If it's your thing, I highly recommend NNT's YouTube channel [1]. He uploads short videos (that he apparently films with a potato), where he explains stat concepts "for beginners." Of course, he also spends lots of time _roasting_ random researchers/statistics users/noobz.
Actually though, I've actually a bunch of his videos super educational and well made - if you can get past the terrible camera / sound quality.
I watched his explanation of the central limit theorem [2] the other day and found it super helpful. I'm not great at math, but I took multiple stat classes at uni (including two where we went through proofs for the CLT at some level). But this single 10 minute video helped me "get it" more than any class. The big observation that I was missing is that the sample mean is pretty much the same as the sample sum, and the sample sum can sum to "things in the middle" in more ways that things not in the middle (e.g. if you have two dice, rolling a 7 is more likely than rolling a 12).
I'm sure I'm butchering it mostly, but that little bit of intuition was just *nice* - especially after 2 stat classes of rote proof following w/ little to no intuition building.
I know of scarcely anything so apt to impress the imagination as the wonderful form of cosmic order expressed by the law of frequency of error. The law would have been personified by the Greeks if they had known of it. It reigns with serenity and complete self-effacement amidst the wildest confusion. The larger the mob, the greater the apparent anarchy, the more perfect is its sway. It is the supreme law of unreason.
- Sir Francis Galton, on the Central Limit Theorem
I have very mixed feelings about these videos. Some don't quite build enough intuition, others are extremely handwavy.
Also, Taleb often disses certain methods without providing alternatives or entertaining the idea that alternatives already exist.
My recent favorite is his rant about correlation where he shows that linear regression can yield a slope of zero on certain piecewise examples.
He just makes fun of regression and moves on without acknowledging piecewise and kernel regression exist to address those cases. Instead he'll make some comment about information theory being the solution without any further elaboration.
Sure correlation can be abused, but so can mutual information and any other information theoretic metric you choose.
If you know a programming language, stuff like Wolfram (or doing it in Python/R) can really help.
Stats suffers quite heavily from having to introduce some ideas, most importantly distributions, which can be tricky to intuit at first. Running through some examples with Wolfram (there is a free version which you can in Jupyter) or using the stats package from scipy in Python was very helpful to me personally.
I agree. I once spoke with a statistics professor in charge of the undergraduate program at my university. I asked him tons of questions about the cursus, and swiftly replied "Yes the program is far from perfect. However understand that statistics is still a new field compared to calculus. I do not think that we know how to teach all of this correctly...". I wish Taleb could write a book about it!
Thanks for the links! Wish he would write a full-length book of "stats for beginners" so I could better understand his critiques of its misuse, but this looks like a good place to start anyway.
Wow! I didn’t know about this, thanks for the recommendation.
I love this guy, it’s a guilty pleasure for sure because of his personality but I find it somehow refreshing.
> Antifragile: what doesn't kill you makes you stronger
No, you've missed the point of the book (or haven't read it). It's about that some things become better in the presence of error/noise/damage/volatility. But not all things, some other, fragile things, become worse and worse as damage accumulates. And that there's no word for such a property, so he calls it "Antifragile". And then the book goes in great detail about lots and lots of examples (with the usual rambling).
> become worse and worse as damage accumulates. And that there's no word for such a property
I think the name for that property is "age". :)
Do you have an example of something that gets better in the presence of accumulated damage? Maybe giving something the "weathered" look on purpose? I can really only think of subjective examples.
His books are definitely repetitive (and somewhat unorganized), but I find that saying the same thing a hundred times in slightly different ways helps drive home the message.
People may "nod their head yes" when you give the one sentence descriptions, and think they understand it. But then they ACT in a different way, especially when they have "skin in the game".
Taleb actually has a name for that: "domain dependence". That is, you can nod your head in agreement while reading a book, but then when confronted with the exact same situation in real life, you act as if you don't have that knowledge, or you act contrary to it.
I've noticed this in programming vs. computer science. In a test or interview situation, someone might say they'll do things one way. But then their production code they do it a different way, just because that's the common way they've seen it done, in that particular situation.
This can be good or bad -- sometimes the textbook way is actually better; sometimes not. You can also flip it around -- it also applies to the person asking the question. They might expect a certain answer of the interviewee, but when they have skin in the game, they do something else.
