The Secret of Fixing US Research Priorities with the Philosophy of Science
The secret of fixing US research priorities is hiding in plain sight, and it has nothing to do with budget percentages or workforce pipelines.
It has to do with what we think science actually is.
Every agency, review panel and program director in the American research enterprise runs on an unexamined theory of knowledge. Most of them don’t know it. The ones who do know it think it doesn’t matter. The rest of us are stuck living with the consequences.
Let me slow down and explain.
The Ghost in the Grant Review
When a senior agency official tells a university audience that they fund “transformative, high-risk, high-reward research,” they mean it. The problem is they don’t want to mean it. Reviewers reading those proposals want to fund research that will pay off in three years. Their incentive architecture, promotion criteria and reputational instincts all lean toward the safe, incremental, and familiar.
This isn’t a moral failing. It’s a philosophical one.
The dominant model of science that governs US institutions was largely assembled in the 1940s. Vannevar Bush’s landmark report, Science—The Endless Frontier, told America that pure basic research flows like water into a reservoir of fundamental knowledge. Applied science later dips a bucket into that reservoir and — presto — you get radar, antibiotics and a booming economy.
That image of science as an ordered, linear pipeline is a philosophical position. It’s not a description of what scientists do. But it has shaped how the National Science Foundation, the National Institutes of Health and the Department of Defense think about “basic versus applied” work for over seven decades.
It was a useful fiction in 1945. It is, today, an expensive one.
What Philosophers Actually Said
Here’s the uncomfortable part. Philosophers of science have spent the last seventy years dismantling the tidy picture that policy makers still rely on.
Thomas Kuhn’s The Structure of Scientific Revolutions arrived in 1962 and threw a wrench into the machinery. Kuhn argued that science doesn’t progress by steady accumulation. It moves through long periods of “normal science”, where researchers solve puzzles within a shared framework. Then anomalies pile up. Crises hit. Eventually the whole framework collapses and gets replaced by a new one.
Kuhn’s story is messier, more human and far less comfortable than the pipeline model. It says that scientific paradigms are often incommensurable. A scientist trained under one framework literally cannot see the world the way someone under another framework does.
You’d think US research policy absorbed this lesson sixty years ago. It didn’t.
The peer review system, the grant review criteria, even the design of academic departments all assume the opposite. They assume that every field sits on one continuous road. They assume that paradigm change is slow. They assume the best predictor of a good research proposal is past performance inside the current paradigm.
This works fine for incremental science. It actively kills revolutionary science.
Here’s the kicker. Kuhn didn’t predict gatekeeping would be a problem. He believed paradigms were necessary for normal science to function. But he would have recognized the modern review panel as an instrument for conserving orthodoxy — not for discovering novelty.
Peer Review’s Philosophical Blindspot
Ask any mid-career scientist about grant review, and you’ll hear variations of the same complaint: “The most innovative proposals get the worst scores.”
That statement gets treated as a bug, something a better scoring rubric could fix. It is, in fact, a feature. It is the inevitable product of putting reviewers inside one paradigm in charge of judging proposals that might establish another.
Think about what a grant review panel is actually doing. It evaluates proposals partly on feasibility, partly on novelty, but mostly on whether the expected results make sense within the existing state of the field. That is the definition of Kuhnian normal science. It should favor predictable, puzzle-solving extensions of the current framework.
The deep truth is that nobody can predict what a genuinely transformative project will deliver. If they could, scientists wouldn’t waste time writing grants. They’d write textbooks.
Feyerabend pushed this even further. In Against Method, he argued that science has no fixed method at all. The only principle that never inhibits progress is “anything goes.” For a man portrayed as an anarchist, Feyerabend made an astute and profoundly practical observation: history shows us that breakthroughs happen through whatever combination of stubbornness, accident and rule-breaking the situation demands.
Try putting “anything goes” onto an NSF FastLane submission form. The system simply cannot process it.
Basic Research versus Applied Science: A False Dilemma
Washington loves the phrase “basic research.” It conjures images of lone geniuses scribbling equations, pursuing knowledge for its own sake. On the other side you have “applied research,” the slightly grubby cousin that actually builds things.
Donald Stokes called this taxonomy nonsense in his book Pasteur’s Quadrant. His argument was elegant: why do we insist on separating fundamental understanding from practical use when most of the world’s most consequential research pursued both at the same time?
Pasteur wanted to understand the chemistry of fermentation. He also wanted to stop wine from spoiling. Darwin wanted to explain the diversity of life and, in the process, transform agriculture through selective breeding.
The basic/applied binary turns out to be an artifact of Bush’s 1945 policy narrative, not a law of nature. But US research priorities still organize around it.
