A Flaw in Human Judgment
We’re scratching the surface here. If you don’t already have the book, get the audiobook for free on Amazon to learn the juicy details.
Daniel Kahneman, Olivier Sibony, and Cass R. Sunstein’s Perspectives
Daniel Kahneman is an American-Israeli psychologist, pioneer of behavioral economics, and 2002 Nobel Prize Laureate. Foreign Policy magazine recognized Kahneman among top global thinkers in 2011, and The Economist named him one of the most influential economists in 2015. Kahneman is a professor emeritus at Princeton University’s Woodrow Wilson School, as well as a founder of the consulting company TGG Group. His book Thinking, Fast and Slow (2011) on cognitive biases and errors in decisions became a New York Times bestseller.
Olivier Sibony is a business consultant and strategy professor holding a Ph.D. degree from Paris Sciences et Lettres University. Having spent 25 years as a senior partner at McKinsey & Company, he is currently an Affiliate Professor of Strategy at HEC Paris and an Associate Fellow at Saïd Business School, Oxford University. Sibony is a co-author of numerous publications such as Harvard Business Review, as well as a book on decision-making traps You’re About to Make a Terrible Mistake!
Cass R. Sunstein is a professor at Robert Walmsley University at Harvard Law School as well as a director of the Program on Behavioral Economics and Public Policy. He used to be an Administrator of the White House Office of Information and Regulatory Affairs under President Obama. From 2016 to 2017, he was one of the members of the Defense Innovation Board at the US Department of Defense. Cass Sunstein is an author of many books, articles, and even law reforms. Two of his publications – The World According to Star Wars and Nudge – are highly-acclaimed New York Times bestsellers.
Introduction to Noise
When people make decisions, they can rarely avoid errors. Many of them are attributed to our biases, whether we realize it or not. However, one more factor comes into play whenever we form judgments – noise.
Noise forces two doctors to make different decisions when they examine the same patient. Noise is responsible for different sentences passed for the same crime by two independent judges or even the same judge on various occasions. Noise accompanies interviewers when they talk to job applicants. It is because of noise we get different results in situations when they must be identical.
Noise: A Flaw in Human Judgment is an attempt to define noise, reveal its source and ways in which it impacts our decisions. The book also proposes a noise audit that relies on measuring the degree of variability. Along with that, it provides practical advice on how to reduce noise using decision hygiene techniques.
StoryShot #1: Noise vs Bias
The first chapter of Noise: A Flaw in Human Judgment describes the differences between noise and bias. Noise is an unwanted variability in professional judgment. In other words, it represents insufficient consistency in decision-making. Bias, in contrast, is rather an individual’s tendency to use the same patterns of decision-making in similar situations. Bias has consistency but is not able to arrive at the correct result. Despite major dissimilarities between these two concepts, both represent errors in judgments.
The authors uncover shocking truths – organizations, whether they be public or private, are subject to noise. A study examining 1.5 million court cases discovered that noise often impacts judges’ decisions. Judges tend to pass harsher sentences in the days following their local football team losses. In the same vein, they become more lenient when their teams emerge victorious. Evidence shows that sentencing decisions vary substantially for the same crimes. Discrepancies can be observed in the decisions of the same judge, as well as in the decisions of different judges having similar cases.
An example of noise in the private sector can be seen in the way insurance companies determine premium rates. When underwriters assessed risks for the same group of cases, the rates they suggested fluctuated within a dramatic range. Some experts believed that $9500 would be a reasonable rate while the estimates of others showed $16,700 – that’s a 55% difference!
StoryShot #2: Noise Audit
If you ask the same insurance company about differences in premium rate estimations before carrying out these estimations, they will say that variability is going to be around 10%. This figure sounds reasonable. However, the factual discrepancy was 55%. The authors call the possibility of measuring variabilities a noise audit.
Again, if you ask a judge if he/she expects the same decision from another experienced judge, the answer is going to be “pretty much the same.” In reality, variabilities are vastly greater than people expect them to be.
