Less Art, More Engineering: It Wasn't Measurement
Two and a half years ago I wrote an article called "Less Art, More Engineering". The argument was simple: software development calls itself engineering but mostly isn't, because it runs on subjective criteria with no verifiable definitions. Readability with no metric. Best practices that contradict each other. Design methods that offer no way to check whether a design is correct. I ended on a promise: the next step is to convert what we know into a set of clearly defined, well grounded rules and criteria. "Ideally measurable," I wrote, "but in either case easily verifiable."
I want to report back on that promise, because I got the diagnosis right and the cure wrong.
None of this began as a plan. The original goal was modest: organize code so its structure would be recognizable and its behavior predictable. I had nothing in mind beyond that. But the moment I asked what actually makes structure recognizable, the question refused to stay small. Answering it meant understanding low-level structure and the forces that shape it, and each answer uncovered the next one. That inquiry grew into Process-First Design, and Process-First Design in turn produced Architecture Synthesis. I set out to tidy functions and ended up deriving systems.
