Why Our Perception of Risk Varies: Lessons from The Art of Uncertainty

When COVID-19 spread in early 2020, governments and media gave wildly different risk assessments, prompting the question of why people interpret the same numbers so differently. The article explains that this stems from personal experience, emotions, and biases that shape our perception of uncertainty.

David Spiegelhalter’s 2024 book The Art of Uncertainty: How to Navigate Chance, Ignorance, Risk and Luck systematically breaks down this cognitive mechanism. The core argument is that uncertainty is not an inherent property of the world but a relationship between the observer and the world; probability therefore measures an individual’s ignorance, not objective randomness.

Uncertainty as a Quantifiable Relationship

The book divides its content into four parts. Part 1 distinguishes “known unknowns” from “unknown unknowns,” emphasizing humility in acknowledging what we don’t know. Part 2 builds probability thinking, showing how vague terms like “likely” can be interpreted very differently and urging conversion of linguistic uncertainty into numbers.

Part 3 focuses on Bayesian inference, causal analysis, and prediction, highlighting the principle of updating beliefs with new evidence. Part 4 returns to decision‑making under deep uncertainty, discussing risk communication and global challenges such as climate change and AI.

Describing Uncertainty Numerically

The book contrasts frequentist and Bayesian philosophies. Frequentism defines probability as long‑run frequencies, suitable for large repeatable datasets (e.g., insurance actuarial work). Bayesianism treats probability as a subjective degree of belief, allowing inference without repeated trials via Bayes’ theorem: P(H|E) = \frac{P(E|H)\,P(H)}{P(E)} An example of medical testing illustrates the power of this approach: with a disease prevalence of 1 %, test sensitivity 95 % and specificity 90 %, a positive result yields a posterior probability of only about 8.7 % for actually having the disease, demonstrating the importance of prior probabilities.

Spiegelhalter also applies variance decomposition to sports, finding roughly 40 % of football match outcomes are due to luck rather than skill, and shows how short‑term investment performance can be dominated by chance.

Three Memorable Applications

Shuffling and Permutations: The number of possible orders of a well‑shuffled 52‑card deck is 52! – an astronomically large figure, illustrating that events with extremely low probability can be practically ignored.

COVID‑19 Risk Communication: Translating a 1 % death risk for 60‑year‑olds into “1 out of 100 infected people will die” improves public understanding compared to raw percentages.

Distinguishing Luck from Skill: If each fund manager has a 50 % chance of beating the market each year, about 31 out of 1,000 managers would appear to outperform for five consecutive years purely by luck, underscoring the need for longer evaluation windows.

Practicing the Skill of Managing Uncertainty

The book does not claim to provide “correct answers” but teaches readers to face unknowns honestly. Spiegelhalter’s key recommendations are:

Express uncertainty with numbers rather than vague language.

Update judgments when new evidence arrives and separate luck from ability when evaluating outcomes.

Maintain appropriate humility in predictions instead of over‑confident certainty.

For mathematical modelers, the book serves as a reminder that model outputs always carry uncertainty, and communicating that uncertainty to non‑experts is a crucial part of the work.

Further Reading

Readers interested in deeper exploration can consult introductory Bayesian texts such as Sharon Bertsch McGrayne’s The Theory That Would Not Die , Daniel Kahneman’s Thinking, Fast and Slow , Gerd Gigerenzer’s Simple Statistics , or Spiegelhalter’s earlier work The Art of Statistics . Additional material is available from the Winton Centre for Risk and Evidence Communication at Cambridge.