Insightful analysis of why we make bad judgments I have been very interested in the work of the psychologist and economist Daniel Kahneman since around where I came across some of the ideas around over-confidence bias on an Executive MBA at Insead, and this was only cemented with his Nobel Prize win with Amos Tversky in What were my first impressions of this book: My first — and negative reaction - was that it was a lot simpler than I was used to from this author and not in a good way.
This book by contrast seems to be light on ideas particularly early on — explaining what to me seemed sometimes very simple ideas in rather excruciating detail.
It felt like the first 80 pages in particular would have almost have been taken as a page of initial definitions in the work. The second — and by contrast positive reaction — was that the book was much more addressed to my own field.
In fact the very first example given in the book is actually from underwriting premium judgements and claims case assessment in an unnamed insurance company which given I run a global team of mathematicians whose key functions are assisting underwriters with the provision of tools to assist in setting premiums, and in carrying out calculations to complement case setting seemed rather relevant.
Whereas much of the earlier work was drawn on social science type examples and often on the aforemention artificial experiments, this book draws heavily both in its empirical data and its recommendations on areas of professional judgement.
Most of the repeated examples — the insurance example is one of a number of one-offs - are drawn from judicial work particularly sentencing , forensic science medical work, and HR areas both recruitment and performance assessment — the former much more mappable to my own work and the latter of course relevant to almost all workers.
So what is the book about? If you look at many judgments, and errors in those judgments all follow in the same direction, that is bias. By contrast, noise is the variability of error. A key assertion of the book is that noise has been largely overlooked — particularly in professional areas as professionals are not prepared to admit quite how noisy their views actually are.
They claim and aim to show from data that in terms of accuracy in post-fact verifiable judgements — noise is a much greater source of error than bias; and also make the point that with non-verifiable judgements, bias is anyway not a concept that can be easily investigated anyway.
Note on the latter though they perhaps miss the point that whereas individual judgements may not be verifiable, aggregate ones perhaps are insurance premiums — which they correctly say cannot be verified on an individual basis — being a case of an area that can be on an aggregate basis.
In terms of noise they later split noise into level noise taking the examples of judge sentencing — the difference in average sentences between lenient and draconian judges and pattern noise variability of judges responses to particular cases. They later split pattern noise into stable pattern noise this could be seen as for example an otherwise lenient judge who is systematically harsh on knife crime, or a harsh judge who is sympathetic to young offenders and occasion noise.
And this goes some way towards explaining why they define noise as so critical as I think many people would more naturally group both level noise and even pattern noise with bias. Some interesting areas: - Although later showing that stable pattern noise is perhaps one of the biggest contributors to error, an earlier chapter gives lots of example of how occasion noise is perhaps the most embarrassing part for professionals to admit: caused either hence its name by judgements being changed by extraneous circumstance weather, time of day, results of local sports teams have all been shown to influence judicial sentences or by internal inconsistency forensic scientists — including fingerprint experts — will commonly reach a different conclusion if given the same case months later, as well clinicians.
From my own professional viewpoint this can seem odd — underwriting professionals are more than happy to have models to complement and ground their assessments - There is an interesting discussion on model sophistication which effectively argues for one of two ends of a continuum. Either simple equal weights or frugal simply weighted models which aggregate a number of assessments known to be partly predictive OR a complex machine learning model when large data sets are available including factors not traditionally assessed.
Jul 23, Camelia Rose rated it really liked it Shelves: sociology , audio , psychology-neuroscience. His previous book, Thinking, Fast and Slow , was an eye-opener to me. Here is my understanding of the core concepts in Noise: A judgement is a decision made when a definite, invariant result can not be obtained at hand. Answering a school math question is not a judgement.
Weather forecasting is still a judgement but it is less likely so because of our improved understanding of weather science and the Noise: A Flaw in Human Judgment is the new book by Nobel Prize winner in Economics Daniel Kahneman. Weather forecasting is still a judgement but it is less likely so because of our improved understanding of weather science and the better measurement of weather data.
