Decision-making can no longer be confined to the very small group at the top. In one way or another almost every knowledge worker in an organization will either have to become a decision-maker himself or will at least have to be able to play an active, an intelligent, and an autonomous part in the decision-making process. What in the past had been a highly specialized function, discharged by a small and usually clearly defined organ—with the rest adapting within a mold of custom and usage—is rapidly becoming a normal if not an everyday task of every single unit in this new social institution, the large-scale knowledge organization. The ability to make effective decisions increasingly determines the ability of every knowledge worker, at least of those in responsible positions, to be effective altogether.
The computer has great impact on strategic decisions. It cannot make them, of course. All it can do—and even that is potential rather than actual so far—is to work through what conclusions follow from certain assumptions made regarding an uncertain future, or conversely, what assumptions underlie certain proposed courses of action. Again, all it can do is compute. For this reason it demands clear analysis, especially of the boundary conditions the decision has to satisfy. And that requires risk-taking judgment of a high order.
There are additional implications of the computer for decision-making. If properly used, for instance, it should free senior executives from much of the preoccupation with events inside the organization to which they are now being condemned by the absence or tardiness of reliable information. It should make it much easier for the executive to go and look for himself on the outside; that is, in the area where alone an organization can have results.
The computer might also change one of the typical mistakes in decision-making. Traditionally we have tended to err toward treating generic situations as a series of unique events. Traditionally we have tended to doctor symptoms. The computer, however, can only handle generic situations—this is all logic is ever concerned with. Hence we may well in the future tend to err by handling the exceptional, the unique, as if it were a symptom of the generic.