Up to 2004 (when this field study ended), the Ethiopian early warning system had not enabled the accurate quantification of harvests or beneficiaries despite the claims of the ‘Net expert system’. At the most, it had made it easier to pinpoint agro-pastoral areas suffering from shortages and to identify broad tendencies relating to harvests and livestock. The EWS results facilitate consensus between institutional actors involved in the allocation of food aid, but it can indeed provide an early warning in the sense that it highlights the need for more accurate evaluations and quantified measurements (harvests, malnutrition, etc.).

It is to be hoped that decision-makers will acknowledge both the current extent of uncertainty and its irreducibility, lead to crude tendencies. This would enable the energy and financial resources now being expended on data production to be reallocated to more pragmatic food security programmes. But Ethiopia is looking in another direction, and reinforcing the status and capacities of field development agents in the collection of basic data.

As part of an evolutionary process in which the EWS is the collective product of actors who participate in a dynamic of negotiation that encompasses every link in the decisional chain, institutional actors are in a position to help improve the performance of the EWS. In this respect, SCF’s integration of socio-economic approaches or nutritional surveys into early warning systems is an encouraging sign. The rapid methods suggested by the WFP seem to merit further development, for they would strengthen the pragmatic approaches already adopted by experts, who have cultivated useful visual estimation skills and adapted them to the time constraints imposed on evaluation exercises. Working with teams as regularly as is possible would also increase efficiency.

In conclusion, humanitarian actors would be advised to regard the EWS as a simple indicator tool, systematically question data production methods, and maintain constant vigilance when investigating and verifying conditions in the field. In other words, they should look beyond the official data and the most convincing investments in form.