Fooled by Randomness

Max Tardiveau writes "I just finished Nassim Nicholas Taleb's Fooled by Randomness. It is an enjoyable book, written engagingly by an interesting character -- the kind of book that makes you think twice about certain things (for instance, the fact that you're not dead: is that really because you're so darn good, or does dumb luck play a part?) Although written all the way back in 2001, this book is more relevant than ever, since one of its major topics is the impact of unpredictable events on markets, insurance, and our perception of life in general. In fact, Taleb makes a living from unforeseen events; these days, that seems like a rather cunning niche." Read on for the rest of his review of this book. Fooled by Randomness author Nassim Nicholas Taleb pages 220 publisher Texere rating 8 reviewer Max Tardiveau ISBN 1587990717 summary Debunking fallacies of observation, Taleb reminds us of the pervasive ineffables that complicate life at every turn.

The main topic of the book is the fact that all humans are simply terrible at judging probabilities. Taleb is a securities trader, so a lot of the book revolves around financial probabilities, but his argument is mainstream and requires absolutely no knowledge of the markets. The book details examples of people wildly misjudging risks and probabilities in many contexts. Often that misestimation is understandable in advance of certain events, but harder to excuse after they've occurred; Taleb hits pretty hard on what he calls "data snoopers," his term for people who back-fit theories to existing data.

One of the most notorious examples is the Bible code (which has been thoroughly debunked), but Taleb argues that analysts who spend their time trying to find patterns in historical market data are no different: if you try long enough and hard enough, you will unavoidably find apparent regularities, which can be extremely compelling when seen in isolation. In context, though, they dissolve into nothing but meaningless statistical anomalies. Taleb rightfully compares these searches for meaning to the famous monkeys and typewriters parable.

Taleb's best example of poor probability intuition is probably the infamous survivor bias, which is our tendency to be disproportionately impressed by success. We almost always ignore the fact that, for one success story, there are many failures. But we seldom hear about the failures (just like we never hear about the many theories that didn't fit the data). So it's all a game of numbers: out of 10,000 traders, a few are statistically bound to be successful, even if they are nothing more than lucky idiots. The fact that they succeeded does not mean anything. It doesn't mean that they are bad traders, but it doesn't mean that they are good traders either, because on average somebody had to succeed.

One of Taleb's hot buttons is that people tend to be too confident in what they know. He argues convincingly that we should take everything, including science, with a grain of salt. Writing about Karl Popper, he points out that there are only two kinds of scientific theories: those that are demonstrably false, and those that are not yet demonstrably false. An irksome (but sadly true) observation, yet most people behave as if what they know is eternal truth. One could of course argue that Popper's observation is but another kind of truth, but I tend to put a lot more trust in people who question what they know than in people who don't.

Another of Taleb's peeves is the human tendency to over-attribute every random event (the old post hoc, ergo propter hoc). For instance, a commentator saying that "the Dow went down ten points today on concerns about Iraq" is talking nonsense: there is no way anyone can tie such a small market move to any particular reason. I found this specific point (which in retrospect is blindingly obvious) especially enlightening, as I am embarrassed to admit that, until now, I just accepted those market comments at face value.

Taleb also has some fun at the expense of economists and analysts, especially those whose predictions turned out wrong, but who claim that the theories were in fact right, and that the facts simply weren't supposed to be that way. This is what he calls denial of history, and is common among investors and gamblers (the two being of course close cousins).

The style of the book is informal and funny, and often meandering. We hop from one topic to the next, which occasionally may detract from the book's continuity, but overall the author's points come through loud and clear. Ironically for a man who advocates self-doubt, Taleb is starkly self-confident, though not in an irritating way.

Taleb is an intriguing, multi-cultural, iconoclastic character that has been around Wall Street for a while, and now runs his own small firm. Malcolm Gladwell (author of The Tipping Point, an absolute must-read for anyone who owns a brain) has written an excellent article that shows how Taleb's reasoning runs counter to just about every bit of conventional Wall Street wisdom. If you're interested in the markets, especially derivatives, and how Taleb trounces most of Wall Street's voodoo doctors, this moderately technical interview from 1996 is worth reading too.

Overall, a warmly recommended book.

You can purchase Fooled by Randomness from bn.com. Slashdot welcomes readers' book reviews -- to see your own review here, read the book review guidelines, then visit the submission page.

Comments

Conservative/Liberal take on it

[SirSlud]

> (for instance, the fact that you're not dead: is that really because you're so darn good, or does dumb luck play a part?)

Thats called the Just World mentality. Its not a Just World .. unfair things happen through dumb luck.

