A new approach to problems of land allocation for livestock grazing, combining both computer science and social science tools has been developed since 1998 in Senegal, especially in the Ferlo area. Here we examine the implications for the different research centres such as the CIRAD1, the Pôle Pastoral Zones Sèches (PPZS), the ESP and the laboratories of the UCO. A computer simulation (MAS), based on sociological data, has been introduced in order to obtain a more neutral evaluation of the possible approaches to the problem, for example the approach of the local people involved, together with that of regional land development and allocation policy. This is preferable to practice that consists of experimenting policy in real-life situations with potentially dire consequences for the population and the environment. This initiative aims both to stabilise the social position of shepherds/herders and to preserve the production potential of ecosystems used for grazing. Its approach is one of sustainable development and the underlying theory must be questioned in order to clarify the empirical scope and pertinence. This applied research (which, owing to the questions it asks, is rooted "in the most immediate reality" [1]), should not be naïve concerning its implications. Is it possible, with theoretical knowledge and an objective approach, to give research results to local management especially when the latter are separated from the decision making process? The project's aim is to modify social behaviour, to rationalise it or to induce new behaviour. Groups or individuals that have to modify their behaviour as a result will naturally give rise to questioning concerning the social benefits of such Changes. Who is destined to benefit from the MAS platform? From the perspective of increased well-being to local populations a long-term follow-up is required together with an evaluation of social relations resulting from its use. It will certainly be necessary to implement a larg