"A model is generally more believable if it can predict what will happen, rather than 'explain' something that has already occurred. […] Model building is not so much the safe and cozy codification of what we are confident about as it is a means of orderly speculation." (James R Thompson, "Empirical Model Building", 1989)
"Although it is difficult to say where problem posing ends and problem solving begins, since in the ideal state there is continuous interaction between the two, model building is more concerned with the former than with the latter. Whereas problem solving can generally be approached by more or less well-defined techniques, there is seldom such order in the problem posing mode." (James R Thompson, "Empirical Model Building", 1989)
"Among the advantages of a simulation approach is principally that it enables us to eliminate time-consuming and artificial approximation theoretic activities and to spend our time in more useful pursuits. More importantly, simulation will enable us to deal with problems that are so complex in their 'closed form' manifestation that they are presently attacked only in ad hoc fashion." (James R Thompson, "Empirical Model Building", 1989)
"Empirical model building refers to a mindset that lends itself to constructing practical models useful in describing and coping with real-world situations." (James R Thompson, "Empirical Model Building", 1989)
"Whenever we axiomitize a real-world system, we always, of necessity, oversimplify. Frequently, the oversimplification will adequately describe the system for the purposes at hand. In many other cases, the oversimplification may seem deceptively close to reality, when in fact it is far wide of the mark. The best hope, of course, is the use of a model adequate to explain observation. However, when we are unable to develop an adequate model, we would generally be well advised to stick with empiricism and axiomatic imprecision." (James R Thompson, "Empirical Model Building", 1989)
No comments:
Post a Comment