Key Principles of Software Architecture: Entropy, Broken Windows, Complexity, SOLID, and Design Strategies

Software architecture is a classic cornerstone of computer technology, yet real‑world projects often suffer from design gaps, aging interfaces, and code decay, making sustainable evolution difficult.

The article introduces the "entropy law"—originally from physics—to illustrate how software inevitably becomes more disordered, leading to technical debt that rarely gets repaid.

It then discusses the "broken‑window effect," showing how small, ignored defects can cascade into larger quality problems, and explains why architects must focus on both behavioral value (delivering functional features) and architectural value (ensuring flexibility and low cost).

Complexity is defined as the product of component complexity and the time developers spend on it; its main causes are dependency and obscurity. Symptoms include change amplification, cognitive load, and unknown unknowns, which accumulate over time and hinder maintainability.

To combat complexity, the article advocates orthogonal design, which isolates concerns, improves productivity, reduces risk, and aligns team structures, as well as the principle of “complexity sinking”—pushing complexity to lower layers (e.g., language runtimes, service meshes).

Classic design principles such as the Ockham razor, consistency, SOLID (Single Responsibility, Open‑Closed, Liskov Substitution, Interface Segregation, Dependency Inversion), DRY, and layered abstraction are presented with practical guidance on applying them in large‑scale systems.

Component coupling rules—no circular dependencies, stable‑direction dependencies, and matching abstraction levels—are highlighted to keep systems modular and evolvable.

Boundary definition, anti‑corruption layers, open host services, and publish/subscribe events are described as techniques for isolating core business logic from technical details like databases, web frameworks, and infrastructure.

The article concludes that treating architecture as a strategic, long‑term investment—focusing on clear boundaries, minimal coupling, and reusable abstractions—maximizes developer productivity while minimizing operational costs.