AN EVALUATION OF THE SYMPOSIUM ON SYSTEMS ANALYSIS
F. H. W. MORLEY
The symposium has provided an occasion for the presentation of models ranging from the growth of a plant, through the growth of pastures and soil-plant- animal systems, to whole-farm and whole-flock decision models embodying economic and managerial elements in objective functions.
The conference has provided an occasion for the presentation of many new facts. Some of these are relevant to production systems. Others may become relevant when appropriate systems are to be analysed. Data on reproductive systems, genetic parameters, responses to supplementary feeding and under- standing of host-parasite-environment systems are fuel for model builders. But, as emphasised by Mullaney, the model builders and data gatherers need to get together.
Corbett, Graham and others must derive encouragement from the use of their estimates by the model builders. Even if the use is not, at all times, quite appropriate, encouragement may rise above irritation, especially if the alternative to this is to be ignored. Model building is a collaborative enterprise and tolerance is necessary. Misfits are a necessary consequence of evolution. Witty aphorisms are being used against systems studies, but this has always been true of new
ventures. .
No model is complete and one must always be conscious of inadequacies.
Some important variable may have been left out, some parameter grossly under- estimated, some relationship critically inaccurate. But, as Trebeck emphasises, a model which assembles and organizes the stock of available information is likely to be more enlightening and useful than alternatives.
Therefore it seems to me that some model builders have sold their goods short. Their imperfect creations may seem vulnerable but what alternatives are available? Certainly the model of Christian et al. is insufficient for decision- making because it studies only part of a production system. But that of Jackson and Turner, although inviting addition and refinement, has important implications in applied animal improvement.
The emphasis is now swinging from the “gadgetry” of model building to the use of models in research and application. Perhaps systems analysis is about to occupy a place in agriculture somewhat similar to that currently occupied by biometry. In the coming years we may see a community marriage of Pearsonian, Fisherian and Bayesian statistics with systems analysis. Curve fitting, inductive inference, subjective probabilities, information from many sources, and the inte- gration of these in interpretation, recommendations and applications may become routine. The time has just about arrived when the building of a model is no longer to be regarded as a major achievement unless this contributes significantly to under- standing or decision-making.
Perhaps the next Biennial Conference, or the one after that, will be the occasion for the presentation of experimental results in combination with models
137.
which encompass these in the frame of a system, which interpret them through the outputs of model simulation, and evaluate treatments through their predicted effects on realistic, perhaps complex, objective functions. This is already the pattern of much research in animal improvement; it will become the pattern of much research in animal and pasture management, and animal production systems generally. Therefore this symposium may be an introduction to a new or expanded outlook in research, extension, and application, activities of profound significance
to
members of this Society.138