Primer

People, Not Projects: Funding Researchers Instead of Research Plans

David Hilmer Rex · July 2026

Summary

Most competitive research funding is awarded to projects: a specific plan, proposed in advance, judged on its merits, and expected to be carried out roughly as written. People-not-projects funding inverts this. It selects a researcher, gives them substantial resources for a long and renewable period, and lets them decide what to work on and when to change direction. The canonical example is the Howard Hughes Medical Institute, whose stated approach is to fund "people, not projects." The best evidence, a matched comparison of HHMI investigators with comparably accomplished project-funded peers, finds that people-based funding produces both more highly cited work and more failures, consistent with longer horizons and a higher tolerance for risk. This primer sets out what the mechanism is, the problem it responds to, the evidence, its design dimensions, the main instruments including the Danish ones, and its limits.

The idea ↑

The difference is in the unit of selection. In project funding, the unit is the plan: a proposal describes a hypothesis, a method, a timeline, and expected outcomes, and reviewers assess whether that plan is worth funding. In people funding, the unit is the person: a funder selects a researcher, commits resources to them for a period of years, and leaves the choice of what to pursue largely to the researcher. The grant follows the person, not a pre-specified deliverable.

The bet behind people funding is that the most valuable research is often the research nobody could have specified in advance. If discovery is genuinely uncertain, then selecting plans rewards the plans that look most feasible on paper, which are rarely the most consequential. Selecting talented people and giving them freedom and time, the argument goes, lets the science follow where it leads. The Howard Hughes Medical Institute has made this its signature: it appoints investigators rather than funding grants, on the principle of backing "people, not projects."¹

The problem with project grants ↑

Project funding is the dominant model in public science, and its documented pathologies are the reason people funding is proposed as an alternative.²

A project proposal asks a researcher to specify in advance what they expect to find, which sits awkwardly with the nature of discovery. To be fundable, a proposal usually needs preliminary data showing the idea already works, so the money often arrives after the risky part is done. Reviewers reward feasibility and penalise uncertainty, which pushes proposals toward the incremental and the safe. Cycles are short, commonly three to five years, and the pressure to produce publishable results before the next application discourages long, exploratory, or tool-building work whose payoff lies further out. Application burden is high and success rates are low, so a large share of researcher time goes to writing and reviewing proposals rather than doing research. And once a project is funded, the researcher is expected to do what they proposed, which discourages pivoting when the science changes underneath them.

People funding is a direct response to each of these: no plan to lock into, a long horizon that tolerates slow-maturing work, and enough security that the researcher is not perpetually writing the next application.

The evidence: HHMI versus NIH ↑

The cleanest evidence on people funding comes from a comparison of the two dominant American models for life-sciences funding: the Howard Hughes Medical Institute, which funds people on long, forgiving terms, and the National Institutes of Health, whose R01 grant funds projects on shorter, renewal-dependent terms.

Azoulay, Graff Zivin and Manso (2011) compared HHMI investigators with a matched control group of similarly accomplished NIH-funded scientists, using the timing of HHMI appointment in a difference-in-differences design.³ After appointment, HHMI investigators produced more papers in the top percentiles of the citation distribution, and more work that introduced new keywords and new directions, a marker of novelty. They also produced more papers that went on to be poorly cited or ignored. The pattern is higher variance in both directions: more hits and more misses. That is the signature of exploration. Researchers freed from short-cycle, renewal-dependent funding took more risks, and the portfolio of outcomes reflects it.

The operative mechanism is not simply more money or more time. It is the combination of a long horizon with a review process that tolerates early failure and rewards long-run success. HHMI review is retrospective and lenient about early setbacks, which is precisely the incentive structure that theory predicts will motivate exploration rather than exploitation.⁴ A long grant with a harsh mid-term review would not produce the same behaviour.

Two caveats matter. HHMI selects people who are already exceptional, so the results speak to what freedom does for strong researchers, not to whether people funding would work for everyone. And the evidence base beyond this comparison is thin: people funding is intuitively attractive and has one strong study behind it, rather than a large body of replicated findings.

Design dimensions ↑

People funding is not one thing. The instruments that use it differ along several dimensions, and the differences change what the mechanism actually does.

• Basis of selection. Proven track record versus demonstrated potential. Programmes aimed at established researchers select on accomplishment; programmes aimed at early-career researchers must bet on promise, which is harder to judge and higher variance.
• Duration and renewal. Five to seven years is common, often renewable. The review at renewal is the key lever: a retrospective, failure-tolerant review sustains risk-taking, while a prospective, milestone-based review recreates the pressures of project funding.
• Size. The award has to be generous enough to free the person from chasing other grants. Below that threshold, people funding becomes a supplement rather than a substitute, and the behavioural effect weakens.
• Degree of freedom. Full discretion over what to work on, or freedom within a broad theme. More constraint moves the instrument back toward projects.
• Portability. Whether the funding follows the person across institutions, which strengthens the researcher's hand and the "people" logic.
• Portfolio position. How many people are funded this way, and how concentrated the awards are. People funding is powerful as part of a diverse portfolio and risky as a monoculture.

The main instruments ↑

The Danish and Nordic instruments ↑

Denmark runs some of the largest people-based research grants in Europe, and it is not a coincidence that they come from the enterprise foundations. People funding needs exactly what those foundations are structurally built to supply: long horizons and insulation from annual budget cycles.⁵

The Novo Nordisk Foundation Laureate Research Grants fund outstanding researchers who relocate to or remain in Denmark, with large budgets over seven years, aimed at the person and their long-term programme rather than a fixed project. The Villum Investigator programme, launched in 2017, awards leading researchers in the technical and natural sciences substantial multi-year grants, up to around DKK 40 million over six years, with wide latitude over direction. On the public side, the Danish Council for Independent Research's Sapere Aude research leader grants back individuals judged to be exceptional early-career talents, a smaller and more selection-on-potential version of the same logic.

The connection to the enterprise foundation model is the important point for anyone thinking about the Danish system. A funder that must justify its spending on a short political cycle finds it hard to commit to a person for a decade with no specified deliverable. A self-owning foundation with patient, insulated capital can. Denmark's unusual capacity to run people funding at scale is downstream of its unusual ownership structure for research money.

Critiques and open questions ↑

The Matthew effect. Because people funding selects on the person, and past success is the most legible signal, it tends to concentrate resources on already-established researchers. This can entrench advantage and make the mechanism poorly suited to early-career or unconventional researchers, precisely the groups whose potential is hardest to read from a track record.

Selection is genuinely hard. Predicting which people will do transformative work is difficult, and a programme that selects on past accomplishment may simply be rewarding yesterday's success rather than identifying tomorrow's. The HHMI evidence tells us what freedom does for people who were selected well; it does not tell us how to select well.

Equity and concentration. Large person-based awards can narrow the distribution of who gets funded and reinforce existing hierarchies of institution, field, and background. A system weighted heavily toward people funding risks elite capture.

Accountability. With no specified deliverables, people funding is harder to justify to funders, boards, and the public, and harder to evaluate. The freedom that makes it work is also what makes it politically and administratively demanding to sustain.

It is one instrument, not a system. The strongest case for people funding is as one component of a diverse funding portfolio, sitting alongside project grants, fast grants, prizes, and mission-oriented mechanisms, each suited to different kinds of work. A research system that funded only people would concentrate risk in selection and lose the coverage that a plurality of mechanisms provides. The value of people funding is what it adds to the mix, not what it would do as the whole.