For 80 years, American science has operated under the Vannevar Bush model that was established in the wake of World War II. In his 1945 report, Science, the Endless Frontier, Bush argued that the government should fund “curiosity-driven basic research,” with panels of peer scientists choosing which projects deserved financial support. His goal was not merely to subsidize scientists’ intellectual whims without regard for visible applications, but to secure American technological supremacy by developing fundamental knowledge before its practical uses were apparent. In practice, however, Bush’s vision mutated into a bloated, risk-averse bureaucracy split between government and universities, stifling the very innovation it was designed to foster.

After eight decades, the weaknesses of Bush’s project-grant model are clear. As the White House Office of Science and Technology Policy (OSTP) points out in its new report, academic researchers spend up to half their time writing, formatting, and defending grant proposals rather than actually doing science. Moreover, because review panels naturally gravitate toward proven concepts, researchers are incentivized to propose incremental work that is guaranteed to succeed, often because they have already accomplished it.

The human cost of Bush’s model is staggering. The average age of a researcher receiving his first major independent grant from the National Institutes of Health has crept into the mid-40s—as the report notes, that is older than the average age of recent Nobel Prize winners when they made their breakthrough contributions. We are forcing the brightest minds of academic science to spend their most creative decades acting as junior collaborators to insiders. The result is steady incremental progress, at least according to peers, but none of the breakthroughs that define true scientific revolutions.

Indeed, the OSTP is arguably much too kind to Vannevar Bush’s scheme: as the careful reader of the report may realize, the technological and scientific progress that powered the American century was produced almost entirely by alternative models based on missions set by the political leadership and their politically accountable deputies.

I therefore propose a fundamental change in how we cultivate intellectual capital. We should largely zero out the current project-grant system for basic research and replace it with a radically simpler philosophy for supporting unguided curiosity: fund people, not projects.

Although the recent OSTP report accurately identifies the bureaucratic rot and systemic sclerosis within our current grant-making framework, we must extend its logic to its necessary conclusion. The report correctly asserts that federal science funding must be politically accountable; in a democracy, the public’s resources must be directed toward the public’s interests by those the people actually elect.

To realize this ambition, Congress should circumvent this ossified establishment with a bold “metascience” pilot: 535 ten-year research fellowships—one formally appointed by every member of the House and Senate. This gives lawmakers direct buy-in. It also dismantles the peer-review cartel, replacing the sterile consensus of risk-averse panels with the robust, structural accountability that the OSTP now recognizes as essential.

These fellowships would be awarded sequentially over a five-year rollout period, amounting to 107 new fellowships per year. Each selected fellow would receive a competitive, stable salary alongside modest lab support—say, capped at 100% of their salary—and would be placed with program help at an institution after winning, much the way Rhodes Scholarship winners are placed at an Oxford college. This funding intentionally excludes substantial infrastructure, which would be funded through mission- or goal-based schemes.

The ten-year horizon of these fellowships is essential. Under the current three-year grant cycle, scientists must immediately begin worrying about their next funding application the moment their current one is approved. A decade of guaranteed support provides the freedom to take on the biggest and most fundamental problems and pursue unexpected insights.

The Howard Hughes Medical Institute, which funds researchers with grants of $10 million over seven years, bets on people rather than projects. It has produced a higher proportion of innovative, high-impact research by a wider margin than the project-based approach. Similarly, the OSTP report notes the long-running NSF Graduate Research Fellowship Program, which funds students independently of their advisors and institutions, has produced over 40 Nobel laureates. (A fair metascientific assessment would require taking into account that it also funded me!)

The selection mechanism must bypass the ossified academic science establishment. Recent Ph.D. graduates (or equivalent independent researchers) would submit their theses and publications, along with a brief vision statement, directly to a joint House-Senate committee. Committee staff, drawing freely on outside experts and advanced analytical tools, would rank the candidates. To guide this process, the committee would set rotating annual priorities.

The actual appointments would be assigned by a randomized lottery tied to congressional seats. When a member’s turn arises, he can either select a qualified applicant he believes in, or he can accept the top staff-recommended name from the staff-ranked list for that year.

While most members would likely defer to the staff ranking, letting them pick fellows if they want avoids inbreeding, gives politicians buy-in, and replicates the highly successful system of congressional nominations to military service academies. For the sake of scope and efficiency, this fellowship scheme centralizes the ranking of candidates at the committee level.

High-capital “big science” fields would continue to be funded directly by their institutional customers, such as mission agencies, the Department of Defense, or private industry. Small goal- or mission-oriented projects would be funded through a wide variety of models, along the many lines proposed in the OSTP report. The congressional fellowships are exclusively for the independent thinkers with the greatest apparent potential—the future Einsteins.

By funding researchers directly with stable, modest support, we can recreate the fertile conditions that historically produced revolutionary advances.

I estimate the cost of this pilot program during its rollout phase at roughly $35 to $40 million per year for the 107 new fellows, with modest administrative overhead. In the context of a federal research budget in the tens of billions, this is a tiny fraction of what we currently spend on the peers’ preferred marginal projects and their administration.

After five years, once all 535 fellowships are awarded, Congress can assess the program, as it should assess all research funding programs, before deciding whether to renew, modify, or expand it.

It is time to move beyond the bureaucratic legacy of the Bush model of peer-review and project-based funding. The congressional fellowships would identify the most promising researchers of their generation and give them the freedom and time to think instead of writing grants. To advance Vannevar Bush’s Endless Frontier, we need simpler, more humane, and more exciting approaches to American innovation.