Today several branches of applied science assist the firm to achieve procedural rationality. 3 One of them is operations research (OR); another is artificial intelligence (AI). OR provides alogrithms [RULES] for handling difficult multivariate decision problems, sometimes  involving uncertainty. Linear programming, integer programming, queuing theory [eg FIFO], and linear decisions rules are examples of widely used OR procedures.

To permit computers to find optimal solutions with reasonable expenditures of effort when there are hundreds or thousands of variables, the powerful [RULES] associated with OR imposes a strong mathematical structure on the decision problem. Their power is bought at the cost [EXPENDITURE] of shaping and squeezing the real-world problem to fit their computational requirements: for example, replacing the real-world criterion function and constratins with linear approximations so that linear programming can be used. Of course the decision that is optimal for the simplified approximation will rarely be optimal in the real world, but experience shows that it will often be satisfactory.

The alternative methods provided by AI, most often in the form of heuristic search (selective search using rules of thumb) [BIAS, PREJUDICE], find decisions that are "good enough," that satisfice. The AI models, like OR models, also only approximate the real world, but usually with much more accuracy and detail than the OR models can admit. They can do this because heuristic search can be carried out in more complex and less well structured problem space than is required by OR maximizing tools. The price paid [EXPENDITURE] for working with the more realistic but less regular models is that the methods generally find only satisfactory solutions, not optima. We must trade off satisficing in a nearly-realistic model (AI) against optimizing in a greatly simplified model (OR). Sometimes one will be preferred, sometimes the other.