Claude Code + Codex re-tested 8 famous research findings overnight — only 1 held up
beglen · x · 2026-10-27
beglen had Claude Code + Codex re-test 8 famous research findings on newer data to the highest academic standards, completing all overnight with a publishable one-page infographic each. Before pulling any data, he wrote a short analysis plan per study (what to test, how, and what counts as holding up/weaker/failing/unclear) and committed it to a private GitHub repo for an independent timestamp. Verdict: 1 finding held up (with caveats), 1 weakly, 1 failed, 1 has mostly stopped being true, and 4 couldn't be settled either way — which he calls the honest headline: most of these effects were never as solid as claimed. A response to Ethan Mollick's suggestion.
More from coding & agent
- xAI releases free Grok Bot course: 18 real-team playbooks, ~3 hours, covering sales to engineering — victor_explore · 2026-10-27
- 100+ commits per day is the new bare minimum in the AI coding era — damnGruz · 2026-10-27
- Rocket launches open app platform to help vibe coders distribute AI apps — alexmacgregor__ · 2026-10-27
- Turn product photos into interactive 3D-style displays with AI video frames, no 3D modeling — aigclink · 2026-10-27
- Full walkthrough: interactive product displays from photos using GPT-6 Astra and AI video as UI — MaziyarPanahi · 2026-10-27
- DeepSeek leads four Qwen3.5-9B models to reverse engineer binaries — team mode costs more than solo — HauntingMoment · 2026-10-27