There is no such science, and there never will be. Our objectives are typically imprecise and multifaceted; they change as we work toward them, as they should. Our decisions depend on the responses of others and on what we anticipate those responses will be. The world is complex and imperfectly understood, and it always will be.

We do not solve problems in the way the concept of decision science implies, because we can’t. The achievement of the great statesman is not to reach the best decision fastest but to mediate effectively among competing views and values. The achievement of the successful business leader is not to foresee the future accurately but to continuously match the capabilities of the firm to the changing market. The test of financial acumen, as described by Buffett and Soros, is to navigate successfully through irresolvable uncertainties.

Mostly, we solve problems obliquely. Our approaches are iterative and adaptive. We make our choices from a limited range of options. Our knowledge of the relevant information, and of what information is relevant, is imperfect. Different people will form different judgments in the same situation, not just because they have different objectives but because they observe different options, select different information and assess that information differently; and even with hindsight it will often not be possible to say who was right and who was wrong. In a necessarily uncertain world, a good decision doesn’t necessarily lead to a good outcome, and a good outcome doesn’t necessarily imply a good decision or a capable decision maker. The notion of a best solution may itself be misconceived. The skill of problem solving frequently lies in the interpretation and reinterpretation of high-level objectives.

There is nothing wrong with using trust as a basis for decision. Finding people you can trust, or establishing trusting relationships with them, is the most effective—often the only effective—means of achieving the delegation that is necessary to accomplish objectives and goals in large organizations. Successful decentralization relies on the transmission of high-level objectives, not just intermediate goals and basic actions, to the agents who will implement them. This is a world apart from principal-agent models that treat social organizations as mechanical systems in which agents respond to the stimuli that incentive structures impose.