Sam Arbisman

Fallible: The Risk of Predictions and the Limits of Expertise

Sam Arbisman, Zachary Carabon, Bethany MacLean, John Rohal, Paul Kudrowski

A diverse panel of experts from finance, journalism, and science convenes to dissect the inherent limitations of predicting complex systems, highlighting the "Gentleman Amnesia Effect" where trust in expertise overrides direct experience of fallibility. Speakers argue that tightly coupled models fail in social and economic contexts characterized by nonlinear interactions, necessitating a shift from precise forecasting to defining bounds on surprise and prioritizing transparent data inputs. Ultimately, the discussion proposes a framework for evaluating predictors based on failure rates and incentives while warning that the human desire for binary certainty often obscures the fluid, probabilistic reality of financial and scientific outcomes.