We promoted developers into architects without informing them
AI coding tools are shifting software engineers away from writing code toward architecture, systems design, and broader responsibilities like product management, QA, and ops—often without changes in title or pay. Many commenters argue that good engineers always did this architectural work, while others warn that organizations treating developers as mere “coders” are now exposed and at risk of replacement by LLMs and agents. The thread also raises longer-term concerns about how juniors will learn, whether “architect” as a separate role still makes sense, and how AI-driven deskilling could erode job quality and compensation across the industry.
Role of “Architect” vs Developer
- Many argue architecture has always been part of competent software engineering; calling this a “promotion” is misleading.
- Others note that in some orgs architect is a distinct, higher‑paid, low‑coding role; in others it’s a “sinecure for failed programmers” or pure “marketecture” (diagrams, PowerPoints, vendor tech choices).
- Several describe a healthier model where architecture is shared across senior engineers / tech leads, with architects focusing on broad principles, boundaries, and dispute resolution, not micromanaging.
AI Coding Tools and “Vibe Coding”
- Common pattern: devs prompt LLMs to generate large chunks of code (“vibe coding”) without understanding performance, data access, or security implications.
- Some see the article’s logging example as simply bad development that would have happened without AI; others say LLMs do tend to over‑log, over‑authenticate, and overuse APIs unless guided.
- Multiple commenters report frontier models can already profile, optimize, and critique code if asked; several say models will improve at architecture and tradeoff reasoning.
Performance, Security, and Tradeoffs
- Broad agreement that architecture is mostly about tradeoffs: performance vs security, CAP, latency vs complexity, etc.
- Critics warn that “hill-climbing” micro‑optimizations via AI is not architecture and leads to local minima and bloated systems.
- Others say AIs can be instructed to care about performance/security, but only if humans provide context (scale, workload, constraints).
Impact on Careers, Juniors, and Labor
- Many report being “promoted” in title only: now simultaneously architect, PM/BSA, QA, and DevOps with no extra pay.
- Widespread fear that hands‑on coders with little design/requirements responsibility are highly replaceable by agents.
- Strong concern about juniors: if they mostly orchestrate AI instead of coding, they may never develop the intuition needed for future senior/architect roles.
- Several tie this to broader labor history: rationalization and deskilling usually reduce worker bargaining power and pay; calls for collective action, licensing, or workplace democracy appear, alongside skepticism about unions’ effectiveness in a globally mobile, AI‑mediated industry.
Quality, Responsibility, and Organizational Patterns
- Contributors contrast “ticket‑banger” cultures (detailed specs handed down, no pushback) with environments where engineers own requirements, design, and tradeoffs.
- Many say AI will flatten architect/dev boundaries: whoever specifies and reviews the system owns the outcome, especially when implementation is trivial to automate.
- Some are optimistic: AI frees them from rote coding to focus on understanding users, systems, and architecture; others feel the job has become shallow, stressful “foible management” of AI output and corporate risk.