I've been reading Skin in the Game. My experience has been similar to yours. There are valuable concepts in the book and he explains them well. All the talk of how smart he is, how dumb almost everyone else is, and the winding anecdotes to nowhere I could do without (or much less of). I think some of these sections are meant to be entertaining satire and not much more?
Thanks for your review!
This is my take: I entered college in 2007, and took a few public policy courses with a professor who was excellent. She spotlighted this book, which I'll admit didn't make a huge impression on me at the time. But it was the first introduction I had to these ideas, and I think they stayed with me. When I reread it a few years ago, I really enjoyed it and thought it stated perfectly a lot of things I'd already picked up on or heard in more obscure ways in the intervening years. I assume that for many, particularly people who don't have any background in this sort of thing, this stuff is new to them or has never been stated in a way that resonates.
I've always disliked discussing statistics and finance, even though I enjoy learning about almost everything. The sense I got was that to understand and use it at all, you'd have to constantly master it and all its tricks--that there was no real in-between. The rules were always changing, and the underlying conditions.
The way Taleb discusses these topics addresses this exact issue, and is very easy for me to follow. The personal tone of the book establishes a feeling of trust...that, I think, is what he signals with those asides. He acknowledges the game being played, even as he plays it. This appeal to a certain type of reader and explains his fans' enthusiasm. It definitely isn't for everyone. But it is definitely my experience that someone like me would not have been familiar with these ideas at the time the book was published. They are much more common now. But Taleb's combative, eccentric style and unique perspective still stand out in general, and remain a big part of his appeal.
I've always disliked discussing statistics and finance, even though I enjoy learning about almost everything. The sense I got was that to understand and use it at all, you'd have to constantly master it and all its tricks--that there was no real in-between. The rules were always changing, and the underlying conditions.
One thing going on there for statistics is that the field greatly dislikes presenting it in any of the unifications which are available, which is something I learned only quite late myself. As often taught or discussed, statistics is treated as a bag of tricks and p-values and problem-specific algorithms. But there are paradigms one could teach.
For example, around the 1940s, led by Abraham Wald, there was a huge paradigm shift towards the decision-theoretic interpretation of statistics, where all these Fisherian gizmos can be understood, justified, and criticized as being about minimizing loss given specific loss functions; the mean is a good way to estimate your parameter (rather than the mode or median or a bazillion other univariate statistics one could invent) not because that particular function was handed down at Sinai but because it does a good job of minimizing your loss under such-and-such conditions like having a squared error loss (because bigger errors hurt you much more), and if those conditions do not hold, that is why the, say, median is better, and you can say precisely how much better and when you'd go back to the mean (as opposed to rules of thumbs about standard deviations or arbitrary p-value thresholds testing normality). Many issues in meta-science are much more transparent if you simply ask how they would affect decision-making.
Similarly, Bayesianism means you can just 'turn the crank' on many problems: define a model, your priors, and turn the MCMC crank, without all the fancy problem-specific derivations and special-cases. Instead of all these mysterious distributions and formulas and tests and likelihoods dropping out of the sky, you understand that you are just setting up equations (or even just writing a program) which reflect how you think something works in a sufficiently formalized way that you can run data through it and see how the prior updates into the posterior. The distributions & likelihoods then do not drop out of the sky but are pragmatic choices: what particular bits of mathematics are implemented in your MCMC library, and which match up well with how you think the problem works, without being too confusing or hard to work with or computationally-inefficient?
And causal modeling is another good example: there is an endless zoo of biases and problems in fields like epidemiology which look like a mess of special cases you just have to memorize, but they all reduce to pretty straightforward and obvious issues if you draw out a DAG of a causal graph of how things might work.
Much of the 'experience' that statisticians or analysts rely on when they apply the bag of tricks is actually a hidden theory learned from experience & osmosis, used to reach the correct results while ostensibly using the bag of tricks: the analyst knows he ought to use a median here because he has a vaguely defined loss in mind for the downstream experiment, and he knows the data sometimes throws outliers which screwed up experiments in the past so the mean is a bad choice and he ought to use 'robust statistics'; or he knows from experience that most of the variables are irrelevant so it'd be good to get shrinkage by sleight of hand by picking a lasso regression instead of a regular regression and if anyone asks, talk vaguely about 'regularization'; or he has a particular causal model of how enrollment in a group is a collider so he knows to ask about "Simpson's paradox". Thus, in the hands of an expert, the bag of tricks works out, even as the neophyte is mystified and wonders how the expert knew to pull this or that trick out of, seemingly, their nether regions.