I've found this is the single biggest commenality in partiasian schools of thought:

Conservatives tend to think you are where you are because you deserved it. Dumb and lazy people are poor, smart and hard working people are rich. Dumb people get hit by trains, smart people comment on how dumb they are. Your situation is a result of your disposition.

Liberals go the other way. Luck, circumstance, and opportunity play huge roles in where you are. Poor people are poor because of luck and circumstance. People get hit by trains because they might have just plain been unlucky. Your situation is a result of your environment, including dumb luck.

Personally, this is the single biggest reason I can't stand conservatives. It bothers me to no end how capable they are of assuming that anybody in a bad situation is there because they deserve it.

Re:Conservative/Liberal take on it

[dracken]

People break down into two groups when the experience something lucky. Group number one sees it as more than luck, more than coincidence. They see it as a sign, evidence, that there is someone up there, watching over them. Group number two sees it as just pure luck. Just happy chance. And surely, the people in Group number two are looking at those fourteen lights in very suspicious way. For them, the situation is fifty-fifty. Could be bad, Could be good. But deep down, they feel that whatever happens, they're on their own. And that fills them with fear. Yeah, there are those people. But there's a whole lot of people in the Group number one. When they see those fourteen lights, they're looking at a miracle. And deep down, they feel that whatever's going to happen, there will be someone their to help them. And that fills them with hope. So what you have to ask yourself is what kind of person are you: are you the kind that sees signs, sees miracles? Or do you believe that people just get lucky? Or, look at the question this way: Is it possible that there are no coincidences?

Huh ? this seems to be a perfect plot for a movie. Wait - nevermind ;)

Re:Conservative/Liberal take on it

[phossie]

the belief that if someone works hard and consistently, they can improve their condition

Unfortunately, a huge number of people who identify themselves as conservative don't seem to realize that condition++ doesn't necessarily meet condition >= decency, or even condition >= survival. It is easily, easily possible to be so low that incremental improvement through hard work is still not enough to get you to the point where you'll be on your own. There is a threshold below which basic 'modern' living is not possible. This is where having basic social services is really handy. If you can get someone above that threshold, they can often stay above that threshold on their own. It's kind of the modern equivalent of knowing how to make fire, or knowing how to fish...

It's also a shame that so few people in the US have any understanding - or put any thought into - the meanings of their tidy political categories (e.g. "liberal" and "conservative"). It's sickening how often those terms are applied in total disregard to their denotations.

Ok, that's all for now. Time for more coffee.

Yep

[Anonymous Coward]

Liberals tend to think that luck, circumstance, and opportunity play huge roles in where you are. Poor people are poor because of luck and circumstance. People get hit by trains while trying to get through the tracks before the train gets there, are just plain unlucky. Your situation is a result of your environment, including dumb luck.

Personally, this is the single biggest reason I can't stand liberals. It bothers me to no end how capable they are of assuming that anybody in a bad situation is there because of bad luck or circumstances beyond their control.

Re:Yep

[mfrank]

It's been theorized that that is the reason why so many actors and actresses are liberal, because for them "luck, circumstance, and opportunity" play huge roles in getting them where they are, and deep down inside, they know it.

Random walks

[watchful.babbler]

As provocative as the book's thesis seems to be -- and I must admit that my information on it comes entirely from this review -- it's not new. In the 1970s, Burton Malkiel's A Random Walk Down Wall Street posited that market fluctuations are mutually independent, though the market follows a general upwards trend, and thus it's impossible (ceteris paribus) to make any bets on short-term stock performance. The upshot is that the only way to beat benchmark indexes is to assume additional risk: trying to beat the market in most cases is nothing more than gambling on an upmarket roulette wheel.

Re:Random walks

[devonbowen]

The problem is that markets are provably not random walks. A random walk would result in a gaussian distribution of price moves but it is well known that the distribution of financial markets have "fat tails" (heteroskedasticity) when compared to a gaussian. Meaning that extreme events like crashes happen more frequently than they are supposed to. One can question whether these deviations from random are sufficient to allow any prediction. But stating that the markets are a "random walk" is really just wrong.

Oh, and the above assumes any general "upward trend" in, for example, a stock market, is taken out of the data. Some markets (like currency, for example) don't have an upward trend.

Devon

Re:Random walks

[richg74]

Actually, the idea is older than that. There is another very amusing book called, Where are the Customers' Yachts?, by Fred Schwed, first published ca. 1940. In an appendix, he describes a coin-tossing championship along the lines of Hitchcock's swindler, which someone else mentioned. At the end, "there are the winners, the coin-tossing experts who have flipped ten heads in a row. They cannot lose."

