Why craft-lovers are losing their craft

Craft vs “make it go” programming

• Longstanding split: some treat programming as a craft (architecture, clarity, long-term maintainability); others see it as a means to an end (“make it go”).
• Several argue LLMs amplify “slopware”: more throwaway, copy‑pasted, or AI‑generated code that craft‑oriented developers must debug or rewrite.
• Others counter that “over‑crafted” code can also be a maintenance burden; simple, boring code is a virtue.

LLMs, quality, and daily workflow

• Concerns:
  • LLM‑written code can be inconsistent, wrong, or just “differently broken,” encouraging replacement rather than real fixes.
  • AI‑generated tests often assert implementation details, not behavior, increasing churn and false positives.
  • Auto‑generated commit messages and reviews can contain fabrications, shifting effort onto reviewers.
• Supportive views:
  • LLMs excel at small, tedious tasks (finding references, boilerplate, merge cleanups).
  • New “craft” emerges around prompt design, building agents, and orchestrating systems of tools.
  • Some find programming more fun with LLM assistance; others report the opposite and feel alienated.

Analogies to other technologies and crafts

• Frequent comparisons to:
  • Photography (film → digital → smartphones): lowers barriers, creates massive mediocre output, but experts remain.
  • Woodworking: hand tools vs power tools/CNC; some value process and skill, others just want a finished product (parodied with IKEA/prints analogies).
  • Industrialization: looms, typesetting, editing film, etc.; software once displaced other crafts and is now being “industrialized” itself.

Labor, value, and “march of progress”

• Some see LLMs as another step in capitalism’s drive to deprofessionalize work, reduce wages, and concentrate gains with capital owners.
• Debates over intellectual property, “training on stolen work,” and fairness vs legalism; strong disagreement on whether this matters or is simply inevitable.
• Broader worry about a shrinking middle class, potential UBI, and loss of meaningful, well‑paid craft work.

Coping, adaptation, and future niches

• Suggested responses:
  • Separate paid work from hobby craft (code for money one way, for love another).
  • Shift careers (e.g., teaching) to protect time and autonomy.
  • Accept that software “craft” may retreat into niches where process is visible and valued, similar to high‑end luthiery or watchmaking.
• Others argue energy and cost realities may limit giant general‑purpose LLMs, favoring smaller, targeted tools.