Engineers will do anything to avoid learning from history
Software’s embrace of AI “agents” is prompting comparisons to long‑established disciplines like project management and engineering, with many arguing that existing methods for requirements, coordination, and risk are being ignored and reinvented under new branding. Commenters debate whether most software developers deserve the “engineer” title, contrasting fields with strict licensing, safety margins, and historical case studies against a tech culture that rewards novelty, hype, and VC‑driven experimentation. Others note that reinventing tools is genuinely enjoyable and sometimes useful, but warn that treating software as exempt from historical and cross‑disciplinary lessons leads to fragile systems and avoidable failures.
Incentives, Hype, and “Novelty”
- Several comments agree that big money flows to things marketed as new and cutting‑edge, not to applying known disciplines correctly.
- VC and IPO dynamics are framed as long‑running “subsidies” that prop up questionable models until broader economic pain hits.
- Some see AI/agent hype as another round of this dynamic, with people retrofitting old pet methods onto the new fad.
Are Software Developers Really Engineers?
- Strong disagreement over whether software developers qualify as “engineers.”
- One side: engineering is defined by formal education, licensing, professional and ethical standards, and safety‑critical responsibility; programmers appropriated the title without that rigor.
- Opposing views: many “real” engineers don’t care about the title; engineering is more a mindset; some empirical polling suggests other disciplines often accept software as engineering.
- Several point out that non‑software engineering also has failures and weak accountability; licensure is not a universal safeguard.
Learning (or Not) from History
- Many argue that ignoring historical lessons is a general human trait, not specific to engineers.
- Others assert that traditional engineering disciplines explicitly teach from past failures, whereas software culture overvalues first‑principles reinvention.
- Some defend reinvention as inherently fun and educational, and resent the idea that only earlier generations got to “discover” things.
AI Agents and Management Analogies
- Some commenters agree with the article’s analogy: orchestrating LLM agents resembles classic management, project planning, and requirements engineering.
- Others argue agent work is meaningfully new: token costs, drift, evaluation, and ultra‑literal “workers” with no common sense differ from human teams.
- A counter‑view says these differences largely map onto existing concerns: time and labor cost, drift from inadequate check‑ins, and overly literal engineers.
Process: Waterfall, Agile, and Specs
- Multiple comments resonate with the idea that using AI effectively is mostly about good specifications; a large share of work becomes up‑front scoping.
- Some caution against literal revival of waterfall or PMBOK; they prefer agile‑style iteration with strong requirements.
- Others note that what is called “Agile” in practice often ignores core ideas like limiting work‑in‑progress and minimizing context switching, especially in meeting‑heavy cultures.