Insight, Agriculture & Method
The Commission's new agriculture and food research strategy says previous work needs stronger links to deployment and uptake. That is correct, and it implies something the strategy does not say: what happens after the pilot is a research question, not a communications budget.
The European Commission has adopted a new research and innovation strategy for agriculture, forestry, rural areas and food systems (as reported), spanning soil and agroecosystems, bioeconomy, forests, rural value chains, artificial intelligence and robotics, circular food systems, consumer behaviour and aquatic food.
The most important sentence in it is not about any of those. It is the acknowledgement that previous research needs stronger connections to deployment, market uptake, user needs, monitoring, governance and public trust.
That is correct and overdue. It also carries an implication the strategy stops short of drawing, and the implication is the whole point: if uptake is the problem, then uptake is the research question, and it cannot be handled by a work package called dissemination attached to the end of a project.
There is a field that already learned this, expensively, and it is worth looking at what it found.
For thirty years, improved cookstoves were among the most confidently promoted technologies in development. The engineering was sound. Laboratory tests showed reduced indoor smoke and lower fuel use. The health case was overwhelming, given how many people cook on solid fuels indoors. Programmes distributed millions of stoves.
Then Rema Hanna, Esther Duflo and Michael Greenstone ran a randomised trial in rural Orissa and followed households for up to four years. The results, published as Up in Smoke, are among the most sobering in the literature.
They found a meaningful reduction in smoke inhalation in the first year and no effect over longer horizons. No improvement in lung functioning. No improvement in health. And no change in fuel consumption, which meant no emissions benefit either.
The stove was not defective. The explanation was behavioural and it is stated plainly in the study: households failed to use the stoves regularly or appropriately, did not make the investments needed to maintain them, and use declined over time.
Read that against the laboratory result. Both are true. The stove reduces smoke when it is lit, used correctly and maintained. In the fourth year, in a real house, with a real repair problem and a competing three-stone fire that has never failed, it is not.
The field responded, eventually, by building a research capability around exactly that gap. Later work has tested combinations of upgraded supply and demand promotion in field experiments rather than assuming that a better product diffuses on its own, and there is now an explicit literature on implementation science for clean cooking, cited within it.
Thirty years and a great deal of wasted money bought one finding: the difficulty is not the device.
Medicine reached the same conclusion earlier and gave it a name. Publishing a trial result changes nothing by itself. Whether a proven treatment reaches patients depends on clinician workflow, procurement, training, incentives, records systems and organisational politics, and none of those are addressed by the trial that proved the treatment works.
So a discipline grew up around the second question, with its own methods, outcome measures and journals. It studies fidelity, reach, adoption, sustainment and cost of implementation as objects in their own right, on the understanding that a treatment delivered at half fidelity to a quarter of the eligible population is a different intervention from the one that was tested.
Agricultural and energy research has broadly not done this. It has demonstration projects, which prove feasibility under favourable conditions, and it has dissemination, which is communication. It does not, generally, have a research function whose subject is what happens when a technology meets an ordinary user with an ordinary budget and an existing way of doing things that works well enough.
That is the gap the Commission's strategy has identified. Naming it is progress. Filling it requires treating it as science rather than outreach, and that means it has to be designed in at the start, funded on its own terms, and reported whether or not the answer is flattering.
Four concrete changes, none of which requires new instruments.
Fund the post-pilot period, not the pilot. The most valuable observations in the cookstove literature came from years two, three and four, after the project would normally have closed. A three-year Horizon project that ends when deployment begins has stopped measuring at the point the interesting variation starts. Even a small extension of monitoring, decoupled from the delivery consortium, would change what is learnable.
Measure use, not distribution. Stoves distributed is an output. Stoves lit is an outcome, and the difference between them was the entire finding. In agriculture the equivalents are hectares under a practice versus hectares still under it in season four, and sensors installed versus sensors read by somebody who acted on the reading.
Report abandonment as a result. A farmer who tried a practice and stopped holds the most informative account available, and no current reporting template asks for it. This is the same selection problem that makes a set of successful case studies an unreliable guide to what will happen next.
Separate the evaluator from the deliverer. A consortium reporting on its own uptake is being asked to hold two positions. The cookstove finding exists because a research team with no stake in the distribution programme measured it independently, four years later, and was free to publish a null result.
That last point is where most of the difficulty sits, and it is organisational rather than intellectual.
There is a version of this argument that European research programmes make politely and then do not act on, which is that comparison with other regions adds value.
The specific value here is unusually concrete. The adoption literature Europe now needs already exists, built over four decades on smallholder agriculture and household energy in Asia and Africa, with randomised evidence, long follow-up and a hard-won understanding that the binding constraints are credit, risk, labour, tenure, information and the quality of the incumbent practice rather than awareness.
We made the same observation about European drought adaptation research, which is reassembling a constraint list published in 1985 about farmers in developing countries.
The productive comparison is not to import findings, because European farms are not smallholdings and the institutional settings differ enormously. It is to import the method: long follow-up, measurement of use rather than receipt, deliberate sampling of non-adopters and abandoners, and a willingness to publish a null result about a technology the funder paid to develop.
That is a research capability rather than a finding, and it is transferable in a way that findings are not.
Continuation at three and five years, for a sample of technologies developed under previous framework programmes. This is retrospective, it is cheap, and it would establish the base rate that every future project is implicitly assuming.
The share of project budgets spent on understanding use rather than producing the technology. If it is under five per cent, the strategy's stated ambition and its resourcing are not aligned, and that is checkable from records that already exist.
What the incumbent practice was, in each case. Whether projects specified it at all is a good proxy for whether adoption was taken seriously at design stage.
Who was asked. Adopters only, or adopters and the people who declined and abandoned. The second group is where the finding is, and they are harder to recruit precisely because nobody has a list of them.
The Commission has said the right thing about the gap between research and uptake. Whether the strategy changes anything depends on a narrower question: whether uptake gets a method, a budget line and an independent evaluator, or a communications plan.
The Lab works on this in regenerative agriculture and through measuring change designed around what happened after the project closed, with field research among the people who stopped.
If you are building a work package on deployment and uptake and want it specified as method rather than dissemination, tell us what you need to know.
This is an independent insight piece by Transitions Lab. For the Lab's applied work, see Regenerative Agriculture. See also Europe Has Enough Demonstrations on why a pilot cannot tell you what will scale, The Trough Before the Dividend on why a set of successful cases is a biased guide, and The Benefits Nobody Was Looking For on outcomes no results framework asked for. To discuss a study, see Contact.