The authors recognize two types of noise. Occasional noise happens when factors such as a football team’s performance or part of the day impact decisions of a person or a group on various occasions. A noise audit is able to recognize this type and help tackle it. Another type – system noise – describes unwanted variabilities that occur when a group of experts tries to individually assess the same events. This type of noise is harder to deal with. It requires greater “decision hygiene”, i.e. noise reduction methods.
StoryShot #3: Noise Hygiene
The authors compare decision hygiene to regular hygiene. When you wash your hands, you don’t immediately see the benefits. You are not aware of the germs you get rid of and the issues they may cause. Still, hand washing protects you from negative consequences. Similarly, decision hygiene protects you from external forces that introduce variabilities to your decisions.
To implement decision hygiene, Noise: A Flaw in Human Judgment suggests breaking a matter into small units and dealing with them independently. We can see the effect of decision decomposing on the example of a corporate merger case study.
Normally, the board of directors delegates bankers or executives to prepare a presentation on the pros and cons of a merger. In this case study, the CEO asked selected senior executives to provide their opinion on several aspects of the merger. Each team had to assess a single unique aspect (for instance, financial benefits, quality of human resources, etc.). Teams didn’t know which assessments their counterparts came up with. Therefore, their judgments weren’t impacted by the opinions of others. At this point, the authors introduce another term – excessive coherence – which describes noise that occurs when we subconsciously support other people’s decisions instead of making our own independent ones.
Besides making decisions as independent as possible, the book highlights the importance of delaying global evaluations until the end. If individuals make decisions early on, they tend to rely on their intuition. However, if we give ourselves more time to weigh facts, we are capable of eliminating the randomness of intuition-based decisions.
Another way to reduce noise is to introduce algorithms and rules that guide decision-making. Still, even the strictest and clearest rules have their drawbacks. The ways in which we present information to algorithms can feature a significant degree of noise and bias.
StoryShot #4: Noise Optimization
The authors acknowledge that noise optimization comes at a price. This price includes not only the financial aspect, but also other resources such as time. The book illustrates noise over-optimization with a story about a company that introduced an annual employee review. It created a feedback questionnaire so complicated (it had 11 dimensions and forty-six ratings) that the review process turned into a disaster.
Before implementing noise elimination techniques, one should weigh their pros and cons. We accept a certain degree of noise when grading a fifth grader’s essay. However, when it comes to a senior’s college application, we should strive to bring the noise down. This is because responsibility is much higher when an applicant’s fate is at stake. So, to help make a decision fair, a college may need five independent people using 10 criteria to assess an application instead of a single person who relies on his/her impressions.
While recognizing that some rules are needed, the authors realize they may entail a loss of dignity. It happens when individuals are treated like numbers. The former CEO of General Electric had a rule to yearly fire the least-performing employees. Some of them still did their job, well, yet they couldn’t avoid firing. The book describes this situation as forcing a rule. Although relative scales and judgments are helpful, companies can’t rely on them exclusively. They need to find a balance between relative and individual ratings.
Perfect fairness is illusory. Sometimes we have to accept some noise in order to achieve progress. Algorithm-based digital body scans to detect melanoma are great, but in many cases, we would rather go to a brilliant doctor. Still, a doctor who benefits from algorithms to make judgments is even better.
Final Book Summary and Review of Noise by Daniel Kahneman
If we are subject to noise, our decisions are like a lottery – we never know which way they’ll go. Even if we manage to get rid of biases, various subjective factors still influence outcomes. Matters get worse when noise creeps into important institutions. We expect medical, child custody, forensic and court decisions to be fair and consistent, but they are shockingly variable.
Noise: A Flaw in Human Judgment opens our eyes to the nature of variables. Understanding that there are forces able to tilt the balance in favor or against something is the first step toward reducing noise. It is not always possible to create noise-free practices. However, with proper decision hygiene, we can achieve a significant consistency in our judgments. It is in our hands to make our decisions more credible and accurate.
We rate this book 4.2/5.
Related Free Book Summaries
Thinking Fast and Slow by Daniel Kahneman
Think Again by Adam Grant