O.R.'s initial development began in the United Kingdom during World War II and was quickly taken up in the United States. This start took place in a … - C. West Churchman

The systems approach to problems does not mean that the most generally formulated problem must be solved in one research project. However desirable this may be, it is seldom possible to realize it in practice. In practice, parts of the total problem are usually solved in sequence. In many cases the total problem cannot be formulated in advance but the solution of one phase of it helps define the next phase. For example, a production control project may require determination of the most economic production quantities of different items. Once these are found it may turn out that these quantities cannot be produced on the available equipment in the available time. This, then, gives rise to a new problem whose solution will affect the solution obtained in the first phase.

A system may actually exist as a natural aggregation of component parts found in Nature, or it may be a man-contrived aggregation – a way of looking at a problem which results from a deliberate decision to assume that a set of elements are related and constitute such a thing called ‘a system.

I am often inclined to put the implementation questions first, ie, "Can anything be changed?" Should the implementation question not accompany the whole process from its very beginning to its very end?