Pages

May 18, 2014

Book Review: "The Black Swan" by Nassim Nicholas Taleb

1. Alex Plays Auto-Psychiatrist

Possibly nothing is more infuriating and terrifying to me than a very particular thought. Let me explain. It's a thought braided together from four essential strands, each strand by itself fairly innocuous, but that in combination - and it's not so rare a combination - constitutes a devastating narrative, psychologically, intellectually, spiritually, and causally. Here goes:

  1. I could have had success in the present if I'd followed a specific past course of action.
  2. I may be hurtling towards failure in the future if I don't change my specific present course of action.
  3. #1 and #2 above are both plausible and - to recognize this - neither requires (or required) any special foresight or ability.
  4. #3 and #2 are true, but I still may be powerless to change course and avoid a future failure.
I wonder whether I even need to elaborate further, or if this is so universal and clearly-framed a alienation quadrilogy that you can instantly relate to my terror. Whatever the case, this braiding of thoughts gives off a Greek tragedy kind of vibe. But it's not even like Sisyphus where there may be something somehow ennobling that you could find in the absurd - it's more like Prufrock or Uncle Vanya: Otherwise-respectable folks hurtling towards failure and oblivion at the speed of light, and noticing that not much of substance has been hurtled so far in one's life. And it's - in large part - the very "otherwise-respectable" part that leads them on the path to doom. It's the heroic tragedy of the non-heroic: It's the feeling of tragedy without even the feeling that one has risen to heights that would at least validate the tragic flaw's value in moderation. 

And, for me, the tragedy unfolding comes from an idleness and an inability to be activated by a hazy future.

Oh, don't get me wrong: I'm pretty happy, stable, and content with what I've done so far, and hopeful for what I'll do in the future. But all this positive energy hasn't always translated into sufficient action in the past, and so I'm right to worry about this larger stuff on occasion, at least as a personal call-to-arms in the visceral present. 

2. Alex Now Directs That Productive Rage Towards Others

And, for all the bluster of the preceding paragraphs, there's a related quinella of thoughts that outdoes all of the previous thoughts in terms of rage, if not so much in existential fear. Here it goes:
  1. I could have done this successful thing that I see before me if only I'd sacrificed a whole bunch of my principles.
  2. #1 is true, and yet, a lot of people I respect see it, and seem to ignore - willfully or not - the latter principles sacrificed for the former success.
And note that this isn't envy: it's me jealously holding onto my principles, and resenting someone who has sold out these principles at apparently no social cost (and social cost as I define it, which adds to the insult). For those that know me, I'm not the kind of person that's rude to workers at fast food restaurants because they didn't get my order exactly right. And I'd like to think that if you saw me screaming at the counter at, say, Chipotle for giving me carne asada when I asked for fucking barbacoa so simple gosh, well, I'd like to think that you that know me would respect me just a little bit less as a human being, or else realize that something is seriously wrong with me today. Ideally, you'd slap some sense into me. And, for the most people, they do. I was a jerk a few weeks ago, and a friend berated me; I slept through something I should've pushed through, and my family told me so. To their great credit. The people around you maintain your standards - it's a large part of what having a Framily really means. It's... très good.

But when I see someone doing more or less the intellectual equivalent of the same fuckery to the public-at-large instead of one unfortunate soul at Chipotle, well, sometimes it doesn't change at all how people see the "perp" - not even for the people I respect the most--not even the people that I'd hope would slap some sense into me if I'd done the same. It's a dirty feeling to watch people I see as charlatans go unpunished in the esteem of those you esteem the most. It's knowing that either you're not living up to your potential and those around you refuse to confront you with suggestions they have no qualms with themselves, or that you're doing these same horrible things as the people you critique, and the people around you are never going to tell you that you're a hypocrite unaware. Or that you're so blinded by your principles that you can't see this legitimately virtuous or intellectually stimulated thing that you - if only you had the presence of mind to see past the bad parts! - could immeasurably gain from. Or, conceivably, that those you respect the most just aren't able to see the same tics and lies and foibles that will haunt you to your grave to see.

And yet, my default response to this infamous quinella of thoughts is not really any of these introspective fears but a kind of instant sympathy for the poor Chipotle worker (and the public) and an instant rage against the screamer. How dare you ruin someone's day for the most inconsequential of reasons?

And, yes, sometimes, the silly and downright evil things that writers and politicians and such will do are dismissed as part of the nature of the game, as inexorable (and thus excusable); sometimes, people just plain love the cattiness and drama of a feud. And, hey, it's stupid to get so mad over carne asada, but it's probably a great story later on! And Obama probably doesn't have as much influence in the U.S. political establishment as we think, and, as far as we of the public know of the influences guiding our politics, maybe civil liberties are the lowest possible cost for a stable society. And maybe it was obvious they were spying on us. And so on. So I try every day to hold my tongue a bit.

Whatever the case, I've been speaking somewhat in generalities so far, and I know that's not ideal. So thank you for your patience. But I'm building towards what to me is the epochal intellectual insult of the last 5 years for me: Trying to read The Black Swan by Nassim Nicholas Taleb.

