1,515 of 1,646 people found the following review helpful
Lost in Extremistan with nothing but a Bell Curve,
This review is from: The Black Swan: The Impact of the Highly Improbable (Incerto) (Hardcover)
If, as Socrates would have it, the only true knowledge is knowledge of one's own ignorance, then Nassim Nicholas Taleb is the world's greatest living teacher. In The Black Swan, Taleb's second book for laypeople, he gives a full treatment to concepts briefly explored in his first book "Fooled by Randomness." The Black Swan is basically a sequel to that book, but much more focused, detailed and scholarly. This is a book of serious philosophy that reads like a stand-up comedy routine. (Think Larry David...)
The Black Swan is probably the strongest statement of enlightened empiricism since Ernst Mach refused to acknowledge the existence of the atom. Of course, in theory, everyone today is supposed to be an empiricist - all right-thinking intellectuals claim to base their views solely on positive scientific observation. But very few sincerely confront the implications of rigorous empiricism. Specifically, few confront "the problem of induction," illustrated here by the story of the black swan.
Briefly: observing an event once does not predict it will occur again in the future. This remains true regardless of the number of observations one adds to the pile. Or, as Taleb, recapitulating David Hume, has it: the observation of even a million white swans does not justify the statement "all swans are white." There is no way to know that somewhere out there a black swan is not hiding, disproving the rule and nullifying our "knowledge" of swans. The problem of induction tells us that we cannot really learn from our experiences. It makes knowledge very problematic, if not impossible. And yet, humans do behave -almost without exception- as though they believe that experience teaches us lessons. This is forgivable; there is no better path to knowledge. But before proceeding, one must account for the limits that the problem of induction places on our claims to knowledge. And humans seem, at every turn, to lack this critical self-awareness.
In one of the many humorous anecdotes that seem to comprise this entire book, Taleb recounts how he learned his extreme skepticism from his first boss, a French gentleman trader who insisted that he should not worry about the fluctuating values of economic indicators. (Indeed, Taleb proudly declares that, to this day, he remains blissfully ignorant of supposedly crucial "indicators" like housing starts and consumer spending. This is a shocking statement from a guy whose day job is managing a hedge fund.) Even if these "common knowledge" indicators are predictive of anything (dubious - see above), they are useless to you because everyone else is already accounting for them. They are "white swans," or common sense. Regardless of their magnitude, white swans are basically irrelevant to the trader - they have already been impounded into the market. In this environment, one can only profitably concern oneself with those bets which others are systematically ignoring - bets on those highly unlikely, but highly consequential events that utterly defy the conventional wisdom. What Taleb ought to worry about, the Frenchman warned, was not the prospect of a quarter-percent rise in interest rates, but a plane hitting the World Trade Center!
Yep, the precise facts of 9-11 were actually presaged by this French gentlemen, as a rogue wave that just might be lurking over the horizon. And, to the contemporary American mind, this is THE quintessential Black Swan. Of course, the Frenchman's insight was just a coincidence - the thing with Black Swans is that they cannot be foreseen.
Taleb explains that conventional social scientists use induction to collect data, which is then plotted on the good old Gaussian bellcurve. With characteristic silliness, Taleb dubs the land of the bellcurve "Mediocristan" - and informs us that it is the natural habitat of the white swan. He contrasts Mediocristan with "Extremistan" - where chaos reigns, the wholly unexpected happens, power laws and fractal geometry apply and the bellcurve does not. Taleb's fictional/metaphorical 'stans' share something with the 'stans' of the real world: very ill-defined borders. Indeed, one can never tell whether one is in the relatively safe territory of Mediocristan or if one has wandered into the lawless tribal regions of Extremistan. The bellcurve can only help you in Mediocristan, but you have no way of knowing whether you have strayed into Extremistan - beyond the bellcurve's jurisdiction. This means that bellcurves are of no reliable use, anywhere. The full implications of this take a while to sink in, and are sure to cause huge controversy. In July, Taleb will debate Charles Murray (author of -what else?- the Bell Curve). I'll let you know who wins.
Taleb frames his whole argument much more entertainingly than I could here, and he bolsters it with an astonishing command of both cutting-edge social science and the entire history of philosophy. This is an astonishing work of serious philosophy, and it reads like pulp fiction. Readers who enjoyed FBR will find here the same dry wit, the same literary erudition, and deep sense of the absurd that made that book so much fun. But this is better, by an order of magnitude - easily the best book I have read in 5 years. I smell a timely pop-science bestseller here to rival Gladwell or Surowiecki, but this is also a classic that will be read for decades to come.