Two factors contribute to the errors in human judgement: bias and noise. The chance variability of judgments is noise. Bias and noise can be both present but they are different. Bias has been studied and policies have been make in an attempt to reduce biases in many areas, but noise gets little attention. Whenever something goes wrong, we look for a cause and often find it. In many cases, the cause will appear to be a bias. Bias has a kind of explanatory chrasma, which noise lacks.
If we try to explain in hindsight why a particular decision is wrong, we will easily find bias and never find noise. Only a statistical view of the world enables us to see noise but that statistical view does not come naturally. We prefer causal stories. The absence of statistical thinking from our intuitions is one reason that noise receives so much less attention than bias does. Another reason is that professionals seldom see the need to confront noise in their own judgments and in those of their colleagues.
Standard vs rules, the pros and cons of using computer algorithms in noise reduction. A well-defined algorithm will reduce noise, but it may also reveal the underneath biases. The authors believe this does not mean we should get rid of algorithms. Instead, we should improve them. Different strategies for different problems. Implementation is key.
There is no recipe for all. The complexity of all kinds of possible human judgement errors makes me wonder if it is possible to reduce noise at all, and at what costs. I find Noise is dry to read compare to Thinking, Fast and Slow. View 2 comments. May 17, Anonymous rated it did not like it. A boring, amateur, and often misleading take on concepts that decision scientists, machine learning engineers, and statisticians have known and systematically studied for decades with far more rigor than these authors do.
The authors are out of their depth here and contribute nothing new to the conversation. For example, their "error equation," which they call the "intellectual foundation" of their book, is a basic concept taught in high school statistics. Their folk, popular-press series of b A boring, amateur, and often misleading take on concepts that decision scientists, machine learning engineers, and statisticians have known and systematically studied for decades with far more rigor than these authors do.
Their folk, popular-press series of books have grown tired and at this point seem mostly like money-making machines for them in which they restate the obvious and botch the nuances and state of the art. Remind me again why we're listening to a psychology professor, a business professor, and a law professor's amateur thoughts on statistics?
View all 7 comments. Jul 27, Cassandra Kay Silva rated it it was ok Shelves: psychology-sociology. I loved Thinking Fast and Slow, so I picked this book up without thinking about it.
However, this was certainly not as well formulated, deep or interesting. Soemthing about the writing style felt disjointed. The thoughts were not cohesive or conclusive. For a book about Noise this felt rather noisy. It read more like a textbook or lecture than I wanted it to. View 1 comment. Oct 22, Athan Tolis rated it it was ok Shelves: psychology , business. I really have no idea who the intended audience was for this book: the authors really, really dumb it down, to the point of explaining what variance is over several pages of prose.
We did not all fail high school. The chapters end with summaries, which was OK for Thinking Fast and Slow, but a bit of an insult when the subject matter is so plain. The style is pompous and paternalistic.
Gun to my head, I could probably get it all down to one page. Let me try: 1. Noise is just as bad as bias in terms of messing up your results 2.
A good way to measure how bad your results are is the mean square error 3. He could be a harsh judge who is less harsh on young women who remind him of his daughter; or he could be a lenient judge who is extra lenient on young women who remind him of his daughter.
Same judge, same crime, same perpetrator, different outcome, because it was a different occasion 4. If you ask people to measure something independently from one another, the more the merrier; but if they talk to each other first, then they will amplify errors for a variety of reasons that lead to groupthink 5. Machines beat people when it comes to cutting noise 6.
In the quest to limit noise, people can fight back by sticking to simple rules 7. Bias can be the source of noise: inconsistency in bias is noise 9.
To improve judgements you need i better judges ii a decision process that aggregates in a way that maintains independence among the judges iii guidelines iv relative rather than absolute judgements There actually is a place for noise: when people are bound to game the system Read something else!
Jun 01, Nekomancer rated it did not like it. This is one of the worst popular press social sciences books I've ever read, and I've read many. It gets a lot wrong about what we know regarding decision-making and basic statistics. While it's true that algorithms are highly useful when applied appropriately, this book massively overstates the case in their favor while neglecting important counterpoints, among other serious problems.
