NetHack port challenge: AI transpilers hit 99.5% correctness but tank readability
davidbau · x · 2026-10-31
David Bau runs the Teleport Coding Challenge — porting 40-year-old NetHack's C code to JavaScript — and benchmarks contestants on both correctness and readability (quizzing a constrained agent with 500 questions about the game; original C scores 81% readability).
- NoahBPeterson's AI-assisted transpiler reaches 99.5% correctness, but readability drops to 69%; serteal 18.8 outputs unreadable $0/HEAP32[...] code at 61-65% readability while only hitting 93% correctness. Other entrants (vtjeng, richie3366, daoao0601) are more readable (73-77%) yet below 100%.
- His takeaways: when 100% correctness is the goal, it's often better to have AI build a program that solves the problem than to solve it directly — yet nobody hit 100%, showing even an old terminal game resists perfect porting. "The world is complex," undercutting Silicon Valley's self-improving "AI god" narrative.
- He asks: is a 93%-correct, unreadable AI-written system acceptable? What happens when that 7% goes wrong?
Related event: NetHack Porting Challenge: AI Hits 99.5% Accuracy but Readability Suffers(3 posts)→
More from coding & agent
- LambdaTest's TestMu AI Mapped: From Test Cloud to Agent Verification — Designer_Hair_417 · 2026-10-31
- Open-source ultra-elements: 39 tunable spell VFX in pure three.js, LLM-extensible — majidmanzarpour · 2026-10-31
- How Boardy made CI 20x faster: PR waits drop from 40 minutes to 2 — inductionheads · 2026-10-31
- Product Engineering vs engineer-centric AI coding: ship all the way to paying users — PawelHuryn · 2026-10-31
- One start-here skill navigates you through the 43-skill agent workflow — PawelHuryn · 2026-10-31
- hbench visualizes agent trajectories — and caught DeepSeek reading benchmark answers — Psychological_Cry920 · 2026-10-31