Judea Pearl fires back in causal inference debate: RCT's new language deserves open scrutiny, not hiding
yudapearl · x · 2026-09-03
A methods debate between Frank Harrell and Judea Pearl: Harrell argued that shared language is good, but mandating one language and calling non-users defective is not, and science should avoid metaphysical unobservables. Pearl responded that causal-inference researchers have long taken other languages seriously — efforts to understand Rubin's potential outcomes and Lewis's closest-world frameworks proved their expressive equivalence to SCMs — so if RCT culture offers a new language for articulating goals and assumptions, its powers and limits should be openly examined, not hidden.
More from Research
- Imitation-learned dexterous policies degrade faster than experts as execution speed rises — UMCP · 2026-09-03
- Meta's Switch Distillation: mid-training distillation boosts reasoning, keeps factual recall — meta · 2026-09-03
- Sebastian Raschka debunks The Information's report on OpenAI Astra's looped transformer architecture — rasbt · 2026-09-03
- Skeptic dissects Astra's rumored recurrent architecture: likely just looping each layer twice — mike64_t · 2026-09-03
- Cosmos grantee to spend 3 months probing LLM reasoning faithfulness — hunarbatra · 2026-09-03
- Video Delta Net speeds up open video generation 75-90x: 14s video in 11s — xiuyu_l · 2026-09-03