Book: The Sciences of the Artificial

Overview
Herbert Simon argues for treating humanly designed systems, artifacts, organizations, programs, as objects of scientific inquiry in their own right. He distinguishes "the artificial" from the natural by the presence of intentional design to achieve goals, and proposes developing general principles that explain how such systems are created, how they function, and how they can be improved. The orientation shifts attention from purely descriptive natural science toward a science concerned with design, planning, and purposive behavior.

Central concepts
Simon introduces key ideas that recur throughout the analysis: complexity and hierarchy, the role of representation in problem solving, the limits of human information processing, and the primacy of design as a mode of scientific explanation. He frames artifacts in terms of their functions and the constraints and means available to designers, and emphasizes that satisfying a goal often requires satisficing, choosing a course that is good enough given bounded resources, rather than optimizing in a mathematically complete sense.

Design and the artificial versus the natural
The artificial is defined by purpose and constructedness: artifacts embody intentions and are evaluated by how well they perform intended functions. Simon contrasts the explanatory strategies appropriate to natural systems with those appropriate to designed systems, arguing that explanations of the artificial often depend on principles of organization and manufacturability, on representations that capture designer intentions, and on a normative component about desirable outcomes. He explores how structure serves function and how functional analysis can reveal multiple realizations of the same purpose.

Problem-solving, heuristics, and representation
Problem solving is central to design, and Simon emphasizes the importance of representation in transforming intractable search problems into ones soluble by heuristic methods. He demonstrates that the way a task is represented dramatically affects the tractability of finding solutions, and he articulates general methods for search, decomposition, and the generation of satisficing solutions. He situates human problem solving within an information-processing framework that lends itself to computational modeling without assuming unlimited rationality.

Bounded rationality and satisficing
Simon challenges the ideal of unbounded optimization and replaces it with bounded rationality: decision makers operate with limited information, limited computational capacity, and limited time. Under these limits, satisficing, accepting an adequate solution, becomes rational. This perspective recasts observed human and organizational behavior as adaptations to constraint rather than as failures relative to unattainable ideals, and it supplies a basis for understanding design choices, routine procedures, and organizational routines.

Hierarchy, complexity, and near-decomposability
Complex systems are often organized hierarchically, with subsystems that interact more strongly internally than with other parts. Simon formulates the principle of near-decomposability, under which complex artifacts become analyzable and designable because they can be broken into modules whose interactions are limited. This insight connects to notions of modular design, evolvability, and tractability, explaining why certain architectures are easier to build, maintain, and adapt.

Implications for AI, management, and engineering
The methodological and conceptual tools Simon develops bridge artificial intelligence, cognitive psychology, engineering, and organizational theory. By treating design as a science, he provides a rationale for modeling cognitive processes computationally, for using simulations and formal models to study organizations, and for developing systematic design methodologies. The emphasis on representation, heuristics, and modularity continues to inform approaches in AI, human-computer interaction, and systems engineering.

Legacy and influence
Simon's synthesis reframed multiple disciplines around shared problems of design and adaptation. The call for a science of the artificial legitimized research on computational models of cognition, normative theories of design, and formal analyses of organizations as artifacts. Concepts such as bounded rationality, satisficing, and near-decomposability remain central to contemporary thinking about decision making, complex systems, and the engineering of adaptive artifacts.