The room where a research programme dies
You submitted the final report fourteen months ago. The results were negative: the intervention didn't work, the compound didn't bind, the model didn't predict. The work was careful, the methodology sound, the write-up honest. And now, without ceremony, the programme has been quietly removed from the council's forward portfolio. No renewal panel. No resubmission window.
Just gone.
This is not an accident of bureaucracy. It is the predictable output of a governance structure that was never designed to value what didn't happen.
Scientific funding councils, whether national research councils, medical research charities, or the grant-making arms of government agencies, share a broadly similar internal architecture. At the top sits a governing board populated by senior scientists, lay members, and sometimes civil servants. Below that, programme-level committees allocate resources across fields. Below those, peer-review panels score individual applications. The question of which null results get permanently defunded is answered not by any single villain in the system but by how information flows, and fails to flow, across those three layers.
The scoring rubric that already decided the answer
Most funding panels score applications on a standard matrix: scientific quality, feasibility, novelty, and impact. The weights differ by council, but impact and novelty together almost always account for more than half the total score. A programme proposing to replicate a failed finding, or to systematically map the boundaries of an effect that turned out not to exist, scores poorly on novelty by definition. It often scores poorly on impact too, because panels conflate impact with the probability of a positive, publishable, citable result. That conflation is not a misunderstanding. It is a choice, and it is the wrong one.
Consider what this looks like in practice. Two research teams, call them the Okafor group and the Lindqvist group, both submit renewal applications after four-year programmes. The Okafor group found that a behavioural intervention modestly reduced hospital readmission rates in a specific elderly population. The Lindqvist group ran a larger, more rigorous trial of a similar intervention across a broader population and found no effect. Both programmes cost roughly the same. The Okafor renewal scores highly on impact: there is a positive finding to build on, clinical partners are enthusiastic, a second phase has obvious momentum. The Lindqvist renewal must argue that mapping why the effect disappeared matters, that the null result changes the evidentiary landscape for a whole class of interventions. That argument requires a panel to think consequentially about what isn't there. Most panels, pressed for time and operating under implicit norms that treat publications as proxies for productivity, do not.
The Lindqvist group's programme gets ranked below the funding line. The decision is permanent because councils, facing flat or declining budgets, rarely revisit ranked lists once a funding round closes.
Where governance structure amplifies the bias
The permanence of defunding is a governance choice, not a scientific necessity. Several structural features lock it in, and they compound each other like interest on a debt nobody admits to carrying.
Programme committees set portfolio priorities before individual panels meet. Those priorities are drafted from strategic plans that emphasise areas of expected advance, not areas of established failure. A null result, by its nature, belongs to the second category. When a programme committee decides that a research area is mature or that a particular therapeutic target has been adequately explored, they are often reading that decision off a pattern of null results, but the conclusion they draw is to reduce investment, not to fund the work of understanding those nulls.
The information that flows upward from panels to committees is heavily summarised. A panel might score fifty applications in two days. The committee receives ranked lists, brief summaries, and the scores. The texture of the deliberation, the argument that the Lindqvist null result is more informative than three positive results from smaller, underpowered studies, does not survive the compression. Committees act on numbers.
Most councils have no formal mechanism for categorising null results differently from inconclusive or failed work. A programme that ran poorly, with attrition, missing data, and compromised blinding, produces a null result. So does a programme that ran flawlessly and found nothing. Governance documents at most major funding bodies treat these as the same outcome class. The Wellcome Trust, the NIH study sections, the UKRI councils: all have published frameworks for research quality, but none routinely distinguishes high-quality null results as a protected category at the panel-scoring stage. Some funders have experimented with registered reports and pre-registered protocols, which commit to funding regardless of outcome, but these remain a small fraction of total grant volume.
Then there is the shadow of the programme officer. At many councils, programme officers, the staff scientists who manage portfolios and advise panels, carry significant informal influence. A programme officer who believes an area is exhausted will signal that to panel members in pre-meeting briefings. Those signals are not recorded in the minutes. They shape scores without appearing in the governance trail.
What a well-designed system would actually do differently
The failure described here is not inevitable, and the distinction between a fixable design problem and an immovable cultural one matters enormously for anyone trying to change it.
A funding council genuinely committed to valuing null results would need to change governance at each layer. At the panel level, scoring rubrics would need a dedicated criterion for evidentiary value, distinct from novelty, that rewards work which resolves uncertainty regardless of direction. That criterion would need to carry enough weight to survive averaging with impact scores. Ask yourself: if a trial definitively rules out an entire class of interventions, saving every subsequent researcher from pursuing a dead end, is that not worth more to science than a modest positive finding that three later labs will fail to replicate?
At the committee level, programme officers would need to present null results to portfolio reviews not as closures but as inputs: what does the Lindqvist finding rule out, and how does that redirect where the remaining budget should go? This requires committees to think about the shape of ignorance in a field, not just the frontier of knowledge.
At the governing board level, annual reports should disaggregate outcomes. How many programmes produced null results? Of those, how many received renewal funding? If the ratio is consistently near zero, that is a governance signal, not a scientific one.
Some councils have moved toward open data mandates and negative results repositories, which is useful but insufficient. Archiving a null result is not the same as funding the work that explains it.
The deeper problem is that funding councils are political institutions as much as scientific ones. They answer to governments, to donors, to public expectations of progress. A council that visibly funds programmes finding nothing, even when those findings are genuinely valuable, is exposed to the accusation of wasting money. That political pressure is real, and it flows downward through every layer of governance until it reaches a panel member who would privately agree that the Lindqvist result matters but will not spend political capital defending it in a room full of people who want to fund the next promising thing.
The null result doesn't lose because science failed. It loses because the governance was never built to win it anything, and the people running that governance have, so far, found it more comfortable to leave things that way.