Tuesday, March 18, 2025

Review of The Signal and the Noise: Why So Many Predictions Fail - But Some Don't by Nate Silver

This book review was written by Eugene Kernes   

Book can be found in: 
Book Club Event = Book List (08/16/2025)
Intriguing Connections = 1) How Does Data Get Use, And Misused?, 2) The Style of Math


Watch Short Review

Excerpts

“Meanwhile, exposure to so many new ideas was producing mass confusion.  The amount of information was increasing much more rapidly than our understanding of what to do with it, or our ability to differentiate the useful information from the mistruths.  Paradoxically, the result of having so much more shared knowledge was increasing isolation along national and religious lines.  The instinctual shortcut that we take when we have “too much information” is to engage with it selectively, picking out the parts we like and ignoring the remainder, making allies with those who have made the same choices and enemies of the rest.” – Nate Silver, Introduction, Pages 3-4



“The most calamitous failures of prediction usually have a lot in common.  We focus on those signals that tell a story about the world we would like it to be, not how it really is.  We ignore the risks that are hardest to measure, even when they pose the greatest threats to our well-being.  We make approximations and assumptions about the world that are much cruder than we realize.  We abhor uncertainty, even when it is an irreducible part of the problem we are trying to solve.” – Nate Silver, Chapter 1: A Catastrophic Failure Of Prediction, Page 20


 

“This book advises you to be wary of forecasters who say that the science is not very important to their jobs, or scientists who say that forecasting is not very important to their jobs!  These activities are essentially and intimately related.  A forecaster who says he doesn’t care about the science is like the cook who says he doesn’t care about food.  What distinguishes science, and what makes a forecast scientific, is that it is concerned with the objective world.  What makes forecasts fail is when our concern only extends as far as the method, maxim, or model.” – Nate Silver, Chapter 12: A Climate Of Healthy Skepticism, Page 403


Review

Is This An Overview?

Having a lot of information does not mean there is a lot of validity in the information.  There is difficulty in understanding large quantities of information, and difficult to differentiate between useful information from misinformation.  While people want useful information, want the Signal, much of the information is not useful, information that is noise.  Noise distracts people from the Signal.  The quality of predictions, or forecasts, depends on filtering the Signal from the Noise. 

 

The data, the evidence, the numbers do not represent themselves.  The evidence is represented by people, who tend to favor the evidence they want to hear.  Confirming their views which limits their decisions, and causes them to miss evidence that can affect the decisions being made.  People are biased, and therefore develop biased predictions.  To improve data-driven predictions, people need to improve their ability to sort the information. 

 

Prediction failures tend to have features in common such as focusing on what is wanted rather than what is, ignoring difficult to measure risks, making inappropriate approximations and assumptions, and misunderstanding uncertainty.  Forecasts tend to improve when people think of various alternative views, and update their views to new information. 

 

What Is Forecasting?

Models are a tool to represent the complexities of reality, they do not substitute for reality.  A prediction is a definite and specific statement of what might happen, while a forecast is a probabilistic statement of what might happen.  Risk is knowing what the options are, while uncertainty is not knowing the options or information that can affect the options. 

 

Systems which are dynamic and nonlinear (chaos theory), made predictions difficult.  People can change their behavior to a prediction, therefore changing the prediction itself into a self-fulfilling prediction as people support the claims or a failed prediction by avoiding the claims.

 

Good forecasts are those which over time make more correct predictions.  A Bayesian analysis is a method of updating beliefs, as the method gets the person closer and closer to the truth. 

 

Caveats?

Advice for how to improve decisions are described using examples, the advice is hidden within the examples.  The examples are noise that the reader needs to engage with to find the signal.   The value of the examples depends on the interests of the reader.  There is not much of a systematic analysis, a lack of a summary for the advice.  


Questions to Consider while Reading the Book

•What is the raison d’etre of the book?  For what purpose did the author write the book?  Why do people read this book?
•What are some limitations of the book?
•To whom would you suggest this book?
•Are statisticians able to predict outcomes? 
•Did Nate Silver predict political outcomes? 
•How did the printing press change the value of information?
•What do people want to hear? 
•Who speaks for numbers? 
•What is the quality of medical hypotheses?  
•What is information overload? 
•What is the rate of change for the noise and the signal? 
•What are common causes of prediction failures?  
•What is the difference between risk and uncertainty? 
•What is the difference between a good and bad forecast?
•What is the difference between a prediction and a forecast? 
•What do rating agencies do? 
•What did the rating agencies predict? 
•What does it mean to be out of sample?
•How does confidence effect predictions? 
•Who are hedgehogs and foxes?
•Who is more likely to appear on major media explaining their views? 
•What is the value of news? 
•What makes for a valuable baseball player? 
•What happened to weather forecasts? 
•What is Laplace’s Demon? 
•What is the value of a computer? 
•What is chaos theory?
•What happens when people do not trust their sources of information?
•Can earthquakes be predicted? 
•What is overfitting? 
•hat is the value of economic forecasts?  
•Can data be ignored? 
•Can the spread of diseases be predicted? 
•What is a self-fulfilling prediction?
•What is the value of a model? 
•How to gamble on sports? 
•What is the purpose of a Bayesian analysis?
•What is the frequentist approach? 
•How did chess change? 
•What is the Mechanical Turk?
•How to play poker? 
•Can the stock market be predicted?
•What are the forms of an efficient-market? 
•How does herding effect the quality of forecasts? 
•What do meteorologist think about climate change? 
•What is the use of skepticism? 
•Could 9/11 be predicted?
•Could the attack on Pearl Harbor be predicted? 


Book Details
Publisher:               Penguin Books [Penguin Group]
Edition ISBN:         9780143125082
Pages to read:          459
Publication:             2015
1st Edition:              2012
Format:                    eBook 

Ratings out of 5:
Readability    5
Content          5
Overall          4