N N Taleb's Probability Questions (UNOFFICIAL) Tail risk of contagious diseases: a conversation with coauthor Pasquale Cirillo (2020)
Nov 11, 2025
Pasquale Cirillo, a statistician specializing in extreme value theory, joins to discuss the intricate relationship between tail risks and pandemics. They delve into why tail properties matter more than point forecasts for decision-making. Cirillo critiques the common use of unreliable historical data and emphasizes the importance of probability training over statistics. The conversation also highlights the rationality of caution in the face of contagion risks, along with the need for policy to focus on tail risk rather than misleading predictions.
AI Snips
Chapters
Transcript
Episode notes
EVT Began With Watermelons
- Pasquale recalls extreme value theory starting with studies of watermelon sizes by Fisher and Tippett.
- Taleb links that history to Dutch hydrology and the Delta project origins of EVT.
Pandemics Exhibit Extremely Fat Tails
- Fat tails make tail observations dominate inference, so few extreme events outweigh many small ones.
- Taleb and Pasquale find the pandemic tail exponent (alpha) below 1, implying infinite mean behavior in practice.
Transformations Rescue EVT For Bounded Risks
- Transformations let extreme value theory handle bounded but heavy-tailed variables robustly.
- Taleb and Cirillo use a phi-transform plus inverse mapping to estimate tail parameters despite upper bounds.