The Software Engineering Identity Crisis

Perceptions of Change in the Ecosystem

• Some participants feel fewer interesting JavaScript/Node libraries are appearing, possibly due to personal filter bubbles or a broader shift of attention toward AI.
• A few developers no longer open‑source hobby code because they don’t want it “slurped” into LLM training, citing unfairness in how a few companies privatize value from public‑commons code.

AI Coding: Tool, Threat, or Shift in Craft

• Experiences range widely: some say AI is “autocomplete++” that fails on anything hard; others claim they regularly ship full features with minimal bugs using AI.
• Many see AI as great for boilerplate, tests, yak‑shaving tasks, refactors, and quick prototyping, but poor at consistent style, integration with existing utilities, or deep architecture.
• Several argue AI output is “average at best,” good enough for business value but not for performance‑ or reliability‑critical systems; others think that’s exactly where humans will focus.
• A recurring complaint: AI makes coding less fun or erodes the incentive to study deeply, even as it becomes hard to stop using because of productivity and workload pressures.

Identity, Meaning, and “What a Software Engineer Is”

• Some strongly resonate with the article’s “identity crisis”: they enjoy being builders, not managers or prompt‑drivers, and feel like they’re now supervising code rather than writing it.
• Others reject the idea that engineers should embrace “business impact” as part of their core identity; they want to build things, not run companies or become product people.
• Counterpoint: some say software engineering was always about driving business value; real creative fulfillment should come from personal projects, not corporate work.

Hiring, Skills, and the Shape of Future Roles

• Interview frustration is high: leetcode‑style puzzles are seen as detached from real SRE/engineering work. Some argue for allowing AI in interviews to test judgment, not recall.
• There’s concern that AI‑assisted coding increases demand for experienced “supervisors” while eroding junior roles, risking a broken training pipeline.
• Predictions differ: some foresee broad, hybrid roles spanning FE/BE/infra/product; others expect bifurcation into “manager‑engineers” with breadth plus a smaller cohort of deep specialists.

Abstractions, Code Quality, and Machine‑Readable vs Human‑Readable

• One thread argues that “good code” is largely about human maintainability; if machines are the primary readers, very different styles or even non‑human‑readable IRs might emerge.
• Others counter that abstraction and structure are valuable for machines too, and that opaque, non‑human representations raise hard questions about debugging, security, and control.