Teachers don't like this because they don't want to defend the philosophies of things like Bayesianism, often aren't trained in them in the first place, and because teaching them is simultaneously too easy (the concepts are universal, straightforward, and can be one-liners) and too hard (reducing them to practice and actually computing anything - it's easy to write down Bayes's formula, not so easy to actually compute a real posterior, much less maximize over a decision tree).
There's a lot of criticisms that can be made of each paradigm, of course, none of them are universally assented to, to say the last - but I think it would generally be better to teach people in those principled approaches, and then later critique them, than to teach people without any principles at all.
Thanks Gwern! I was wondering if you had any pointers as to where beginners should start in terms of understanding statistics paradigmatically? I've not come across statistics explained this way before, and I am quite interested to learn more.
Thanks for the explanation---that all makes sense. I guess what I was getting at is that as you said, it can be done in a completely sensible way by people who know what they're doing, but it tends to become split up in awkward ways.
I'm not sure what "I already know about hindsight bias" is about. Hearing about hindsight bias alone has no use for not getting affected by it. You can express the idea of hinsight bias in a paragraph but that still doesn't help. What's important is learning about the idea in a way that actually affects actions. That likely needs hard reflection with the actual content and having multiple examples can be helpful for that.
Survivership and hindsight bias isn't Talebs central message. To me his central message is about many distributions being fat-tailed. That insight is also resulted in Taleb's current higher popularity because it lead to him seeing COVID-19 as a serious threat in January.
Christian, thanks for your comment. However, I do not understand what your first paragraph is referring to, as I do not think I claimed not being affected by hindsight bias, or anything similar. Whether additional examples of the hindsight bias (or anything else) are helpful is up to any potential reader to decide; I'd just say that I find the signal-to-noise ratio of the book low, and would probably start reading https://en.wikipedia.org/wiki/Hindsight_bias (or Kahneman) instead.
The central message of Taleb's oeuvre in general may be about many distributions being fat-tailed, but just judging from FbR, I think you will not learn very much about the concept. Searching the google books version for the word "fat" seems to indicate that the word "fat-tailed" only appears in the preface.
My point is that signal-to-noise ratio is a metric that rewards mentioning many ideas over exploring the ideas in more detail. For important ideas like specific biases that can be started shortly, I think it's valuable to explore them in a longer way and signal-to-noise is no good complaint.
I see. Using "signal" and "noise" figuratively here, I ran the risk of being understood that way. But to be clear: I do not regard explanations and illustrations as "noise", because they help understand the signal. The book has a lot of text that is counterproductive and has, in my opinion, a very loose relation with the concepts Taleb (presumably) aims to explain.
The book "Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets" by Nassim Nicholas Taleb was published in 2001. On Amazon, the book has a rating of 4.3, rated by 1,124 people, with 62% of reviewers giving 5 stars. On Goodreads, it has a rating of 4.07 from 1,968 reviews. "The book was selected by Fortune as one of the 75 'Smartest Books of All Time.'", notes Wikipedia. So this should be a great book. I forgot who wrote this, and I forgot the exact words, but I remember reading on twitter that Taleb is something like a modern genius because in his books he develops a worldview based on fundamental insights on randomness, and derives important conclusions including a system of ethics from that. Now that sounds promising! This worldview is developed in Taleb's Incerto series of five books. In the description of the German publisher you can read that the order in which you read the books does not matter, but then it seemed like a good idea to start with the first one nonetheless ("The Black Swan" surely is the better-known title and has 2,083 reviews).
The book seems to have been a revelation to many, judging by the enthusiastic reviews. In a five-star Amazon review, reviewer Alex Bush writes (November 3, 2015) that the book "revolutionized how I view the world. In multiple ways. It's hard to overstate how rarely a book changes your ideas about how the world works once, let alone multiple times". He thinks that Taleb "has managed to weave a fantastically engaging and entertaining book out of what could very easily be a dry and technical topic", that FbR is "the most general and therefore most widely applicable" book by Taleb, and suggests that it teaches people "the survivorship and hindsight biases, as well as the difference between conditional and unconditional probability". Indeed, these are things you may learn from the book in a non-technical way. You may also learn that people find causality where there is mere randomness; that the past of a time series can lead you to a false feeling of certainty if rare events are relevant; that it is often disregarded that the expected value of a random distribution may differ from the median, when in fact it would be critical to keep it in mind if possible rare events would have strongly negative effects. These things come together when people attribute a financial trader's performance to his ability in selecting the right stocks, their value increases over a long time, but he disregards that a crash of the stockmarket may ruin him. So the fact that I did not find the book so eye-opening may be related to the fact that I already knew some things about statistics, probability, econometrics and behavioral economics, that I had read Thinking Fast and Slow, blogposts on Lesswrong, and some other things Taleb writes about. To be fair, the book is from 2001, and maybe many concepts described in the book were very innovative back then.
