Programming Is Not a Skill

Programming’s central claim is that writing code by hand has not been the enduring skill of software work; successive abstractions have steadily automated it. In the 1950s, automatic coding and languages such as FORTRAN were expected to eliminate low-level coding. Instead, they removed tedious work and elevated programmers toward design and analysis, while the separate coder role disappeared. Later advances - from BASIC and spreadsheets to frameworks, no-code tools, and now LLMs - extended the same process, letting more people produce software while reducing the hand-written code needed per system. The author argues that these predictions were wrong only when taken to mean the profession would vanish: the specific work of writing code was repeatedly displaced, while the job title survived.

A parallel shift automated checking. Structured programming, shared design patterns, automated tests, and continuous delivery made code and its behavior easier for people or machines to inspect, verify, and ship. The author’s explanation is that machines take over tasks that are costly for people to perform but cheap to check; what remains human is costly to evaluate, especially deciding what to build, specifying requirements, and judging whether results are correct or useful. Code-writing is described as manual labor that can obscure the deeper craft of software design, a distinction supported by Fred Brooks’s separation of conceptual design from implementation difficulties. The argument allows that coding can itself reveal design problems, but locates the lasting skill in judgment rather than syntax.