Some things just take time
AI coding tools promise unprecedented speed, but many engineers argue that rushing work simply lets teams create the wrong things faster while eroding hard-won practices like careful design, testing, and iteration. Commenters contrast “vibe-coded” AI output with long-lived projects, open source communities, and even centuries-old trees, emphasizing that trust, quality, and good products emerge from sustained effort and real-world feedback that cannot be compressed. Others note that while AI can eliminate drudgery and enable more experiments, the real constraint has shifted to human judgment, focus, and the social systems that decide how increased productivity is used.
Speed, Direction, and “Friction”
- Many argue that speed is only useful if you’re heading in the right direction; otherwise you just get lost faster.
- Others counter that speed makes it cheaper to be wrong: if course-correction is easy and judgment is good, moving fast lets you explore more options.
- Several people say old “friction” (time/effort to code or ship) forced better thinking; AI removes that friction and tempts people into shallow, poorly considered work.
- Metaphors about jars, rocks, sand, and even fish are debated; some see them as helpful reframing, others as empty “wise-sounding” talk.
LLMs, Coding, and “Vibe Slop”
- Strong concern that devs are accepting LLM code with little scrutiny: code “works” superficially but root causes and design are ignored.
- Multiple commenters report that AI can quickly generate PoCs, boilerplate, tests, and one-off scripts, but still can’t replace deep domain understanding, product vision, or fun/game design.
- Others share frustration: long prompts, broken code, and more debugging than writing it themselves. Model quality and usage mode (agent vs chat) matter a lot.
- Some say AI accelerates being wrong and getting stuck in “doom loops” of bad assumptions. Good results require tight scoping, specs in context, restarts, and strong human steering.
- There’s a recurring pattern: AI feels like a productivity high, but may not actually improve long-run output or quality.
Time, Value, and Status
- Disagreement over whether luxury goods are valued for “embedded time” or just as status symbols.
- Several distinguish between market value and personal/emotional value (e.g., a grandmother’s hand‑knit sweater).
- Some broaden the idea: value often reflects how many people, skills, and industries a thing passes through over time.
Work, Productivity, and Capitalism
- Commenters note that productivity gains historically haven’t translated into more leisure for workers; instead they fuel layoffs, quotas, and burnout.
- Tech workers describe coding becoming “sweatshop-like,” with AI used to justify higher expectations rather than better work.
- “Productivity” is seen by some as a weaponized, fuzzy concept used to rationalize squeezing labor.
Trees, Time, and Open Source
- The tree analogy (you can’t fake a 50‑year oak) sparks nitpicking but broadly resonates: trust, community, and mature OSS projects need years.
- Some open‑source maintainers describe decade‑long efforts, slow compounding improvements, and the difficulty of sustaining projects once they gain users.