This paper describes the learning selection approach to enabling innovation that capitalizes on the complexity of social systems at different scales of analysis. The first part of the paper describes the approach and how it can be used to guide the early stages of setting up a “grassroots” innovation process. The second part of the paper looks at how the learn selection model can be used “top-down” to guide research investments to trigger large-scale systemic change.

This paper (Part I) present a case study of work conducted by the International Centre for Tropical Agriculture (CIAT) to adapt network mapping techniques to a rural and developing country context. It reports on work in Colombia to develop a prototype network diagnosis tool for use by service providers who work to strengthen small rural groups. It is complemented by a further paper in this issue by Louise Clark (Part II) which presents work to develop a network diagnosis tool for stakeholders involved in agricultural supply chains in Bolivia.