In 2010, Fred Brooks published The Design of Design. It is his best book and his least read.
The argument: conceptual integrity is the most important property of any designed system. Everything else serves it.
"I will contend that conceptual integrity is the most important consideration in system design. It is better to have a system omit certain anomalous features and improvements, but to reflect one set of design ideas, than to have one that contains many good but independent and uncoordinated ideas." — The Mythical Man-Month, 1975
The tradeoff: omit useful features to preserve coherence. Say no to good ideas that don't fit.
Definitions
Conceptual integrity. The system feels like one mind designed it. The user forms one mental model. They predict behavior in new situations.
Orthogonality. One way to do each thing. No overlapping concepts. Non-redundancy.
Propriety. Nothing unnecessary.
"The essential skill of the designer is saying no — repeatedly, to smart people with good arguments — and having the authority to make it stick. Every added feature is a subtraction from all existing features, for it adds complexity without corresponding benefit."
Feature cost is multiplicative. Each new thing makes every existing thing harder to find, learn, and use.
Generality. No arbitrary limits. Primitives compose to handle unanticipated cases. Generality without orthogonality is a kitchen sink. Orthogonality without generality is a toy.
Brooks was not alone. Knuth: programs should be readable by humans. Dijkstra: elegance prevents bugs — transparent structure hides nothing. Kernighan and Plauger on clarity vs. muddle in The Elements of Programming Style (1974). Wirth: Pascal as deliberate integrity — one mind, one language. Lampson: the Alto's GUI worked because one designer decided what to leave out.
Brooks's contribution: the structural precondition. One mind. In Computer Architecture (1997, with Blaauw), he showed this governed even ISAs, where orthogonality of operations and addressing modes was the explicit goal.
Proof
"Reims Cathedral has conceptual integrity. Built over eight generations of architects, each stuck to the original plan. The result is a unified work, coherent in every detail. Most cathedrals are not like this; their conflicting concepts produce architectural chaos."
Reims worked because the original architect's plan had authority that outlived him. Two centuries of successors submitted. It should not exist. That it does is the evidence.
Most cathedrals: Gothic nave, Renaissance facade, Baroque chapel on Romanesque transept. Each generation had a vision. Nobody had authority. Chaos.
Software systems are cathedrals built over decades. Unix: Thompson and Ritchie, one resonant pair. Lisp: McCarthy, one mind. Go: Thompson, Pike, Griesemer. The Macintosh: Jobs. These are Reims. Thompson: "One of my most productive days was throwing away 1,000 lines of code."
Part 2 → · Part 3 · Part 4 · Part 5 · Part 6 · Part 7
Fred Brooks, The Mythical Man-Month (1975, Anniversary Ed. 1995), No Silver Bullet (1986), The Design of Design (2010). Brooks & Blaauw, Computer Architecture: Concepts and Evolution (1997).
Brooks's principles apply beyond software. Conceptual integrity, the one-mind rule, the empiricist method — these are engineering principles that hold across any designed system. The building, the organization, the codebase, the protocol. The medium changes. The principles don't. That is the definition of engineering: principles that hold across domains.
Conceptual integrity is not about consistency. It is about coherence. A system can be consistently bad. A coherent system feels like one mind designed it, even if many hands built it.