Slow Coding and the Decline of Technical Books
A reactionary pivot is emerging against the "slop cannon" of AI-assisted development. While many engineers use LLMs to rapidly output low-quality code, a growing contingent is advocating for "slow coding"—using AI agents as meticulous, pedantic code reviewers to improve overall codebase health.12 This shift highlights a deeper anxiety: as developers trade the manual "struggle" of coding for high-abstraction convenience, they risk losing the foundational mental models required to guide and debug these systems.
This anxiety is mirrored in the rapid decline of the physical programming book. With sales of technical books plummeting as chatbots and coding assistants take over, the industry is shifting away from the slow, structured learning that forced deep comprehension. Critics warn that "knowledge is the residue of typing," and that by optimizing away the friction of learning and coding, developers are becoming cognitively brittle, unable to effectively steer the black box.
-
An instance of The rush to automate coding inevitably forces developers back to slow, deliberate craftsmanship. — In response to high-abstraction AI code generation and falling technical book sales, developers are championing slower, highly deliberate coding practices to preserve critical mental models. ↩︎
-
An instance of You cannot manage AI-generated codebase debt without a systematic move to formal specification. — Developers must pivot from raw output generation to systematic code verification and design guidance to prevent severe context degradation. ↩︎