knowledge that will not cross the fence

Problem statement

Farmer field schools (FFS) — season-long, field-based groups of 20–25 farmers who run their own experiments under a facilitator — are one of the most widely funded agricultural-extension models in low- and middle-income countries: introduced in Indonesia in the late 1980s and expanded chiefly by FAO and IFAD, they have reportedly trained an estimated 12 million farmers in over 90 countries. Their theory of change has always rested on a second step: trained farmers would spread what they learned to the much larger number of neighbours who never attended, making a costly, small-group method scalable. The 3ie/Campbell systematic review (92 impact evaluations, 15 of adequate quality, plus 20 qualitative studies) found that this second step does not happen: "Neighbouring farmers who do not participate in FFS projects do not benefit from diffusion of knowledge about IPM from trained farmers," with "no improvement in IPM knowledge among neighbouring, non-participating farmers," no change in their pesticide use, yields or income — "true for both kinds of projects: those that supported diffusion through processes, such as community institutionalisation in India and Pakistan and training of farmer trainers in Indonesia and China, or those that left diffusion to happen by word of mouth and observation, as in Nicaragua." The unsolved problem is that FFS knowledge is experiential — built by watching insects, counting predators and comparing plots over a season — and, in the review's words, "even where there is high awareness of IPM among non-participants, it is difficult to convey through verbal communication."

Why this matters

Participants clearly gain: in China, the Philippines and Pakistan FFS graduates "used 23 per cent less pesticide than neighbouring non-participants," with yield and income gains in pilots. But an extension model that only reaches those who sit in the school for a season cannot reach hundreds of millions of smallholders, and the review found that "for larger FFS programmes implemented at national scale over longer periods there is no evidence of positive effects" — the only two national IPM programmes evaluated (Indonesia and Viet Nam) showed no significant impact "because adoption was not sustained." Diffusion failure therefore sits at the root of a decades-long argument about whether FFS are worth their cost, and it matters beyond agriculture: any programme whose scaling logic depends on peer transmission of tacit, practice-based knowledge (community health workers, water-user groups, cooking-practice change) inherits the same weakness. Where any short-run neighbour adoption did occur, "any initial adoption among neighbouring farmers in the short term falls considerably over time."

What’s been tried and why it hasn’t worked

Programmes have tried both formal and informal diffusion routes. Only around 40% of projects ran follow-up activities such as field days and exchange visits, only 30% built platforms such as farmer clubs, and more than 20% trained FFS graduates as farmer-trainers — and none of these produced measurable neighbour effects in the pooled quantitative evidence (the review notes short-run diffusion of simple practices such as reduced spraying among relatively educated cotton growers in China and Pakistan, but that it "falls considerably over time"). The qualitative record explains why: in Indonesia, "despite trained farmers teaching their neighbours about IPM during pest outbreaks, the ideas were not properly explained or understood"; in India, "non-participants did not have the confidence to implement the new practices they had heard about from their neighbouring FFS graduates"; in Cameroon and Cambodia low social cohesion limited communication, and in Indonesia socio-economic differences between participants and non-participants blocked it. Targeting more educated farmers as early adopters "may backfire when it comes to diffusion" — one Indonesian study found participants "communicated to a 'selective audience in the villages' and made no deliberate efforts to train other members of the community." Observability is the crux: farmers adopted when they saw with their own eyes that unsprayed plots yielded the same, and neighbours who did not run the comparison had no such evidence — the mechanism that makes FFS work for participants is the same one that prevents it from spreading. Programmes were also undercut by lack of back-stopping: without ongoing technical support, even graduates lapsed "in the face of pest outbreaks and continuing recommendations to use pesticides."

What would unlock progress

The gap is a communication-design problem: how to give a non-participant the observable evidence and confidence that a season in the field gives a participant, at a fraction of the cost. Candidate reframings the review's own findings suggest are (a) making the comparison plot itself the diffusion medium — public, signposted, visited paired plots at village scale rather than private learning inside the group; (b) redesigning the farmer-trainer role around demonstrations rather than explanation, and selecting for the "inquisitive farmers" who in one Indonesian case became recognised village "farmer professors"; and (c) building the follow-up and back-stopping that most projects skipped, since diffusion appears to require a persistent local source of confidence, not a one-off transfer. Adjacent fields with a lead: village-level demonstration plots in seed marketing, "model household" approaches in nutrition, and the video-mediated farmer-to-farmer extension model (Digital Green), which turns experience into a shareable artefact and whose evidence base is a natural comparison.

Entry points for student teams

A team could build the "diffusion kit" for one FFS group — a village-visible paired-plot protocol, a simple photo/video record format for the season's comparisons, and a demonstration-based script for graduates — and hand it to an NGO or extension service as a pilot-ready package, since actually measuring neighbours against a comparison village takes a full FFS season and a two-village survey abroad; in-semester the kit's neighbour knowledge and adoption-intention instrument can be validated by cognitive interviews with extension agents, agricultural-education faculty, and farmers reachable at home, and the measurement plan (allocation, outcomes, power) pre-registered with the AEA RCT Registry (socialscienceregistry.org) or OSF Registries (osf.io/registries) for the partner who owns the field access. A research team could re-code the review's qualitative studies — the underlying evaluations are indexed and largely open through the 3ie Development Evidence Portal (developmentevidence.3ieimpact.org) — to build a taxonomy of what non-participants actually heard versus what graduates knew, identifying which IPM concepts are verbally transmissible and which are not; this door needs nothing but the published record. A design team could prototype low-literacy visual tools (predator/pest count cards, plot-comparison boards) that let a graduate show rather than tell. Relevant skills: agricultural extension, communication design, development economics/impact evaluation, video and visual design, and field survey methods.