Modularity

Modularity is the design principle of decomposing a system into subsystems — modules — that interact through well-defined interfaces while hiding their internal complexity. It is the central strategy by which intelligent agents, whether biological or artificial, manage complexity without requiring comprehensive knowledge of the systems they interact with. A module is a black box: you need not know how it works, only what it promises to do.

The principle is visible across scales. The brain is organized into functionally specialized regions that communicate through standardized neural codes. Software is built from libraries and APIs that expose functionality while concealing implementation. Markets are composed of firms that transact through contracts rather than continuous internal monitoring. In each case, modularity reduces the dimensionality of the interaction problem: instead of tracking every internal state of every component, the system tracks only interface states.

The cost of modularity is information loss. When you treat a module as a black box, you sacrifice the ability to exploit internal structure for optimization or prediction. A tightly integrated system — a monolith — can be faster, more efficient, and more coherent than a modular one. The history of technology is a pendulum between modular and integrated architectures, driven by which cost (complexity or efficiency) dominates at a given moment. Modularity is not an unqualified good. It is a bet that the complexity you are avoiding is more dangerous than the efficiency you are sacrificing — a bet that is sometimes wrong but that makes most large-scale systems possible at all.