Questions for you:
- When have you made important decisions based on probabilistic forecasts rather than certain predictions – and how did you handle the inherent uncertainty?
- In what areas of your life or work do you demand more precision than is actually achievable, and what would change if you accepted irreducible uncertainty?
- How do you balance the need to act decisively with the reality that you cannot eliminate all uncertainty through additional data or analysis?
- Think back, when has your attempt to gather “just one more piece of information” before deciding actually been a way of avoiding the discomfort of making choices under uncertainty?
Organisational applications:
Project forecasting and planning: Abandon single-point estimates for project timelines and budgets. Present ranges of probable outcomes with explicit confidence levels, helping stakeholders understand that precision beyond certain thresholds is illusory regardless of planning effort that’s been invested.
Strategic scenario planning: Develop multiple plausible future scenarios rather than attempting to predict the most likely outcome. Assess strategic options based on robustness across various probable scenarios rather than optimising for a single predicted future that may never materialise.
Risk assessment frameworks: Distinguish between measurable risks (where historical data enables probability estimates) and genuine uncertainty (where the future is fundamentally unpredictable). Adjust decision processes accordingly – don’t pretend uncertain situations are merely risky ones requiring more data.
Meeting culture and decision-making: Challenge requests for “more precise forecasts” by explicitly stating the limits of prediction for complex systems. Shift conversations from “what will happen” to “what might happen and how we’ll respond” – building adaptive capacity rather than predictive accuracy.
Further reading:
Chaos theory and prediction limits
- Chaos: Making a New Science by James Gleick – accessible introduction to chaos theory explaining why tiny measurement errors compound exponentially in complex systems, making long-term weather prediction mathematically impossible regardless of computational power.
- The Weather Machine by Andrew Blum – explores the global infrastructure of weather forecasting, documenting how forecasters acknowledge and communicate uncertainty whilst the public often demands impossible precision.
- “Deterministic Nonperiodic Flow” by Edward Lorenz in Journal of the Atmospheric Sciences (1963) – the foundational paper establishing sensitive dependence on initial conditions, demonstrating mathematically why perfect long-term weather prediction requires perfect measurement which remains perpetually unattainable.
Hurricane forecasting and communication
- Divine Wind: The History and Science of Hurricanes by Kerry Emanuel – meteorologist’s explanation of hurricane physics and why prediction beyond five days remains fundamentally limited by chaotic atmospheric dynamics rather than inadequate technology.
- National Hurricane Center’s forecast verification reports – actual historical data showing how forecast accuracy degrades with time, providing concrete evidence of prediction limits that no technology improvement can eliminate.
Decision-making under uncertainty
- Superforecasting by Philip Tetlock and Dan Gardner – documents research on forecasting accuracy, showing that good forecasters explicitly acknowledge uncertainty ranges rather than providing false precision, and update predictions as new information emerges.
- The Black Swan by Nassim Nicholas Taleb – argues that the most consequential events are inherently unpredictable, suggesting we should design systems for robustness to unknown scenarios rather than attempting precise prediction of known risks.
About the image
It’s not a hurricane. It’s just some clouds, off the Suffolk Coast, at Southwold. One of my most favourite places to stare at where the sky meet the sea.
Photo and photomontage Matt Ballantine 2026.
Random the Book 