08-17-2026, 04:34 PM
Uncertainty: The Soul of Modeling, Probability & Statistics
Author: William M. Briggs
Publication date: July 2016
Publisher: Springer International Publishing
William M. Briggs’s Uncertainty is less a conventional statistics textbook than a philosophical challenge to the way probability and statistical modeling are commonly practiced. Briggs argues that probability is fundamentally a branch of logic and is always conditional on specified information. Uncertainty, in his account, represents what we know—or do not know—rather than an intrinsic property residing in objects themselves. Starting from questions about truth, logic and induction, he develops this position through discussions of probability, randomness, causality and statistical models. Mathematics is present, but the emphasis is primarily conceptual rather than computational.
The most provocative part of the book is Briggs's attack on mainstream statistical practice. He is strongly critical of p-values, null-hypothesis significance testing, excessive attention to model parameters, and attempts to infer causation merely from statistical models. He advocates what Springer describes as a “Third Way” beyond the usual frequentist-versus-Bayesian division. Models, in this view, should primarily make testable predictions about observable outcomes, and their usefulness should ultimately be assessed against reality rather than by whether estimated parameters achieve conventional statistical significance. Briggs summarizes two of his central principles particularly succinctly: all probability is conditional, and probability itself does not tell us what decision to make.
The result is deliberately controversial. It should therefore not be approached as a neutral introduction to statistics: even a Mathematical Association of America review characterizes Briggs's positions as unusually strong compared with mainstream probability and statistics. Its value lies precisely in forcing statistically trained readers to reconsider assumptions that are often taken for granted—what probability actually means, what a model can legitimately tell us, whether statistical association provides evidence of causation, and whether conventional significance testing answers the scientific questions researchers really care about. For readers interested in the foundations and philosophy of statistics, rather than simply learning statistical techniques, Uncertainty offers an unusually provocative perspective.
Key takeaways
Goodreads — Uncertainty
Author: William M. Briggs
Publication date: July 2016
Publisher: Springer International Publishing
William M. Briggs’s Uncertainty is less a conventional statistics textbook than a philosophical challenge to the way probability and statistical modeling are commonly practiced. Briggs argues that probability is fundamentally a branch of logic and is always conditional on specified information. Uncertainty, in his account, represents what we know—or do not know—rather than an intrinsic property residing in objects themselves. Starting from questions about truth, logic and induction, he develops this position through discussions of probability, randomness, causality and statistical models. Mathematics is present, but the emphasis is primarily conceptual rather than computational.
The most provocative part of the book is Briggs's attack on mainstream statistical practice. He is strongly critical of p-values, null-hypothesis significance testing, excessive attention to model parameters, and attempts to infer causation merely from statistical models. He advocates what Springer describes as a “Third Way” beyond the usual frequentist-versus-Bayesian division. Models, in this view, should primarily make testable predictions about observable outcomes, and their usefulness should ultimately be assessed against reality rather than by whether estimated parameters achieve conventional statistical significance. Briggs summarizes two of his central principles particularly succinctly: all probability is conditional, and probability itself does not tell us what decision to make.
The result is deliberately controversial. It should therefore not be approached as a neutral introduction to statistics: even a Mathematical Association of America review characterizes Briggs's positions as unusually strong compared with mainstream probability and statistics. Its value lies precisely in forcing statistically trained readers to reconsider assumptions that are often taken for granted—what probability actually means, what a model can legitimately tell us, whether statistical association provides evidence of causation, and whether conventional significance testing answers the scientific questions researchers really care about. For readers interested in the foundations and philosophy of statistics, rather than simply learning statistical techniques, Uncertainty offers an unusually provocative perspective.
Key takeaways
- Probability is conditional: probabilities only make sense relative to specified evidence or assumptions.
- Probability ≠ causality: statistical relationships alone cannot establish what causes what.
- Briggs rejects conventional significance testing: particularly routine reliance on p-values and hypothesis tests.
- Models should face reality: the important question is how well a model predicts observable outcomes, not merely whether its parameters appear statistically significant.
Goodreads — Uncertainty
┌────────────────────────────────┐
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘
│ KONSTANTINOS MICHAILIDIS │
└────────────────────────────────┘