Tracked by 3 customers
Sort: Oldest first | Newest first
Showing 1-10 of 69 posts in this discussion
Initial post: Apr 24, 2007 4:52:36 AM PDT
[Deleted by the author on Apr 26, 2007 3:52:13 AM PDT]
In reply to an earlier post on Apr 24, 2007 8:23:05 AM PDT
Last edited by the author on Apr 24, 2007 8:31:49 AM PDT
Keith M. Robertson says:
Mr. Jones, I beseech you as another gentleman to read more closely. The only person described as being French is Taleb's first boss, not Taleb himself, and the "Frenchman" further mentioned is not Taleb but this Boss. The ethnicity of Mr. Taleb is never commented upon, thus the gentleman reviewer, our Dilettante, is not guilty of making a mistake and thus not in need of correcting.
In reply to an earlier post on Apr 24, 2007 9:21:36 AM PDT
Last edited by the author on Apr 24, 2007 9:34:46 AM PDT
Thank you both, gentlemen! Mr. Robertson is indeed reading my post as I intended it. Forgive me for any ambiguity.
Posted on Apr 25, 2007 1:22:27 PM PDT
Last edited by the author on Apr 25, 2007 1:23:54 PM PDT
If one sees a million white swans, it does not disprove that a black swan exists. But it gives you million-to-one odds that the next swan you'll see is white.
In reply to an earlier post on Apr 25, 2007 1:52:30 PM PDT
John Maynard Keynes would say that the odds are greatly in your favor and that the weight of the evidence supporting the reliability of the odds is extremely high.However,we haven't mentioned what the outcome's magnitude is.One of Taleb's points is not only that the normal distribution severely underestimates the probability of so called outlier(extreme)events but that the effect or impact of these events are far,far greater than the average events that hover around the mean.It is like a white swan's occurrence would result in a gain or loss of $5.00 but the gain or loss resulting from a black swan is $5 million.
In reply to an earlier post on Apr 25, 2007 3:55:49 PM PDT
Last edited by the author on Apr 25, 2007 6:24:39 PM PDT
If I understand Taleb correctly, there are two separate issues here. First, I read him to explicitly disagree with Razzi - that a million observations do NOT give you million-to-one odds (though this is indeed the assumption that we humans usually operate under). I may be wrong here, but my read was that past observations literally have no predictive value. (He illustrates this with a funny story about a turkey - but just read the book.) This is counter-intuitive, but I think it is the strong empiricist position.
Second, he emphasizes the problem of infrequent events of great magnitude - the problem identified by Mr. Brady.
In reply to an earlier post on Apr 25, 2007 7:30:56 PM PDT
It may be that Taleb is taking the extreme nihilist position of G L S Shackle(that the weight of the evidence supporting a probability assessment is always equal to 0 so that we make our decisions in complete ignorance rather than the partial ignorance emphasized by Keynes).However,I think he is stating that this is the case some of the time and not all of the time.It is the case if we have no knowledge of the process or mechanism that is generating the stochastic time series.The problem is that we do not know.,except for periods of time that are short run and stationary, what is generating the process in financial markets due to constantly changing expectations about the future
In reply to an earlier post on Apr 25, 2007 8:40:18 PM PDT
That is a very useful explanation, thanks. I will look into the Keynes/Shackle debate. I am tempted to question you further on this, but I sense that it would lead us into an eternal regression of hairsplitting distinctions at the fractal boundary between knowledge and certainty [or ignorance/partial ignorance]. I don't want to put too fine a point on it, so I will leave it at that.
(You say pot-ay-to, I say po-tah-to!)
In reply to an earlier post on Apr 26, 2007 2:14:16 AM PDT
I see your point.However,let me just write down Keynes's weight of the evidence index,w,where w is normalized on the unit interval between 0 and 1(Keynes,1921, A treatise on Probability,p.315 and p.315,ft.2).If you have a complete information or data set or knowledge base or can specify a specific probability distribution,w=1.If you are completely ignorant,w=0.Partial ignorance(knowledge) is between 0 and 1.Keynes was the first to come up with such an index.Daniel Ellsberg came up with a very similar index,rho, in 1961(see his 2001 book,Risk,Ambiguity and Decision,pp.194-196).Following the English philosopher L Jonathan Cohen's lead ,Amos Tversky presented a similar measure called degree of support,s,in 1994.Both rho and s are defined on the unit interval [0,1].
You are correct that there could be a lot of disagreement about where you are on the index.Nevertheless,it is very useful because it shows the limitations of using/relying on probability alone,however defined.
In reply to an earlier post on Apr 28, 2007 6:43:22 AM PDT
Eclectic Reader says:
In reply to Razzi's post,
This is a very common statistical fallacy. In fact, seeing a million white swans, without more, tells you absolutely nothing about the probability that the next swan you see will be white or not. Bayesian probability theory disproves this conclusively. For example, if the white swans I saw were all one "community" of swans that did not mix genetically with other swans, then the next flock I see might or might not have black swans in it--it would certainly not be a 1-in-a-million shot.