However, amazon reviewer Alex Bush also writes that "it's often hard to determine whether the heart of the book is the ideas ... or the author. I can't stress how much I learned from this book that has nothing to do with probability or statistics, just random asides from an erudite and meandering mind." This may make you a bit skeptical.
And indeed, the author does meander. A common criticism in reviews is that the book is way too long; suggestions are somewhere between a paragraph (which is a bit mean), a New Yorker article and 50 pages. More importantly, the book lacks structure. In the preface, Taleb writes that he "hates books that can be easily guessed from the table of contents (not many people read textbooks for pleasure)". And so it is a bit obscure what kind of book Taleb wants this to be. It seems he aims at being illuminating and, at the same time, entertaining. The book contains many stories and anecdotes, usually using a story about some person exemplifying a certain kind of behavior, but it is sometimes unclear how representative these stories are.
If nothing else, you will certainly know a lot about Nassim Nicholas Taleb when you have finished reading the book. He likes to go to the gym. He does not want to be a janitor. He is intelligent and cultured, and people around him are ignorant and shallow - people in the financial industry and in the media in particular - and they usually do not appreciate his insights when he explains them. As a cultured, intelligent person he also admits some emotional weaknesses. You will also know the names of some people whom Taleb likes and several people whom he dislikes. Maybe you read somewhere that Taleb is arrogant and likes to start fights, but that he has profound insights. (I had read something like that somewhere.) Well, I don't mind arrogance per se, if it is just some topping on the cake of insight. What I find a bit annoying, however, is that some people (or at least some reviewers) seem to understand the style of the book as a signal for the insight.
And indeed, Taleb's self-descriptions seem to aim at signaling some kind of insight. This can be a bit lengthy. When Taleb writes that he had a lucky career choice, he notes that "one of the attractive aspects of my profession as a quantitative option trader is that I have close to 95% of my day free to think, read and research". As this does not seem to be clear enough, he also adds: "(or 'reflect' in the gym, on ski slopes, or, more effectively, on a park bench)", and then: "I also had the privilege of frequently 'working' from my well-equipped attic." Over the book, the self-descriptions of Taleb, his descriptions of people around him, and many other digressions add up. In a book whose topic is the problem that we often confuse noise with signal and should be mindful of distinguishing them, this is a bit disturbing. Of course - reading about an author's life can be interesting, and anecdotes can be illuminating; Daniel Kahneman's writing is a good example of that. But the entertainment highly depends on an author's style, and writing anecdotes in an illuminating way seemingly is not easy.
(Just to add one example: At some point, Taleb says that he sat in many meetings where traders had to offer their interpretations of what was currently moving the markets. He says he found these meetings to be a waste of time. And he adds that he himself talked a lot in them, to make them less boring, though he did not listen to what other people had to say: "I have to confess that my optimal strategy (to soothe my boredom and allergy to confident platitudes) was to speak as much as I could, while totally avoiding listening to other people's replies by trying to solve equations in my head. Speaking too much would help me clarify my mind, and, with a little bit of luck, I would not be 'invited' back (i.e, forced to attend) the following week." It is up to every reader to decide what the confident platitudes are at this point; but it is a little weird that it does not seem to occur to Taleb that the other participants of these meetings may have the same justification for their "platitudes" and the same view of themselves and the others.)
I stopped reading the book after chapter 9 (that is, after 202 paes of the German epub edition, which is 302 pages long, excluding some back matter parts). At this point, the book had improved, the last two chapters had been more readable and more interesting, which might have made me continue reading. But then again, what had these last chapters been about? Chapter 9 nicely illustrates the problem of regression to the mean, and this served as a reminder to lower expectations. The expected value of the rest of the book did not seem high enough.
Summing up
You may like the book if you need a story-based introduction to some biases in understanding random events, and if you find a certain joy in an author classifying others as idiots.
(At several points, I think things that Taleb writes are wrong or misleading, but I will not put time and effort into arguing about these points.)
What did I like about the book?
There are nice illustrations about regression to the mean and survivor bias. Also, the mentions of several financial-market crises of the last decades served as a nice reminder to put things into perspective.
What did I learn from the book?
The most interesting thing that the book emphasized to me was that it is often hard to rate the quality of books using a single "quality" dimension. For a fair judgement, it seems necessary to know the background of the typical reader, to know what the author wants, and to understand the intellectual environment in which he wrote the book. Fooled by Randomness may be a good book for someone to whom the survivorship bias is news, it may transport Taleb's message, and it may have been revolutionary 19 years ago.
What should you read if you like the topic?
Daniel Kahneman's "Thinking, fast and slow"