Code Quality Learned via Long Horizon RL, Not Granular Graders
dylanbowmanSF · x · 2026-08-29
The author argues that code quality gets learned through long-horizon tasks because high-quality code is implicitly useful for the goal. Even if accumulating "slop" causes agents to stumble now, the solution is not more granular graders, but more ambitious RL tasks.
While agreeing that continual learning is technically challenging, the author believes it clearly unlocks capabilities needed to automate most labor. In a reply, the author notes that the difficulty lies in whether the judge (LLM) can capture subtle qualitative judgments, which depends on pretraining scale and data coverage.
Related event: The S-Curve Debate: Can AI Progress Keep Stacking New Paradigms(16 posts)→
More from coding & agent
- Marc Lou praises PostShiba, a cute indie alternative to AWS SES for email infra — marclou · 2026-10-16
- RExBench: best coding agent completes only 33% of autonomous research extensions — najoungkim · 2026-10-16
- Wiring Pstack skills from a Grok bot into Hermes agent loops — alexcovo_eth · 2026-10-16
- past.dev launches long-term memory API for AI agents, tops BEAM public benchmark — thisiskp_ · 2026-10-16
- Google's CLASHarness benchmarks multi-agent conflicts in computer use, finds frontier VLMs struggle — google · 2026-10-16
- mitsuhiko's AI agent solved multiplayer with PeerJS before he could explain it — mitsuhiko · 2026-10-16