soviet Union, the state elevated Trofim Lysenko’s politically- convenient claims and crushed modern genetics, leading to real scientific stagnation and agricultural harm following the politicization of biology. In apartheid-scarred South Africa, AIDS denialism at the highest levels helped delay effective treatment, contributing to more than 330,000 premature deaths from obstructing life-saving therapy.
The point is not that the United States is fated to become any one of these places. However, when you weaken independent expertise, you do not create a freer society. You get a society governed by weaker information, narrower debate and larger errors.
Those mistakes do not stay in Washington. They appear as slower drug reviews and regulatory backlogs, with the FDA offering bonus payments to reviewers as the agency deals with roughly 20% staff attrition. They appear in public health disruptions and in growing uncertainty among young scientists about federal careers.
And they appear on campuses like Dartmouth’s.
The consequences become real when you are a senior applying to Ph.D. programs or a Geisel researcher trying to keep a clinical trial alive. Federal science policy shapes who stays in research and who decides to leave it.
Congress rejected the steepest proposed cuts this year. But the larger question is whether the United States continues to treat independent expertise as a public good and something to be protected even when it complicates politics.
A nation does not remain a scientific leader by accident. It remains one by building institutions that outlast administrations and by allowing evidence to constrain power rather than serve it. “10,109” is not just a headline; it is a warning. If the country allows the normalization of governing without evidence, we will still be governed by something. It just won’t be the truth.
Opinion articles represent the views of their author(s), which are not necessarily those of The Dartmouth.
