The Sciences of the Artificial
Source: https://mitpress.mit.edu/9780262691918/the-sciences-of-the-artificial/ ↗
Simon's argument is that designed systems — artefacts, organisations, software, economies — require their own science, distinct from the natural sciences: a science of the artificial, concerned with the interface between inner and outer environments.
The book introduced the concept of "satisficing" (good enough rather than optimal), the idea that complex systems are nearly decomposable, and the argument that design is the core intellectual activity of the professions.
For product direction this is the most foundational book on the shelf: every product is an artificial system, and Simon's framework for thinking about how such systems are designed, bounded, and adapted is more powerful than anything written since.
Read alongside Administrative Behavior for the organisational companion and Alexander's Notes on the Synthesis of Form for the design complement.
Short, dense, Simon at his most ambitious.
Central argument
Simon argues that artefacts — from software to organisations to economies — are not governed by natural laws but by the interface between their inner constitution and the outer environment they are designed to serve, and that this distinction demands an entirely separate science of the artificial. The central move is showing that design is not engineering intuition but a rigorous intellectual discipline shared across all the professions, unified by the logic of how artificial systems are shaped to meet goals under constraints. Two foundational concepts follow: 'satisficing', the claim that designers and decision-makers do not optimise but instead seek solutions that are good enough given bounded rationality, and near-decomposability, the idea that complex systems are organised into semi-independent subsystems whose internal interactions are far stronger than their cross-subsystem ones.
Critique
Simon's framework was built largely from observation of mid-twentieth-century administrative and computational systems, and it shows: the 'inner environment' he theorises is implicitly stable and relatively well-specified, which sits poorly with digital products whose core design problem is that both the artefact and its environment co-evolve at speed, often driven by network effects and user behaviour that the system itself generates. Near-decomposability is a powerful heuristic, but Simon offers little account of how to handle the cases — common in platform and ecosystem design — where the most consequential dynamics are precisely the cross-subsystem couplings he treats as weak. The theory of satisficing also tends to naturalise constraint rather than interrogate where the bounds come from, which can make it a conservative framework when the design challenge is to reframe the problem entirely.
Why it matters for product
The concept of near-decomposability is arguably the most practically underused idea in product organisation: it is the theoretical basis for why team topologies built around bounded domains reduce coordination cost, and it gives CPOs a principled rather than intuitive argument for how to carve product surface into teams. Simon's insistence that design is always about the interface between inner and outer environment reframes discovery work — the question is never just 'what do users want' but 'how does the system's current inner structure fail to meet its outer environment, and what adaptation is required'. Satisficing, read carefully, is also a direct argument against OKR maximalism: it suggests that setting targets at 'good enough' thresholds and redirecting cognitive resources to the next constraint is more rational than pursuing marginal optimisation on a metric that has already cleared its threshold.