Why LLMs can't make your code simpler, per Naur's "Programming as Theory Building"
math_rachel · x · 2026-08-19
Answer.AI publishes Pol Alvarez Vecino's essay applying Peter Naur's classic "Programming as Theory Building" to LLM coding limits:
- A program is the Theory in engineers' minds — the constraints, trade-offs and design rationale; code and docs are incomplete downstream artifacts.
- LLMs raise codebase complexity: duplicated methods, over-defensive code for impossible edge cases, premature optimization. Metrics like LoC can't fix it because the target is Theory complexity, which is subjective and unmeasurable.
- Conclusion: further training can't solve this — the needed information doesn't live in the code.
More from coding & agent
- Everything crashes, but Omarchy sends crash reports to Claude in 1 click — indragie · 2026-10-16
- Six-month prediction: hiring for generic agent harnesses built from scratch will cool — yi111 · 2026-10-16
- Omarchy ISO wouldn't boot on NVIDIA, so he had Claude patch it — indragie · 2026-10-16
- Dev uses Grok to clean up 3D printer firmware, catches it sneaking in nonsense code — yacineMTB · 2026-10-16
- A rude AI agent took 89% of a shared resource in 102-run multi-agent personality experiment — Brave-History-6502 · 2026-10-16
- Guardrails can make risky delegation work — but it takes a full-time team, says iandanforth — iandanforth · 2026-10-16