Sciences of the Artificial
What is science for contingent and constructed things?
2026-01-07 — 2026-01-07
Wherein Simon’s argument that designed systems are best studied empirically is presented, and nearly‑decomposable hierarchies are offered as a means to model complexity in organizations.
I read Simon (1996) back in 2010, and it seemed cool. Now that the world is full of AI systems and complex engineered socio-technical systems, it seems like I didn’t notice how prescient it was.
First published amid the rise of cybernetics, early AI, and systems theory in the post-WWII era, the book challenged the dominance of the natural sciences (looking at you, physics) by advocating empirical study of designed, human-made systems.
1 Historical Context
In the 1960s, amid Cold War computing advances and behavioural psychology’s shift from behaviorism, Simon positioned the artificial sciences as complementary to the natural ones, drawing on his work in bounded rationality and organization theory. He responded to debates on whether engineering and design qualified as “science,” arguing synthetic systems—like economies or AI—warrant rigorous study despite their goal-oriented, adaptive nature. This sounds like some kind of pushing the edges of the scientific method, which often includes a universality that he explicitly rejects here.
2 Methodological Approach
Simon defines artificial systems by their interface with their environments: inner states adapt to outer constraints via functions and goals, enabling prediction through simulation. This yields an information-processing view of cognition, rejecting perfect rationality in favour of empirical models testable via computers, as in his “nearly decomposable hierarchies” for complexity.
3 Philosophy of Science
Philosophically, Simon elevates design as a science—devising actions to change situations—bridging “what is” (natural laws) and “what ought to be” (goals), thus unifying the social sciences, economics, and engineering under empirical rigour. Maybe.
He critiques reductionism, favouring hierarchical structures where wholes exceed their parts.
4 Post hoc relevance
In hindsight, a lot of things look like Herbert Simon call-outs. AI Evals and ML Benchmarks are about evaluating designed systems. institution design, movement design, and incentive mechanisms are about designing socio-technical systems. Epistemic systems are about designing knowledge systems. All of this feels very relevant.