would help to shape and confirm decisions. Evaluation does not necessarily imply that producers are not capable of making sound decisions or that input suppliers are untruthful in their claims. Research in the area of decision theory indicates that sound judgments are particularly difficult for humans in situations where there is considerable variability, where there are time lags between actions and results, and where there are multiple and complex cause-and-effect relationships. These three characteristics seem to apply to precision agriculture and its adoption and evaluation.
Several factors will make it difficult for public agencies to carry out such evaluations in the area of precision agriculture. The technology is changing so rapidly that evaluations of specific technology components have a very short useful life. Producers will not be well-served if equipment or products embodying technologies which have been evaluated cannot be compared with newer, more sophisticated versions that have not been evaluated. Products that have been validated in the field before submission for evaluation, whether prototypes by industry or federal expertise, should be compared on the same basis as existing commercial products. System evaluations are appropriate on technologies that are installed, maintained, and operated as specified by the manufacturer. Because the area is evolving so rapidly, technology developers may be reluctant to expose newly developed technology to public evaluation, risking loss of proprietary and trade secret information. Meaningful collaboration between private firms and public agencies, and between agencies, may not be forthcoming without considerable effort.
Over the long term, there is no substitute for carefully designed observation of economic and environmental results obtained by actual producers in real field conditions. For these experiments to be useful, side-by-side treatments and statistical control methods need to be used to distinguish precision agriculture technology's contributions from normal variation in resources, weather, and management. Given the systems nature of precision agriculture techniques and the importance of site-specific variability, on-farm experimentation performed in collaboration with producers will be necessary and desirable, compared with more traditional farms of plot-based research design (Alliance on Agricultural Information Technology, 1996). Research collaborators can mine a wealth of on-farm data and use regression analyses and other multivariate statistical methods to isolate the multiple sources of variations that influence economic and environmental outcomes of precision agriculture. These findings can provide invaluable guidance to producers on the expected benefits from adoption of precision agriculture technologies in their particular setting.
The accuracy and reliability of methods for collecting precision data need to be evaluated to ensure confidence in grid soil-sampling schemes, directed sampling, and yield monitor results (Blackmer and Schepers, 1996; Lamb et al., 1995). Similarly, the accuracy and reliability of methods for making precision applications of fertilizers, pesticides, irrigation water, and other inputs also need confirmation