----
I also don't agree with the one sentence description of the Black Swan. I'd say the most important idea is that outliers are often ignored, but they're precisely what drives history. [1] And also that they're more likely than naive mathematics would suggest, i.e. "fat tails".
To use an example that's HN friendly, every successful company is an outlier. And we always seem to be talking about the biggest outliers of them all (e.g. FAANG, none of which existed when I started using computers.) Making statements about the average new company isn't informative, because the average one fails and doesn't matter.
[1] Thiel has a similar sentiment, something like "every big moment in business happens exactly once. The next Google isn't making a search engine, etc."
I don't think this is a good characterization. Black Swan is more about "Looking just at probability is misleading and dangerous, you need to look at severity too". I haven't read antifragile, but I don't think that statement summarizes it (also it's a completely false statement). My understanding was that it's message is <<resilient systems are those that have "fragile" components but can adapt to/ learn from failure; not those that are designed to be "very strong">>.
While I have a similar impression (that he is overly verbose), my take from Black Swan was “even though you can put any data into the formula for calculating mean and standard deviation of a bell curve, lots of stuff isn’t actually on a bell curve, and if you assume it is when it isn’t you will have a lot more bad days than you planned for”.
Yeah the way I remember it is to compare "casino math" vs. math for real life problems like the stock market. Before reading the book I didn't appreciate how much these are two entirely separate domains, and how much it matters.
And even people who "know" that they're two separate domains fail to ACT like they know it.
If you're trying to predict what happens with 52 cards or a six-sided die, well that's easy. Everything works out nice and cleanly, with no ambiguity.
But if you're trying to predict markets, the math is intractable, for fundamental reasons. An example is that Michael Lewis talks about the "Value At Risk" that Wall St. metric in one of his books, leading to the 2008 crash. That's clearly casino math applied to real life, failing catastrophically.
The charitable interpretation is that people saying "well this only happens once in 100 million years" are genuinely naive. The other interpretation is that they know that their gains will be privatized and their losses will be socialized, so they have no incentive to use math that's not nonsense.
This is basically the "ludic fallacy" [1], although I think "casino math fallacy" is easier to remember. You're applying the math of games to real life, which is wrong exactly where it matters (in the fat tail).
I took one of his twitter polls where he asked a mathematical question. And the little Nasim shouting 'imbecile' worked in perfect symbiosis with the little Kahneman to engage System 2 and think about my answer.
I thought the same for a long time about Antifragile and the oft overlooked Skin in the Game. Both books are now on my yearly reading list. Reading those two along with Fooled by Randomness and to an extent the Black Swan has made me infinitely less foolish and perhaps wiser.
The ideas in Antifragile are to me an good application of the bets to make in life and love. Skin in the game is to evaluate the bets (actions) of others. Doctor prescribing statins...hmmm...what are his incentives around the upsides/downsides of it?
Fooled by Randomness is a prescription of how to evaluate systemic performance (or failure). Black swans is eh...just about black swans or how outlier events in power law distributions can fool us by not showing up for a long time and then ...watch out.
Nassim can appear to be overly verbose but his books defy summarization. I recommend reading them but given my own experience can empathize by the contra opinions.
I'd love to see more antifragile ideas but on more precise terms. I also think it's an idea that's starting to blossom in some fields. Hormesis (controversial) is the same idea at the cellular level. I think I also saw the idea in material science (maybe structural batteries, I forgot).
This is an odd summary of antifragile. I would go with an angle that justifies the word. "When building something you don't want to be fragile, it is not sufficient or even wise to go with established antonyms of fragile."
This was the thinking of some people who didn't want to inconvenience themselves in the face of a respiratory pandemic. Some of them died, but more to the point many of them are still suffering long-term effects even though they've "recovered" from their illness.
Fooled by Randomness by Nicholas Taleb is probably the most influential book I have read so far. It completely changed my world view. But at the same time it made life less meaningful. Are there any books to minimize the damage made by the book? Books that give propose and meaning back to life?
It's the opposite of Taleb. Basically, if people were rational and used financial instruments properly, a lot of the risks of our lives can be mitigated. The trick isn't so much to control everything but to trade the less desirable forms of risk for the ones you can handle.
> Books that give propose and meaning back to life?
Ride The Tiger by Julius Evola.
