Some issues in the design of agricultural decision support systems

The development of crop and soil process models, and their incorporation into decision support systems (DSSs) for the practical management of agricultural production, have made a significant contribution to the focus and integration of research by agricultural scientists. However, such DSSs should not necessarily be used as vehicles for communicating with other groups, such as farmers, with a view to effecting directed change in their behaviour. Other, sometimes more effective and efficient (and often more enjoyable and elegant) models are available. The historical, and seemingly unwarranted, escalation of organisational commitment to the provision of agricultural DSSs appears to have been based on a fundamental category mistake of confusing process models (for professional research) with model-based DSSs (intended as a guide to practical action). This escalation of commitment has had substantial implications for the design and implementation of both modelling and DSS projects, and has precluded the development of distinctive criteria for evaluating DSS success. Design issues for DSSs include: the need for an analytical phase in DSS development to deconstruct professional models, resolution, validation and appropriateness in relation to intended purpose, in order to avoid over-engineered solutions; its contribution to communication practice; and the role of professional responsibility. The SIRATAC cotton pest management system, used in Australia between 1980 and 1993, is presented as a case study. Copyright (C) 1996 Published by Elsevier Science Ltd