Carson Gross, creator of htmx, argues that while AI has made code generation cheap and fast, understanding code has become the scarce and expensive resource. He warns that LLMs, unlike compilers, produce non-deterministic output faster than humans can comprehend, creating a complexity trap similar to the Sorcerer's Apprentice. His proposed response is the 'subtractive, constraining engineer' — someone who says no to unnecessary code, simplifies systems, and removes layers rather than adding them. The key insight is that complexity grows geometrically with system size, and LLMs are prolific coders with no fear of complexity.

5m read timeFrom htmx.org
Post cover image
Table of contents
Understanding is Expensive(er)The Sorcerer’s Apprentice TrapComplexity: Still BadThe Subtractive, Constraining Engineer
12.4K Impressions1 Comment