Unlike most of these suggestions this book will actually give you purpose and meaning in life, but you will be vilified for even mentioning it. That's how you know it's good.
I have the same question. I didn't get the feeling that life was less meaningful out of it.
It has been a long time since I read it, but my basic understanding was that a lot of success is from luck and showed how that works. I remember clearly the math around investment managers being mostly lucky to have returns above the average yet they get a lot of praise and then their luck runs out. But there are still some that stay lucky for a long time.
If I was a successful investment manager I might question my success.
I couldn't make it past the beginning because it felt uppity, self congratulatory, and pointless compared to something like Thinking Fast and Slow. How did this book change your worldview?
I have not read it. The black swan was interesting. But in the end I may have liked books from Mandelbrot more. I have met professional traders that did not know (and became very angry) when told that far out of the money options are underpriced because it is not a random walk but a fat tail distribution.
I have not read Thinking Fast and Slow ether but really disliked "The tipping point" by Malcolm Gladwell.
Haha! I had a similar reaction when I started reading Taleb. You might checkout Antifragile, which offers a more positive view of randomness. It's easy to get a dire impression from Taleb, but to use his words, "don't read it *too* well." If you really dig into Taleb, specifically Skin in the Game, he is a proponent of risk-taking: his closing statements in Skin in the Game include "Start a business."
It may be easy to jump to the fatalistic conclusion that we are all clueless lemmings waiting to be wiped out by some random event or another but the book ultimately suggests to not ignore tail events. What you make out of those is up to you. Some of those are disasters for which one should hedge against but some others are opportunities to reap outsize benefits.
Have you tried Meditations by Marcus Aurelius? TL;DR - do something for other people, don't whine about your current position, don't whine about stupid people, believe in god(s). If you don't know what to do for other people, start at any local homeless shelter or food service. They always need someone to help. If you don't like my list, read that book yourself.
I used to like stoic writings until I realized that they rarely applied those principles to own life. Meditations is basically a description of painful experiences even when one is a rule of Ancient Rome empire.
I would suggest to try to read writing of Epicurus instead. At least he did followed own principles as far as we know.
I recommend 'The Hero with a Thousand Faces' by Joseph Campbell as it also deals with pre-history but focuses on mythology and the universality of a story of the hero across cultures.
For those who are confused by Taleb, I recommend reading some Eastern philosophy. It offers insights for coping with circular and contradictory life. Tao Te Ching, Yi Ching, Buddhist texts.
I got the opposite feeling from The Black Swan.
Taleb finished the book by pointing out that life on earth is a black swan and very precious. That was meaningful to me.
I was looking for this answer and found it at the bottom downvoted. I wonder how many people actually know what Peterson says as opposed to what the NYT says about him.
To sum it up: do what is at the limit of your competence and what you feel very drawn too (as in very interested in it). That’s where you’ll find a balance of your existing knowledge and pressure to expand your horizons, plus you are following your passion. That gives meaning.
It’s certainly worked for me recently. It’s very good advice.
Yes; certainly a politically controversial figure, but "Maps of Meaning" is, obviously, about meaning. Could also try an influence of his, Jung, but he can be a challenge to understand. I'm a bit disappointed that Taleb seems to be unwilling to even talk with figures such as Peterson or Pinker so I could better understand all of their positions, but oh well.
Suspect the down-votes aren't related to the factualness of the statement, but the perceived value of Peterson as a recommendation for books worth reading in this context (i.e. what GP asked for).
Yes, one of the central points of his books ("Maps of Meaning", "12 Rules for Life", and "Beyond Order") is that meaning in life is found by taking on personal responsibility. A good starting point is probably his "12 Rules for Life".
I came here to say the same! I’ve been thinking about this recently. It began after hearing the Weinstein brothers talk about Taleb[1]. My hunch is that a significant part of Taleb’s popularity stems from how dogmatically polarizing his views are. I’m not sure if he’s doing so authentically (I think he is), or: if he’s dogmatic in his opinions primarily because he knows doing so will result in wider distribution of his views.
I struggle with this dilemma.
On one hand, there’s a case for being strong-willed, dogmatic and polarizing. It takes courage.
On the other hand, there’s a case for being open-minded, flexible in your thinking, and uncertain. Yet, this is generally perceived as weak-willed and doesn’t attract nearly the same attention as the polarizing stance.