Consider the Department of Energy. The Office of Science funds fundamental physics and chemistry. It also oversees the applied energy programs. They are separate directorates, separate budgets, separate cultures.
Scientists cross these boundaries anyway. When they do, they hit administrative walls. A solar cell researcher might have dual funding streams and dual reporting requirements. Why? Because the categories were invented by people who believed a man writing equations in one building had nothing to do with a man building a prototype in another.
The Positivist Legacy in Peer Review
Let’s dig down one more layer, because the philosophical underpinnings go deeper than Bush.
The US research system is built on a simplified and simplified version of logical positivism. Positivism, especially the Vienna Circle version, held that science is the accumulation of observable facts, that knowledge grows by verification and that the scientific method is essentially the same across all fields.
Policymakers absorbed this in diluted form. It shaped everything:
- Research proposals are evaluated partly on their “hypothesis-driven” character, as if all science proceeded by hypothesis testing.
- Observational, descriptive and exploratory work gets systematically devalued, because it seems less scientific.
- The biomedical world treats replication labs as second-class. The funding system rewards novelty over verification.
- The larger cultural assumption is that more raw data automatically equals more knowledge. This is not an empirical claim. It’s a metaphysical one, and the Stanford Encyclopedia of Philosophy entry on scientific method reminds us how contested the whole terrain is.
The scientific method, described as a single monolithic procedure, doesn’t exist. There is no algorithm that scientists follow. What exists is a grab-bag of practices: experiment, modeling, fieldwork, computation, taxonomy, luck, skepticism and the occasional leap of intuition.
Once you accept that, a lot of what passes for research policy starts to look strange.
The National Institutes of Health, for example, insists on hypothesis-driven research in most of its grant mechanisms. This biases the portfolio toward confirmatory studies and away from discovery-based science.
For decades, grant review boards have asked “what is your hypothesis?” as if every worthwhile investigation began with one. Some of the greatest biomedical discoveries — penicillin, the role of Helicobacter pylori in gastric ulcers, the structure of DNA — began as observations that broke out of the hypothesis-testing mold.
A hypothesis can be an excellent tool for certain kinds of research. It is not an universal scaffolding for all knowledge production. Treating it as such is philosophical dogma wearing a lab coat.
Values Are Already in the Room
There is another myth that needs dispatching: the idea that science is value-free.
Philosophers like Helen Longino have argued that scientific knowledge is socially constructed in ways that reflect the values of the scientific community. When those communities exclude certain perspectives, the knowledge they produce is not neutral. It is partial.
What does that mean for research priorities? Quite a bit.
When the US funds research on the biology of race while neglecting research on the social construction of racial categories, values are at work. When agencies pour billions into defense-related AI ethics while starving civilian AI infrastructure, values are at work. When NIH weighs individual-level cancer treatments against community-level public health interventions, values are at work.
None of these choices are purely technical. They are choices about what kind of world we want science to help build.
US research policy often pretends those choices are made downstream, in Congress, by the officials allocating broad budget categories. But they are actually being made upstream, by review panels, program officers and advisory committees who either recognize their value commitments or fail to see them entirely.
How the System Actually Fails
Want a quick diagnosis? The American research ecosystem systematically underfunds the very ideas that would upend both existing theories and existing priorities. It achieves this through a series of interlocking philosophical mismatches.
Consider four of them:
- Foresight Paradox: The less a proposal resembles current knowledge, the less capable reviewers are of evaluating it. But “novel” is one of the highest-ranking criteria in every grant call. Agencies essentially ask reviewers to predict the unpredictable.
- Collaboration Trap: High-profile funding programs push for convergence and multidisciplinary teams, yet hiring, tenure and promotion still depend on traditional disciplinary structures. Universities want researchers to collaborate but reward the lone investigator.
- Replication Neglect: Peer review celebrates new counterintuitive findings, while direct replications languish unfunded. Yet replications are the only way the community confirms a finding isn’t the product of measurement error or unconscious bias. The Replication Crisis in psychology shows what happens when the incentive system ignores this.
- Data-First Fallacy: Agencies push open data and data sharing as if total data volume were equivalent to understanding. Philosophers have long noted that raw data cannot make claims. Theoretical frameworks must interpret data. The infrastructure to generate data is expensive; the theoretical capacity to make sense of it is treated as an afterthought.
Each of these mismatches trace back to a philosophy of science, one that assumes knowledge accumulates like compound interest. Real science, like real innovation, moves in zigzags, dead ends and full reversals.
How many research administrators does it take to change a lightbulb? None. They’ve already replaced the lightbulb program with one that funds nine different types of illumination but none of the sockets.