Kahneman's "Thinking, Fast and Slow" remains one of my favorite books on research in psychology and this is an This is one of the worst popular press social sciences books I've ever read, and I've read many.
Kahneman's "Thinking, Fast and Slow" remains one of my favorite books on research in psychology and this is an extremely disappointing step down. I recommend skipping "Noise" entirely and looking elsewhere if you're interested in the subjects it touches on.
Want a book on statistics? Try "Naked Statistics" by Charles Wheelan. Interested in decision-making? Want critical thinking with a healthy dose of data interpretation? Just, whatever you do, skip "Noise" and spend your time elsewhere. May 30, Nick Lucarelli rated it liked it. Doesn't add enough to "Thinking, Fast and Slow" to warrant another book. Feels like one of those books where the author gets paid for every time they use a specific word in this case, "noise" and have said it to themselves so much it has become a cult-like world view.
In this instance, noise refers to the variations in human decision making which Kahneman attributes to a mixture of situational and systemic cognitive biases that covers old territory in the behavioural psychology world. He makes Doesn't add enough to "Thinking, Fast and Slow" to warrant another book.
He makes a case for a utopian rules-based slash AI system to guide decision making in spheres including law, medicine and HR, which can work to a degree to eliminate noise and bias but can also mute gestalt and out-of-the-box thinking.
Aside from the odd forcefully inserted and admittedly interesting behavioural psychology study The 5 page conclusion at the end is all that's worth your time here.
There is a lot of talk about bias and it is definitely important but another important issue is noise i. To introduce it, the book starts with a study by a judge Marvin Frankel, who showed on several almost identical criminal cases quite different rulings, e.
The first man was sentenced to fifteen years, the second to 30 days. His study was more case by case, but soon statistical studies started and they showed a great diversity.
In , federal judges were exposed to the same sixteen hypothetical cases so they judge the same case! In only 3 of the 16 cases was there a unanimous agreement to impose a prison term. Even where most judges agreed that a prison term was appropriate, there was a substantial variation in the lengths of prison terms recommended. In one fraud case in which the mean prison term was 8.
This works in private business as well: in insurance companies there are qualified underwriters or claims adjusters, who evaluate expert judgments. Then they shift to the question why there is so much noise and how to lower it. System noise can be broken down into level noise and pattern noise. Level noise is the variability of the average judgments made by different individuals. Regardless of the average level of their judgments, two judges may differ in their views of which crimes deserve the harsher sentences.
Their sentencing decisions will produce a different ranking of cases. This variability is pattern noise the technical term is statistical interaction. The main source of pattern noise is stable: it is the difference in the personal, idiosyncratic responses of judges to the same case. The main suggestion for reducing noise in judgment is decision hygiene.
Noise reduction, like health hygiene, is prevention against an unidentified enemy. Handwashing, for example, prevents unknown pathogens from entering our bodies. In the same way, decision hygiene will prevent errors without knowing what they are. Think statistically, and take the outside view of the case. Resist premature intuitions.
Obtain independent judgments from multiple judges, then consider aggregating those judgments. Favor relative judgments and relative scales. The Behavioral Decision Theory BDT authors of Thinking, Fast and Slow, and Nudge are the authors of this review of theory and research on errors, biases, and noise in human decision making. These are hugely important ideas that are often poorly understood by many readers, including many who should know better.
The book is well written and entertaining, with lots of examples and clear approaches for making use of these somewhat arcane ideas in our everyday decisions. Towards the end of the book, The Behavioral Decision Theory BDT authors of Thinking, Fast and Slow, and Nudge are the authors of this review of theory and research on errors, biases, and noise in human decision making.
Towards the end of the book, there are chapters that focus on particular professional areas and their difficulties in handling noise. The chapter on medicine is particularly good, including its discussion of noise issues in psychiatry.
The chapters on more common problems of noise encountered in work settings, such as in hiring and performance evaluation, are also excellent. Even if one follows the research literature, this is an extremely useful book and well worth the time to read. The book also has numerous diagrams and some useful exhibits, along with numerous references, for those wishing to learn more. May 21, Viktor Lototskyi rated it liked it. This book might be interesting if you're new to the topic, but overall, there's much less food for my brain than I would expect based on the previous "Thinking, Fast and Slow" Half of the book is describing multiple experiments that prove that people are biased and don't act rationally or make the right judgements all the time.
