Funding high-risk research

What changes when the funding rules change?

By Anete Vingre (Technopolis)

What kinds of research does dedicated funding for high-risk, high-reward research actually support? Are grant lotteries legitimate? What happens when research funders treat failure not as a flaw, but as part of the model?

These questions are often debated in principle. Technopolis recently worked with the Austrian Science Fund (FWF) to evaluate its 1000 Ideas programme – a dedicated funding programme for high-risk, high-reward (HRHR) research with an unusual design. Applications are anonymised and first undergo an initial plausibility screening by the FWF Scientific Board, which assesses whether proposals meet the programme’s criteria for high-risk, high-reward research and scientific plausibility. Shortlisted proposals are then assessed by an international jury, which identifies the pool of fundable proposals from which the final funding decisions are made through partial randomisation. Acceptance of failure is clearly signalled in advance to applicants and awardees. What makes this evaluation interesting is that the programme has now been running long enough (since 2019) for us to examine what happened in practice and provide answers based on empirical data to some of these interesting questions.

Beyond perceptions: what does the lottery actually change?

Much of the discussion around lottery funding rightly focuses on perceptions – do researchers think it is fair? Is it legitimate? There is also evidence on how the use of lottery changes the demographics of applicants. That matters, but it does not tell us whether introducing a lottery actually changes what gets funded.

We wanted to answer that question. To do so, we combined surveys and interviews with bibliometric and econometric analyses of funded and unfunded applicants, and comparison with the FWF’s conventional benchmark funding programme. That meant looking not only at what people said about the programme, but also at publication profiles, interdisciplinarity, citation performance, follow-on funding, and something more unusual – bibliometric outcome variance.

Our econometric analyses suggest that the lottery element changes what gets funded. Compared with conventional project funding, introducing partial randomisation reduces the extent to which funding decisions align with traditional indicators such as applicants’ previous citation performance. At the same time, characteristics associated with more exploratory research, particularly bibliometric interdisciplinarity and novelty indicators (for example, publications that combine ideas from previously unconnected scientific fields or introduce unusually new combinations of knowledge), become more strongly associated with funding success.

This doesn’t mean that the lottery replaces scientific judgement. Rather, scientific assessment first establishes which proposals meet the required quality threshold. Randomisation then reduces reliance on fine-grained ranking among proposals that experts themselves consider difficult to distinguish. The result appears to be a portfolio that better reflects the programme’s objective of supporting genuinely unconventional research ideas and proposals with uncertain outcomes.

Looking at the outcomes produced by the programme awardees, we found greater variance in bibliometric outcomes. Compared with conventional FWF funding instruments, projects funded through 1000 Ideas showed significantly greater variation in citation outcomes. That may sound technical, but conceptually it is important. High-risk, high-reward funding should not simply produce better average outcomes; it should produce a broader spread of outcomes, including projects that fail, projects that redirect, and projects that prove exceptionally successful. In that sense, the programme appears to be selecting not simply for excellence, but for productive uncertainty.

As for perceptions, the lottery itself is probably the most provocative part of the programme. But most awardees understood it for what it was - not a replacement for scientific assessment, but a mechanism used after proposals had been judged fundable. Many interviewees argued that in highly competitive funding systems, there is often “hidden randomness” anyway, arising from small differences between reviewers, panel dynamics, or strategic prioritisation. The lottery simply makes that uncertainty explicit.

When novelty and failure go hand in hand

Researchers repeatedly described their projects as “hop-or-drop” science - ideas where it was genuinely unclear whether the methods would work, whether the underlying hypotheses were correct, or even whether the original research questions could ultimately be answered.

Our interviews showed that novelty emerged in several different ways (see figure). For some projects, it involved developing new methods or technologies, for others, it meant reframing established scientific questions or bringing together disciplines that rarely interact. Many projects also explored previously unstudied phenomena or unusual empirical cases. Rather than following a predictable pathway towards predefined results, novelty often emerged through exploration itself.

This perception of uncertainty was shared by applicants themselves. In the survey, a large majority of applicants considered it likely that their proposed research might not succeed as originally envisaged, suggesting that the programme effectively communicates its expectation that high-risk research entails a genuine possibility of failure. Among awardees, around one-quarter subsequently reported that their projects had indeed failed to achieve their original objectives. Rather than indicating a weakness of the programme, these findings suggest that it is succeeding in attracting and supporting research where uncertainty is real, not merely rhetorical.

But failure, in this context, rarely looked like a waste of time. Instead, we saw what might be called productive failure - failed experiments that nevertheless generated new methods; abandoned pathways that led to new collaborations; and null results that reshaped subsequent grant proposals.

This matters because most monitoring systems don’t see these things. They tend to count publications and citations. They rarely count redirection, learning, or methodological dead ends that prove scientifically useful. That creates a mismatch between how high-risk research works and how it is evaluated.

One concern often raised about high-risk research funding is whether projects that don’t achieve their original objectives leave researchers, especially early-career researchers, with little to publish and with publishers unwilling to accept null results. This is often considered one of the career risks associated with pursuing more uncertain research.

Interestingly, we found little evidence of this in the 1000 Ideas programme.¹ Most awardees did not report difficulties publishing results, even where projects had substantially changed direction or the original hypotheses proved incorrect. Instead, researchers described publishing methodological advances, unexpected findings, or insights that emerged from following alternative research pathways.

Whether this reflects broader changes in scientific publishing, greater acceptance of negative and null results, or simply the particular level and type of risk supported by the programme remains an open question. Nevertheless, the findings suggest that productive failure may be more publishable than is often assumed.

Should this be mainstream?

One of the evaluation questions was whether 1000 Ideas should remain a stand-alone programme, or whether its features should be integrated into conventional schemes. The context is important: 1000 Ideas provides relatively small, short-term seed grants (€50,000–175,000 for up to two years), whereas FWF’s mainstream Principal Investigator projects typically provide substantially larger, multi-year awards intended to support full research programmes rather than exploratory proof-of-concept work.

The dedicated scheme matters. It creates a protected space where uncertainty, unconventionality, and even failure are legitimate. That is hard to replicate inside mainstream funding programmes.

But some elements, especially anonymised first-stage review or transparent randomisation among equally ranked proposals, may have wider relevance. Not for every grant scheme, but perhaps in places where we pretend certainty is greater than it really is.

Research funders spend a lot of time trying to eliminate uncertainty from decision-making. But if the goal is to support transformative ideas, some uncertainty is not a problem to solve.

1. The programme attracted a substantial proportion of early- and mid-career researchers, suggesting that the possibility of failure did not discourage applications from researchers at earlier career stages. However, the evaluation was not designed to compare researchers' willingness to pursue high-risk research across career stages or with applicants to conventional funding programmes.