Economist: most published math papers are right, econ papers are wrong — and that's why AI lags
paulnovosad · x · 2026-10-10
Economist Paul Novosad argues AI finds it harder to surpass humans in economics than in math because most published math papers are correct while most econ papers are wrong. Worse, the written record misleads: it suggests diff-in-diff studies failed over standard errors or negative weights, when the real issue is that most DiD studies were poorly conceived and violated the method's assumptions — knowledge that lives mainly in researchers' heads. So even very good models will keep giving doctrinaire advice like "use distance to coast as an IV."
More from AGI Musings
- The average technologist isn't a technologist: on loving outputs, not the craft — fkasummer · 2026-10-10
- Anthropic researcher parodies singularity-worship rhetoric in viral rant — suchenzang · 2026-10-10
- Would you trade thousands of researcher jobs for an AI cancer cure? — VraserX · 2026-10-10
- Economist: hyper-accurate prediction models will reshape economics like asset pricing — soumitrashukla9 · 2026-10-10
- Miles Brundage: No Frontier Lab Has Even a Concept of a Plan for Safe Self-Improvement — Miles_Brundage · 2026-10-10
- Lean maxxing as hyperutilitarian superintelligence — and a coming aesthetic revolt — akbirthko · 2026-10-10