You can see a reprinted edition at Amazon:

http://www.amazon.com/exec/obidos/ASIN/0471119784

You might also want to have a look at Extraordinary Popular Delusions and the Madness of Crowds, by Charles Mackay.

Surfing on the Noise Floor

[Alien54]

A lot of times you can see a legitimate trend in a noisy signal. The problem is to determine when you are looking at variations that are in fact noise produced, vs legit trends. And humans seem to be geared to picking out a signal from the noise in the first place, even if they often get it wrong.

The stock market, day traders, etc are all dependant of this sort of thing. even though the possibily that their successes are often not distinguishable from random luck gives them all nightmares.

kinda difficult to pick out signals where they are buried in the noise floor to begin with.

he has some good points, but also overlooks a few

[AssFace]

The main issue I have with what he says is comparing Time Series analysis (a topic that is near and dear to my heart), to the people that read the bible and then find prophetic phrases in there.
Aside from that being a bad analogy (but it sounds good if you don't know anything of time series analysis), it is also misleading in that time series analysis can be proven if it will work or not.

There is an equation (that is f'ing driving me nuts right now that I can't recall the name of it) that will tell you if your time series data is truly random or if it leads to predictability.
It has been shown over and over again that the stock market, were it a perfect market would in fact be random - but since there is human intervention that drives it, it is no longer truly random and there are non-linear patterns that are within the data.

The human brain is wired to pick up on linear patterns (to the point we see them where they are in fact not really there - it is advantageous to us to see a lion in the brush and run - when in fact there isn't a lion there, than the other way around and get mauled by a lion that we never saw).
That is more to what he speaks of, humans seeing what isn't there (there are a crapload of great books written about this - How We Know What Isn't So: The Fallibility of Human Reason in Everyday Life by Thomas Gilovich immediately comes to mind) - but the way he speaks, he also includes computational analysis in there and is wrong in his generalization.

Now it is going to bug me all day as to what that frickin' algorithm is called - the one that shows the strength of your time series data... I know it doesn't start with an A, or a Z... and I don't think there is a C there... other than that, I'm pretty much at a total blank.

hitchcock stock scam

[louabill]

Hmm... there was a Hitchcock episode about a scammer who chose 2^n folks (I think n=12 or 13), and

1. mailed 1/2 the folks a prediction that stock X would rise, mailed the other 1/2 the opposite.
2. threw away the 1/2 who ended up with the incorrect prediction
3. repeated until there were a few folks left who had gotten 10 correct predictions (either 4 or 8 people, depending on n)
4. asked the remaining folks for investments.
5. split with the cash

Now imagine stock analysts all making predictions. Eventually there will be a star.

Wasn't there something in Ecclesiastes about this?

[dpbsmith]

...something about seeing under the sun, that the race is not to the swift, nor the battle to the strong, neither yet bread to the wise, nor yet riches to men of understanding, nor yet favour to men of skill; but time and chance happeneth to them all?

I guess it just goes to show that there is nothing new under the sun.

estimation

[Parsec]

Something else I've noticed people being bad at is estimating ratios. Example: back in high school an English teacher asked the class what the population ratio of blacks to whites in Michigan was. Some guessed as high as 50%, but I was the only one with the correct (lowest) answer of around 10%.

I have always wondered what affected this the most. It could be that we were a suburb of Detroit, or that the local television stations' news programs may have featured more black people (scaring white viewers makes for good ratings), or that these kids hadn't travelled very far from the Detroit area, or that they were just really lousy at estimation of distances and population and had no concept of the size of the state.

Certainly, some of the error may have come from cultural bias. I didn't notice any overt racism in the school, but the specter of the "dangerous negro" was/is probably still present.

Is this still on-topic? Anyway, it seems that we need to do better in our educational system of teaching people how to get their heads around the concepts of chance and estimation. Letting any bias influence these will cause errors.

Quantifying the randomness

[Inflatable Hippo]

Fantastic sounding book, I've ordered it.

I've long since been a believer in the "largely random" nature of the world and this puts me in mind of a wonderful piece of "research" I came across in New Scientist a few years ago.

It cut through a lot of bullshit to ask the question:

"can you predict the longevity of a system or state if you know almost nothing about it"?

i.e. life's pretty random and bothering to analyse the details will only get you so far :-)

So how can we address, from a probabalistic standpoint, questions like:
"How long will the pyramids stand"?
"How long before my bank goes bust"?
"How long before we're hit by an asteroid"?

The reasoning went like this:

Take any "thing"/"system" etc. with a finite existance and ask the question:

"With a 95% certainty, how much time does it have left"?