Taleb's insights are wonderful and I've learned a lot from his writing, but his writing style is so obnoxious that I always have trouble finishing his books. It is very clear that he thinks he is one of the smartest people on the planet and he wants everyone to know it.
You have to understand that he probably mostly reads books he enjoy and is statistically and factually correct. The 1 star reviews is when finds lots of statistical fallacies and misuse or they're factually incorrect. He doesn't take lightly on these kinds of things. Personally I think it's great when people finds mistakes in book and publish them for others.
He’s far from the only person to spot that idea, though. My favourite take is from a different author:
"""An Outside Context Problem was the sort of thing most civilisations encountered just once, and which they tended to encounter rather in the same way a sentence encountered a full stop."""
Readit News
Actually though, I've actually a bunch of his videos super educational and well made - if you can get past the terrible camera / sound quality.
I watched his explanation of the central limit theorem [2] the other day and found it super helpful. I'm not great at math, but I took multiple stat classes at uni (including two where we went through proofs for the CLT at some level). But this single 10 minute video helped me "get it" more than any class. The big observation that I was missing is that the sample mean is pretty much the same as the sample sum, and the sample sum can sum to "things in the middle" in more ways that things not in the middle (e.g. if you have two dice, rolling a 7 is more likely than rolling a 12).
I'm sure I'm butchering it mostly, but that little bit of intuition was just *nice* - especially after 2 stat classes of rote proof following w/ little to no intuition building.
[1] https://www.youtube.com/channel/UC8uY6yLP9BS4BUc9BSc0Jww
[2] https://www.youtube.com/watch?v=bfM9efdStN8
- Sir Francis Galton, on the Central Limit Theorem
Two interactive demos...
http://mfviz.com/central-limit/
http://www.ltcconline.net/greenl/java/Statistics/clt/cltsimu...
Also, Taleb often disses certain methods without providing alternatives or entertaining the idea that alternatives already exist.
My recent favorite is his rant about correlation where he shows that linear regression can yield a slope of zero on certain piecewise examples.
He just makes fun of regression and moves on without acknowledging piecewise and kernel regression exist to address those cases. Instead he'll make some comment about information theory being the solution without any further elaboration.
Sure correlation can be abused, but so can mutual information and any other information theoretic metric you choose.
Stats suffers quite heavily from having to introduce some ideas, most importantly distributions, which can be tricky to intuit at first. Running through some examples with Wolfram (there is a free version which you can in Jupyter) or using the stats package from scipy in Python was very helpful to me personally.
Antifragile: what doesn't kill you makes you stronger.
Black Swan: Forecasting is really hard, perfect forecasting is impossible.
No, you've missed the point of the book (or haven't read it). It's about that some things become better in the presence of error/noise/damage/volatility. But not all things, some other, fragile things, become worse and worse as damage accumulates. And that there's no word for such a property, so he calls it "Antifragile". And then the book goes in great detail about lots and lots of examples (with the usual rambling).
I think the name for that property is "age". :)
Do you have an example of something that gets better in the presence of accumulated damage? Maybe giving something the "weathered" look on purpose? I can really only think of subjective examples.
Things that get worse with accumulated damage definitely abound in physics. One example that came immediately to mind. https://en.wikipedia.org/wiki/Hot-carrier_injection
People may "nod their head yes" when you give the one sentence descriptions, and think they understand it. But then they ACT in a different way, especially when they have "skin in the game".
Taleb actually has a name for that: "domain dependence". That is, you can nod your head in agreement while reading a book, but then when confronted with the exact same situation in real life, you act as if you don't have that knowledge, or you act contrary to it.
I've noticed this in programming vs. computer science. In a test or interview situation, someone might say they'll do things one way. But then their production code they do it a different way, just because that's the common way they've seen it done, in that particular situation.
This can be good or bad -- sometimes the textbook way is actually better; sometimes not. You can also flip it around -- it also applies to the person asking the question. They might expect a certain answer of the interviewee, but when they have skin in the game, they do something else.
----
I also don't agree with the one sentence description of the Black Swan. I'd say the most important idea is that outliers are often ignored, but they're precisely what drives history. [1] And also that they're more likely than naive mathematics would suggest, i.e. "fat tails".
