Are researchers, academics and policy makers from within a project country the best people to facilitate sustainable education development? Based on experiences of a project in a rural community in South Africa, I suggest in this paper that even same-country 'experts' may inhibit local ownership. In 1998, a colleague and I initiated an education development project in a rural area in South Africa. It was known as the Estcourt project and was based on a model of collaborative action research. Two years later, the introduction of a new project in the area saw the demise of the Estcourt project and questions about the project methodologies and its inherent assumptions became paramount. The arguments presented in this paper rest in the story of myself as researcher, attempting to find answers to the Estcout project 'failure' through a case study of a school in the Estcourt project and its communities in Black rural South Africa. The paper interrogates a notion of the interconnectedness of multiple agendas in collaborative action research, contextualised within an understanding of political relationships between the researched and the researcher in a post-apartheid environment of transformation in South Africa, and in relation to the inherent politics of global agendas for education development. It describes how this is played out in both in policy and practice in South Africa which leads to development that is ultimately not locally owned. It uses the Inclusive Education policy development process and the Estcourt project as particular examples of this.