What A Philosophical Fix Would Look Like
So — are we doomed to oscillate between boring incrementalism and the occasional catastrophic funding shift? Not necessarily. But fixing the system requires that we make the implicit philosophy explicit and then update it.
This isn’t a call to abolish peer review or to grant everyone equal funding. It’s a demand for intellectual honesty. The reforms below follow naturally from what philosophers of science have been saying for decades:
Stop pretending research will “pay off” predictably. Every nation that funds basic curiosity-driven science accepts that the return on investment happens on timescales of decades, or never. Agencies should stop evaluating proposals using the improbable calculus of short-term ROI and instead support a wide portfolio with rough guidance from previous performance.
Restructure proposals around research practice, not research fantasy. Great scientists often do not know exactly what they will find. Let them describe their approach without forcing them to construct a sterile hypothesis. Fund observational, qualitative and modeling-based science on equal footing with hypothesis-driven confirmatory research in areas where those approaches are methodologically appropriate.
Reward the anomalies. A cultural shift needs to happen inside review panels. The proposal that challenges an established belief should get extra scrutiny and be given the benefit of the doubt. Reviewers need training in epistemic humility. They should ask not only “is this feasible?” but “what would happen if it were true?”
Bring the sociology of science into science policy. Institutions like the Society for Social Studies of Science have generated decades of research on how laboratories actually work, how discoveries actually happen and how funding regimes shape the questions scientists ask. This work shouldn’t be merely cited in a footnote of an anthropology journal. It should be read by everyone designing new research programs.
Adopt Pasteur’s quadrant as the default. Instead of asking whether a proposal is basic or applied, ask whether it has potential to advance fundamental understanding and whether it could solve a pressing practical problem. Both axes matter. Funding both extremes is far less valuable than funding research at the intersection.
Let successful program managers take bigger risks. The talent of a good program officer lies in recognizing people with good taste who are asking genuinely promising questions. These people are the venture capitalists of the scientific funding world. Give them more discretion, trace their portfolios and normalize the understanding that several of their investments will fail.
An Office of Philosophical Audit?
Here is a bizarre proposal that might actually work. Every government science agency should hire a small team of philosophers of science, not as compliance officers but as institutional gadflies.
Their job would be to ask uncomfortable questions. Which fields are we underfunding because they lack the right buzzwords? Are we funding the infrastructure needed to understand the data we collect? Do our grant criteria actually match the way research is performed in this field? Why are we setting priorities according to one model of science when other robust models are available?
This is not national science policy in the hands of professional skeptics. It is introducing epistemic pluralism into bureaucratic practice.
Agencies like the NIH fund many studies of biomedical ethics. They fund almost none on the philosophy of the research they support. That’s a strange blind spot.
The Political Economy of the Inside
I should be clear about why none of this gets fixed easily. The harder truth is that the US research system is not designed for science. It’s designed to make science predictable enough to fit into a congressional appropriations cycle.
Congress wants to tell constituents that money produced a specific thing — a vaccine, a chip, a planetary probe. The “Endless Frontier” model promised that basic science would eventually deliver those goods while never asking the public to get too close to the messy process through which science works.
When you dig into history, you find that every period of scientific and technological flourishing — the postwar Manhattan Project, the Apollo program, the biotech boom — involved a messy, urgent, problem-focused push rather than a serene flow from basic science to application.
What if we judged US research priorities by whether they advance a more just, resilient and knowledgeable society, rather than whether they maximize that vague and easily gamed metric of market share and economic dominance?
Philosophers of science call this the “proper aim of inquiry.” A society whose research establishment thinks carefully about that aim will make better priorities.
The Secret Is Not a Secret
Here’s where we land. There is no one scientific method, no pure basic/applied boundary and no way to bring “transformative” research through a review system designed to reproduce the status quo.
The secret of fixing US research priorities is not investing in AI, quantum computing, or any of the other trendy fields that will dominate headlines for the next decade.
The secret is building an institution capable of epistemic humility. That means acknowledging the limits of prediction. It means funding a wide range of research styles, some of which will look foolish until they don’t. It means treating science as what it is: a genuinely collective, contested, human process, conducted sometimes by loners but overwhelmingly by communities of practice with their own values, blind spots, and unexamined commitments.
Policymakers do not want to hear this. The prevailing take is that America is at risk of falling behind China, and that what we need is more money, faster, with clearer metrics and no time for navel-gazing philosophy.
But that urgency is precisely why the philosophy matters. A country that pours $200 billion a year into research without ever asking what counts as scientific knowledge will keep producing more incremental findings,