To use an example that's HN friendly, every successful company is an outlier. And we always seem to be talking about the biggest outliers of them all (e.g. FAANG, none of which existed when I started using computers.) Making statements about the average new company isn't informative, because the average one fails and doesn't matter.
[1] Thiel has a similar sentiment, something like "every big moment in business happens exactly once. The next Google isn't making a search engine, etc."
And even people who "know" that they're two separate domains fail to ACT like they know it.
If you're trying to predict what happens with 52 cards or a six-sided die, well that's easy. Everything works out nice and cleanly, with no ambiguity.
But if you're trying to predict markets, the math is intractable, for fundamental reasons. An example is that Michael Lewis talks about the "Value At Risk" that Wall St. metric in one of his books, leading to the 2008 crash. That's clearly casino math applied to real life, failing catastrophically.
The charitable interpretation is that people saying "well this only happens once in 100 million years" are genuinely naive. The other interpretation is that they know that their gains will be privatized and their losses will be socialized, so they have no incentive to use math that's not nonsense.
This is basically the "ludic fallacy" [1], although I think "casino math fallacy" is easier to remember. You're applying the math of games to real life, which is wrong exactly where it matters (in the fat tail).
[1] https://en.wikipedia.org/wiki/Ludic_fallacy
The ideas in Antifragile are to me an good application of the bets to make in life and love. Skin in the game is to evaluate the bets (actions) of others. Doctor prescribing statins...hmmm...what are his incentives around the upsides/downsides of it?
Fooled by Randomness is a prescription of how to evaluate systemic performance (or failure). Black swans is eh...just about black swans or how outlier events in power law distributions can fool us by not showing up for a long time and then ...watch out.
Nassim can appear to be overly verbose but his books defy summarization. I recommend reading them but given my own experience can empathize by the contra opinions.
Antifragile: be convex when applicable.
Black Swan: not all distributions are made the same.
"It is my ambition to say in ten sentences what everyone else says in a whole book"
https://www.mayoclinic.org/diseases-conditions/coronavirus/i...
It's the opposite of Taleb. Basically, if people were rational and used financial instruments properly, a lot of the risks of our lives can be mitigated. The trick isn't so much to control everything but to trade the less desirable forms of risk for the ones you can handle.
Ride The Tiger by Julius Evola.
Unlike most of these suggestions this book will actually give you purpose and meaning in life, but you will be vilified for even mentioning it. That's how you know it's good.
It has been a long time since I read it, but my basic understanding was that a lot of success is from luck and showed how that works. I remember clearly the math around investment managers being mostly lucky to have returns above the average yet they get a lot of praise and then their luck runs out. But there are still some that stay lucky for a long time.
If I was a successful investment manager I might question my success.
I have not read Thinking Fast and Slow ether but really disliked "The tipping point" by Malcolm Gladwell.
Most impressive book: Thus spoke Zarathustra.
You may also try Memories, Dreams, Reflections by Carl Jung
But in general there is no such book as there is no universal meaning. One has to find it oneself.
I would suggest to try to read writing of Epicurus instead. At least he did followed own principles as far as we know.
Edit: love how I’m getting downvotes for a factual statement
To sum it up: do what is at the limit of your competence and what you feel very drawn too (as in very interested in it). That’s where you’ll find a balance of your existing knowledge and pressure to expand your horizons, plus you are following your passion. That gives meaning.
It’s certainly worked for me recently. It’s very good advice.
I came here to say the same! I’ve been thinking about this recently. It began after hearing the Weinstein brothers talk about Taleb[1]. My hunch is that a significant part of Taleb’s popularity stems from how dogmatically polarizing his views are. I’m not sure if he’s doing so authentically (I think he is), or: if he’s dogmatic in his opinions primarily because he knows doing so will result in wider distribution of his views.
I struggle with this dilemma.
On one hand, there’s a case for being strong-willed, dogmatic and polarizing. It takes courage.
On the other hand, there’s a case for being open-minded, flexible in your thinking, and uncertain. Yet, this is generally perceived as weak-willed and doesn’t attract nearly the same attention as the polarizing stance.
[1]: https://youtu.be/XxlVAbc1Vjo
"""An Outside Context Problem was the sort of thing most civilisations encountered just once, and which they tended to encounter rather in the same way a sentence encountered a full stop."""
https://en.wikipedia.org/wiki/Excession#Outside_Context_Prob...