Like, happy and fed judges do less sentencing and so on. The rest talks that mood, weather and other factors creating noise and affect our judgements. And that's pretty mu This book might be interesting if you're new to the topic, but overall, there's much less food for my brain than I would expect based on the previous "Thinking, Fast and Slow" Half of the book is describing multiple experiments that prove that people are biased and don't act rationally or make the right judgements all the time.
And that's pretty much it. Even the practical part is too generic to add anything. There's an obvious difference between bias and noise, but the latter could nicely fit in the other format than a book. It took some effort to finish it up. Interesting look at noise- anything and everything from time of day, to weather, to unconscious preconceptions- that causes inconsistencies in judgement.
The authors go through several studies and cases including the judiciary branch, actuary science, and medicine and take a look at examples of noise in the decision-making processes. They call for a hygiene makeover for the way that judgments and decisions are handed down.
They maintain that too much noise has permeated our society and it is a m Interesting look at noise- anything and everything from time of day, to weather, to unconscious preconceptions- that causes inconsistencies in judgement. They maintain that too much noise has permeated our society and it is a major contributor to societal injustices.
An interesting, quick, nonfiction read. Jun 05, Patrick rated it it was ok. This book was a long slog. The topic of noise variability error —not to be confused with bias error , is important and has serious consequences on human judgements. Unfortunately, the novel insights in this book are buried within many pages of uninteresting, poorly edited text. The bottom line is that people make noisy decisions most of the time.
Most of us tend to believe we make rational decisions. In some cases such as the judicial and medical systems noisy decisions can have undesired and even tragic outcomes. The reasons for this noise are categorized in detail in the book but they all come down to the fact that people are emotional and easily influenced, sometimes by non-obvious tangential things like the weather and if their home team won the game the night before.
Our decisions are much, much more emotional and noisy than we realize so we should first be aware of this, and second take steps to reduce noise for judgements and decisions that impact our lives and the lives of others.
The book gives some examples of how to do this but many seemed impractical, involved, and not easy to implement e. Many of the solutions presented come with their own downsides such as reducing noise errors but increasing bias errors. The last chapter of the book provides an excellent summary for those wanting the cliff notes.
I think both of these books on influence could help reduce noisy decision making. Jun 29, James rated it really liked it. While a lot of great podcasting employs journalism ethics, best practices, and principles, not all journalism makes for good podcasting.
There are a few reasons for this. In journalism, there are two basic kinds of stories: stories that tell the news and stories that forward our understanding of events and the people involved in them.
Headline news is a ubiquitous commodity. When an aging celebrity keels over from a heart attack, there are literally hundreds of sources who tell you, often as breathlessly quickly as possible, that the death has occurred.
Let me provide an example. Recently a friend of an acquaintance called me for advice on starting a podcast. When I asked what the podcast was about, she told me they had done some investigative work on a local doctor who had been accused of molesting young female patients—very young female patients. But none of that was a reason to listen to the story. These are examples of noise: variability in judgments that should be identical. Sunstein show how noise helps produce errors in many fields, including medicine, law, public health, economic forecasting, food safety, forensic science, bail verdicts, child protection, strategy, performance reviews and personnel selection.
And although noise can be found wherever people make judgments and decisions, individuals and organizations alike commonly ignore to its role in their judgments and in their actions.
Packed with new ideas, and drawing on the same kind of diligent, insightful research that made Thinking, Fast and Slow and Nudge groundbreaking New York Times and international bestsellers, Noise explains how and why humans are so susceptible to noise in judgment — and what we can do about it.
Bazerman, author of Better, Not Perfect. He lives in New York City. CASS R. Horbachevsky hbgusa. Little hbgusa. For UK media inquiries, please contact : Katherine. Patrick HarperCollins. Wherever there is human judgment, there is Noise. Ad Libris. Mighty Ape.
0コメント