For the sake of argument lets say you plant a tree, and you know nothing about trees.

Ten years later you come back and the tree is still alive and so you ask the above question about the tree.

Well, the tree is clearly alive but you don't know where "now" is in it's lifespan.

From a purely probabalistic point of view you've got a 50% chance that "now" is the middle half of it's life, i.e. you're not in the 25% of it's "youth" or the 25% of it's old age. So thats a 50% chance that it's had 75% of it's life already (middle age + youth).

Turning this into an equation, if:
x = current age
y = total age
p = probability

then: (it helps with a diagram)
x = py + (y-py)/2
or, rearranging to solve for y:
y = 2x/(p+1)

Going back to our tree then, if it's 10 years old
it's got:
95% chance of making it to 10.26 years
50% chance of making it to 13.33 years

The magic figure I keep in my head is that if you know NOTHING about something, you've got a 95% chance that it will last 1/39th as long again.

I find this little nugget invaluable when considering how much to spend on insurance or investments - cos I know/care NOTHING about them.

After all, why bother with the details, life is random :-)

Re:Quantifying the randomness

[Anonymous Coward]

I don't know who wrote the article you read but it sounds like Richard Gott.

If I recall correctly, this approach predicts that the demise of a thing that has lasted x years has a 95% probability of coming between x/39 and 39x years later. So if human civilization has lasted 4000 years (just to pick a number) then it should, with very high probability, last more than ~100 years, but less than 156000 years, from today. There's a 5% chance of it lasting either less than 100 years or more than 156000 years from today.

Comment #1: That's a pretty wide span, so it isn't all that useful a prediction.

Comment #2: At what point should the measurement of a phenomenon begin? Did civilization begin with the bronze age or the stone age? One could pick different points depending on how civilization is defined.

Comment #3: It doesn't tell us what happens at the end. Will civilization come to a complete end at that time, or will it be replaced by something else? In other words, maybe we are only predicting the "terran phase" of civilization.

Comment #4 (related to #3): Doesn't work with cyclic phenomena. What if civilization develops in a series of waves, with the most recent wave starting from the Renaissance and ending with another Dark Age N years from now? Perhaps you can apply it to the current wave but not to all of civilization.

Comment #5: If there is any a priori information that is knowable (but perhaps hidden from you) then your estimate is useless. For instance, that 10-year old tree could have quite a different life expectancy if it's a fir on a Christmas tree farm, than if it's an oak, or a sequoia, or a bristlecone pine.

So, while there's some validity in Gott's thinking, use some discretion in applying it.

More like this...

[toganet]

Check out these books by Robyn M. Dawes, another heavywieght in the emerging field of 'decision science.'

Rational Choice in an Uncertain World: The Psychology of Judgement and Decision Making

Everyday Irrationality: How Pseudo-Scientists, Lunatics, and the Rest of Us Systematically Fail to Think Rationally

House of Cards: Psychology and Psychotherapy Built on Myth

Choices.

[Lemmy Caution]

The thing is that "choices" is bandied about in this situation as a complete black-box, without asking how people make choices, how they percieve what those choices are, and when the understand what the consequences are. Largely, the "choice-making" facility that the poor and unfortunate are failing at are failures of education to begin with (both formal - understand just what is possible and what input creates what output - and informal - understanding social patterns and behaviour outside of one's own in-group).

It's why people from "better" backgrounds can recover from poverty quickly - why good business-people can create new businesses from catastrophes repeatedly - and why very hard working people who don't know how to work tactically can still die poor. Knowledge and social networks are what keeps one out of poverty, not hard work.

As an aside, drug addiction exists in that gray area between choice- and not-choice, since it can be thought of practically as a "disease of the choice-making mechanisms of the mind." When drug addiction is responsible for poverty, there's little chance of escaping the poverty without addressing the addiction; likewise for mental illness.

Re:Choices.

[osgeek]

It's why people from "better" backgrounds can recover from poverty quickly - why good business-people can create new businesses from catastrophes repeatedly

It could be equally argued that they were in those "better" backgrounds because of some genetic advantage. You'd really need to do a large number of controlled studies to determine the order of the cause and effect there. My suspicion is that both genetic propensities and social frameworks play a part.

Knowledge and social networks are what keeps one out of poverty, not hard work.

Hard work is an important element, and you can't just discount it out of hand.

As an aside, drug addiction exists in that gray area between choice- and not-choice

What is a choice, then? I know smart people who are "addicted" to EverQuest to the point that their work quality seriously suffers. Are they making a choice? If I go home early to watch TV or read a book, am I making a choice or just a slave to some electrochemical impulse in my brain?