{"slug": "what-objective-is-reasoning-solving", "title": "What objective is reasoning solving?", "summary": "A blog post by Alex Wang argues that reinforcement learning for reasoning models maximizes the probability of finding any path to a correct answer rather than making individual chains of thought converge, so longer chain generation does not necessarily improve accuracy. The post concludes that forcing longer reasoning traces is a weak control knob for production systems and recommends sampling multiple paths, using verifiers and rerankers, and treating chain-of-thought text as an unreliable artifact rather than a faithful execution trace.", "body_md": "[Hacker News](https://alexwang.ai/posts/reasoning-as-pathfinding/)\n\n### What objective is reasoning solving?\n\nWhich summary reads better? Pick one — models revealed after.Both summaries are AI-generated.\n\nReinforcement learning maximizes the probability of a model finding any path to a correct answer, not the probability of individual paths converging to a correct answer, which means longer chain generation doesn't necessarily improve accuracy. This matters because it implies that models may not improve with more computation per chain, limiting the effectiveness of simply increasing chain length. Models may still improve with more samples, as they can find alternative paths to the correct answer.\n\nReasoning RL is better understood as shifting probability mass toward token paths that end in verifiably correct answers, not making any single chain-of-thought converge as it gets longer. For production systems, forcing longer reasoning traces is a weak control knob and can degrade outputs; invest instead in sampling multiple paths, using verifiers/rerankers, and treating CoT text as an unreliable artifact rather than a faithful execution trace.", "url": "https://wpnews.pro/news/what-objective-is-reasoning-solving", "canonical_source": "https://www.snipvote.com/story/cmwd84ih9000c3w1krj8gd6j0", "published_at": "2026-11-12 07:52:08.373238+00:00", "updated_at": "2026-11-12 07:52:10.118435+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "large-language-models", "ai-research"], "entities": ["Alex Wang"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/what-objective-is-reasoning-solving", "markdown": "https://wpnews.pro/news/what-objective-is-reasoning-solving.md", "text": "https://wpnews.pro/news/what-objective-is-reasoning-solving.txt", "jsonld": "https://wpnews.pro/news/what-objective-is-reasoning-solving.jsonld"}}