8 reasons why people are terrible at predicting the future, from forecasting to black swans

Dan_Jeffries1 · x · 2026-09-09

A thread lays out 8 reasons people fail at predicting the future: even top superforecasters' accuracy collapses past a 3-year horizon; black swan inventions like the printing press can't be seen coming; people change one variable while holding all others fixed (AI advances, nothing else does); they assume unlimited resources and zero real-world friction; domain expertise doesn't transfer to forecasting ability; "Jules Verne syndrome" ignores diffusion curves; exponential curves are mistaken for infinite ones; and compute/memory/learning limits are assumed away.

Original post →

More from AGI Musings

AGI Musings channel →