Hacker News debates coding without AI agents

The growing developer resistance to AI agents is not mere nostalgic luddism, but a structural reaction to productivity theater and codebase degradation in corporate engineering.

• –**Agent management fatigue**: Developers report that orchestrating, babysitting, and auditing autonomous coding agents feels like an unwanted transition into middle management rather than software engineering.
• –**Codebase bloat and technical debt**: Teams relying heavily on AI agents face massive diff churn, with small feature requests triggering dozens of extraneous file modifications, redundant tests, and tangled abstractions.
• –**Erosion of code ownership**: In automated workflows where implementation details are deferred to LLMs, codebases quickly become black boxes that no single developer understands or feels accountable for maintaining.
• –**Top-down mandates versus practical utility**: Corporate pushes to mandate AI usage and track token metrics often incentivize automated volume over system reliability, fueling frustration among engineers who value craft.