One of the most astonishing aspects of science policy over the past 30 or so years is the consistency of R&D funding levels as a proportion of the discretionary budget [Figure above]. (Discretionary spending is the part of the budget that is subject to annual congressional decisions about spending levels.) Since the mid-1970s, nondefense R&D budgets have constituted between 10 and 12% of total nondefense discretionary spending. Total R&D (defense and non-defense) shows a similar stability at 13 to 14% of the total discretionary budget. This consistency tells us that marginal changes in the R&D budget are tightly coupled to trends in discretionary spending as a whole.
Sarewitz argued that the long-term stability in R&D funding can be traced, in part, to a bipartisan consensus that R&D, especially support for basic research, was broadly in the public interest. He explained:
[T]he political case for basic research is both strong and ideologically ecumenical. Unlike applied R&D, basic research appeals to the political left as an exemplar of the free expression of the human intellect, to the political right as an unambiguously appropriate area of government intervention because of the failure of the market to provide adequate incentives for private-sector investment, and to centrists as an important component of the government’s role in stimulating high-technology innovation.
During the administration of George W. Bush, R&D funding averaged 12.5% of discretionary spending — Obama, 11%; Trump 1, 9.5%; and Biden, 10.5%. The FY2026 budget proposal by the Trump administration cuts R&D funding to 7.3% of the discretionary budget, which would be by far the lowest since at least World War 2.
Of course, presidents don’t decide appropriations, Congress does. Neither the executive nor legislative branches make decisions about the overall R&D budget, which is a summation after the legislative sausage is made.
According to AAAS, the Senate has proposed reversing some of the proposed cuts to R&D in the president’s budget, but these would only increase the proportion of R&D spending to ~8.8% of discretionary spending, still a post-WW2 low.
Sarewitz explains why the U.S. does not have a national science policy:
In total, the decentralization of influence over S&T budgeting in the federal government precludes any strategic approach to priority setting and funding allocations. Although an “R&D” budget can be—and is—constructed and analyzed each year, this budget is an after-the-fact summation of numerous independent actions taken by congressional committees and executive-branch bodies, each of which is in turn influenced by its own set of constituents and shifting priorities. From this perspective, if science policy is mostly science budget policy, then one can reasonably assert that there is no such thing as a national science policy in the United States.
However, the actions of the Trump administration indicate that while there may be no strategic approach to priority setting, there is a strategic direction — Less.
“If one cannot choose among alternative policies in terms of what they may achieve, then policy preferences are revealed as nothing more than expressions of parochial values and interests” — Dan Sarewitz
In May, Michael Kratsios, director of the White House Office of Science and Technology Policy in the Trump administration, made a case for why less is more at the U.S. National Academy of Sciences:
We are seeing diminishing returns. For example, despite biomedical research budgets soaring since the 1990s, scientific progress has stalled—new drug approvals have flatlined or even declined, more researchers are needed to achieve the same outputs, and workforce training has stagnated. More money has not meant more scientific discovery, and total dollars spent has not been a proxy for scientific impact. . .
[P]rivate money’s growing place in America’s scientific enterprise presents opportunities. In particular, in a period of fiscal constraints and geopolitical challenges, an increase in private funding can make it easier for federal grantmaking agencies to refocus public funds on basic research and the national interest.
What we target is what we measure, and what we measure is what we get more of. To get more bang for America’s research bucks, we need to enhance the creativity and precision of our funding. Spending more money on the wrong things is far worse than spending less money on the right things.
This brings us back to Sarewitz.
His central argument was that the science community’s longstanding efforts to lobby for more R&D funding misses what matters most in science policy:
What is the capacity of a particular science policy decision to advance a given desirable outcome? This should be the most fundamental science policy question, because if one cannot answer it, then one cannot know whether any particular policy is likely to be more or less effective than any alternative policy. And if one cannot choose among alternative policies in terms of what they may achieve, then policy preferences are revealed as nothing more than expressions of parochial values and interests.
A science policy focused on LESS is just as empty as one focused on MORE.
Sarewitz explains:
It is not very difficult to imagine the types of questions that might help to inform a transition from science policy based on “how much” to science policy based on “what for,” though it is certainly the case that such questions may be rather unwelcome in national R&D policy discussions . . .
He suggested those questions might not be welcomed because their answers would very likely lead to changes in funding priorities, with some existing areas of R&D seeing more funding and others less — upsetting the science community’s longstanding truce across agencies and disciplines to argue for everyone’s funding to increase.
Instead of more or less, Sarewitz suggested 10 questions that we should ask of federal science policy:
What are the values that motivate a particular science policy?
Who holds those values?
What are the actual goals that the policy is trying to achieve?
What are the social and institutional settings in which the R&D information or products will be used?
What are the reasons to expect that those are settings for effectively translating the results of R&D into the goals that justify the policy?
Who is most likely to benefit from the translation of the research results into social outcomes?
Who is unlikely to benefit?
What alternative approaches (through either other lines of research or nonresearch activities) are available for pursuing such goals?
Who might be more likely to benefit from choosing alternative approaches?
Who might be less likely to benefit?
With only a few exceptions (e.g., nuclear energy and artificial intelligence) the Trump administration has apparently not considered questions such as these.2 As a result, the administration’s proposal to cut R&D spending by more than $50 billion (from a 2025 total of >$190 billion) is not justified in terms of policy.
The result is that we are left with science policy without policy, leaving partisan politics — both in Washington and in the scientific community — to take center stage.
A follow up post will take a close look at how the Trump administration considers (or not) experts and the institutions of expertise within the executive branch.
