Book Talk: "Noise - A Flaw in Human Judgment" (Kahneman, Sibony, Sunstein, Brockman) | DLD All Stars
Vložit
- čas přidán 22. 02. 2021
- Speakers:
Daniel Kahneman, Professor of Psychology and Public Affairs, Princeton University
Olivier Sibony, Professor, HEC Paris
Cass Sunstein, Robert Walmsley University Professor, Harvard University)
moderated by John Brockman, Publisher & Editor, Edge.org
There’s a lot of noise around us - distractions of all sorts that can be found wherever people make judgments and decisions. Too often, individuals and organizations ignore this. They show “noise neglect”, as Nobel Prize Laureate Daniel Kahneman and his co-authors Olivier Sibony and Cass R. Sunstein argue. In their new book Noise - A Flaw in Human Judgment they show how noise can lead to errors in many fields, including medicine, law, public health, economic forecasting, food safety, management and human resources. In this DLD talk, moderated by John Brockman, the authors discuss simple remedies that we can all use to reduce both noise and bias in order to make far better decisions.
DLD All Stars is our virtual kick-off event to the DLD year. On February 21-23 we hosted inspirational speakers and role models from 15 years of DLD history. They gave us their call-to-action for 2021 in accordance with this year’s DLD motto, “What the World Needs Now
The DLD Conference channel features all talks held at past conferences and our digital format DLD Sync as well as the highlights of our events.
DLD Website: dld-conference.com
DLD on Instagram: / dldconference
DLD on Twitter: / dldconference
DLD on Facebook: / dldconference - Věda a technologie
I remember the first day in the physics lab where the demonstrator clearly explained the concepts of bias and error and standard methods of avoiding it. It's remarkable that simple concepts are ignored by everyone.
I was curious about how to measure noise, in my own decisions.
It is great to be able to live in this time, which allows me access to such brilliant minds.
Brilliant discussion on the topic of errors in judgement borne out of subjective variability and biases. What bothers me though is the use of the word "noise" to characterize unpredictable deviations in subjective judgement. In the area of statistical classifiers, each classifier is characterized through the notion of "bias" -- a constant deviation in the output in favor of one of the options, and "variance" -- the deviation that is random in nature. Why our esteemed speakers chose to talk of "noise" rather than "variance" in the (human) classifier output is beyond me. This is more bothersome since the notion of "noise" usually carries the connotation of "unwanted or misleading signal" -- e.g. as in Nate Silver's popular exposition in "The Signal and the Noise".
It is also interesting to note that the solution proposed to cover subjectivity in decisions is similar to what has become the staple in machine learning -- namely the use of ensemble classifiers to reduce the overall classification variance.
It will be very important in the financial audit!
What a group of awesome scientists 🖖🤗. Grandios
I wish this book to be translated in other languages of Asia to be profitable to the world . It's amazing , it's a work of genius what I just listened to. Thanks a lot.
This has huge implications for public policy, reforms, how they impact people and how people respond to policies.
The recording had lot of noise? 😊
Bias in scientific researches toward positive results is a big issue and need to remedy. I wish they discussed it by providing examples in this area too as when it mix with noise can resulting wrong conclusions.
7:04 - show the slide
A picture is worth a thousand words.
Looks like a much needed addition to the thought on decision making. I'm not too happy that Sibony keeps rushing into framing the book in terms of its business relevance (and consulting work) - in the process cutting off Kahneman and Sunstien, who are trying to develop this complex topic for us. You can see Kahneman visibly wrestling with the sunk cost of bringing onboard co-authors that are much too eager-beaver for his taste, I guess. Hope the book is biased towards Kahneman and Sunstein's ideas.
I am more interested at this point in my life (retired) in how to reduce noise on a individual level. It seems like they are missing something in the conversation.... doesn't what may be system noise become often bias noise when evaluated on the individual level. In other words a gun is limited at some point to its accuracy. No matter how well you shoot at the target your ability to hit the bulls eye is limited by the tool. There is randomness about the target. But lets say you are a parole officer and you want to audit your decision making and you notice in the morning you grant more probations in the morning (when you are well rested) and less in the afternoon (when you are tired). Now this is no longer noise but bias (I think). It is actionable. I can be something to try to right this situation. So what at a system level (all parole officers) was noise... becomes at the individual level... bias. ??? I guess I would love to had heard Edward Demming's take on this discussion. Something seems to be missing.
I’m wondering don’t these professionals have guidelines isnt that basically an algorithm. I imagined these professionals that make decisions base it on shared knowledge, the knowledge of their profession. Are these random professionals actually equal in intelligence,knowledge ,experience, temperament and expertise ? How do you test that? And are they suggesting that intuition or the culmination of knowledge and experience of a particular issue would not produce a similar decision with the same facts completely understood. Is the variance just a variance in understanding? Meaning 2 professionals disagree... one is less professional or experienced or neither of them is either.
WELL DONE
Poor Cass :( Silly Harvard give that man some fiber
I don't understand the concept of squaring absolute error which makes it difficult to understand why improving noise should be done even if you don't know what the true value is. Here is my example. If the true estimate is 50, and there is $1 of cost for being wrong by 1 point, and the guesses are normally distributed around 30, then how does eliminating the noise (everyone guesses) improve anything? It doesn't make anything worse, but it also doesn't improve anything (in this example). If the errors are squared then it does improve things because you eliminate the largest errors, but it doesn't make any sense if the cost of an error is $1 why would you square it?
Interesting mention about structures (meetings) can favor noise @~31:10, and then how noise is conveniently or actively avoided 34:50.
Perhaps noise is actually a feature, not a bug, in some cases? Favoring incompetence has benefits for some population in an organization. It's not hard to find errors in judgment long-term made in favor of short-term profits, either personal or organizational. Michael Lewis' latest book, "Premonition" is a pretty good epic of this (as is Cass' "This is Not Normal" to some extent.)
Perhaps you're looking through the wrong end of the telescope?
May I suggest that y'all's next book studies why noise is so prevalent, sometimes actually encouraged, in society.
I'm also curious about why the book is titled the anodyne "Noise" when you are talking about "errors" which seems to be much clearer.
Classic Cass
skip introduction
3:18
Before I read this book I thought, no way judges or doctor's or c-suite managers of large corporations could be wrong or make mistakes. Now I have learned about noise, and I realize that, my God, someone with a PhD can actually be an idiot.
Gentlemen you have changed the way I view this world forever. The next drink at Epstein's island is on me!
4:30 - The annoying moderator is looking all around while Kahneman speaks like he is annoyed or distracted. Hey - if you can't do the job then get up and get out, or pay attention and stop distracting people.
🚶
Useless ... these old codgers cannot even set up their internet right! ;-) What a waste of time for smart guys.