3. "The Black Swan"

Now, The Black Swan isn't the worst book in the history of civilization, and this isn't the kind of cutesy blog where I follow that up with "At least it's not Atlas Shrugged LMAO". No; The Black Swan is not some monstrous entity that is going to cause a war in five years or something.

And it's not even that bad.

And, for some people, it's even pretty good.

It's quite hard to sum the book up in a way that both respects the book's complexity and acknowledges its contradictions. Nevertheless, the basics go like this: Taleb delves into a large web of influences--personal, intellectual, practical, anecdotal, historical, and otherwise--to make a case that the governing forces of our society boil down to rare events known as "Black Swans". These "Black Swans" are unforeseen catastrophes such as war, market crashes and natural disasters--and Taleb makes the case that such catastrophes are not nearly as inevitable as they appear in hindsight--and they dominate our lives in ways that human beings are not cognitively wired to appreciate. And nearly without exception, our formal and informal institutions--academia, risk management, financial systems, investments--are largely built around a vision of society that is misleadingly stable, and whose misleading stability is self-perpetuating, is fed to us by a host of false analogies that prevail. 

And, therefore (the prestige, if you will, for you Nolan fans) Taleb proposes that the way to thrive in a world of "Black Swans" is to embrace an extreme overall perspective of extreme uncertainty in the world, to use exponential growth to exploit ultra-high-upside opportunities, to manage risk with an extreme attitude of skepticism towards even the most basic assumptions about the future, to think about the world in a fundamentally different ideological way, and in doing so to reject the established ideology of the staid and outmoded traditions built around familiar forms and skinny-tailed distributions that we are all taught to swallow wholesale. So far so good.

So that's where the book finds itself, from a content standpoint (with necessarily huge oversimplifications; I only read ~60% of the book in 2011, and have only recently started skimming it again). 

Of course, as I noted when I tried to review it a few years ago, my most powerful takeaways were in the absurd way that the book proceeds to attempt those content goals:
 Now, this narrative is innocuous enough, and the empiricism and humility and skepticism the narrative seems to engender should render the book fit for any readers looking for marvelous exposition. But there's a problem. Taleb is a shockingly unpleasant human being, who writes of his exploits with the empiricism of a sales pitch, the skepticism of a high-falutin' charlatan, and the feigned, egregiously false-hearted humility of a pick-up artist. Everyone Taleb decides is wrong is a "fool" or a selfish "fraud", or both. Deprecating humor without real self-deprecation. Human nature causes us to be stupid, okay, but simply by his special life story, he's able to be a lot less stupid, and tell us so, is his main point. He takes a broad, often thoughtful, sweep of the history of ideas, which is commendable. However, his reduction of this history of ideas into his self-serving (to the point of arrogance) and others-tearing-down (to the point of narcissism) is disgusting, and it's hard to take his work seriously. People and ideas he doesn't like are trapped by their own delusions, people he does like are polymaths with good senses of humor. The fox is better than the hedgehog, is what he gets from Isaiah Berlin's famous dichotomy. Everything is about whether these people and ideas grasp "the point" (whatever Taleb is writing about at the time) and if they don't, then they are overrated and discarded. ... In Taleb's hegemony of the Black Swan, Umberto Eco's library and the idea of a library containing vastly more unread books than read books, trumps all of the worthfulness of Plato ("Platonicity", or a tendency towards reductive categorization, is (often rightfully) mocked mercilessly), Kant, and all of economics. Everything that confirms the Black Swan is elevated above everything that doesn't yet understand the brilliance of the Black Swan.
All of that quotation holds up three years later, but I'm little bit of a different person than when I wrote that. Beset by the same haunting beliefs and triggers I started this essay with, and still bothered by any random section I happen to pick up of The Black Swan, but with 34 extra months of humility and life experience to my name outside the university, I feel obliged to illustrate what I see as dangerous and harmful about Taleb's book, with a cooler head.

And a cooler head is necessary. Taleb's book is remarkable: It's the rare book that provokes in me both the rage-inducing braids of thought of the first two sections - the tragic quadrilogy and the infamous quinella. I feel the existence of this book almost as a personal betrayal, which is absurd, of course, but I've at least laid out a subjective framework for my feelings. It's wrong to write a book as he has, he shouldn't have extra social capital because of how he wrote it, and my overall impression of Taleb as a human being is still as petty, self-promoting grifter. And I still think that --with some massive gains in intellectual honesty--what he did could possibly be salvaged as something that I could respect. And so I envy Taleb the opportunity at the same time I shudder at him for having wasted it. So some complex, subjective feelings going on. 

But publishing is a stupid, stupid industry at times, and twice as ruthless. If Taleb is half as bad as he presents himself with his book, he's just an awful guy. But maybe he isn't. Maybe he wanted the book to be two hundred pages longer, but they wouldn't let him (perhaps wisely for his sales). Maybe he wanted to present to a larger audience, and good ideas got oversimplified and broadened until they became the kinds of "Four Legs Bad, Two Legs Good" sloganeering that required an internal glossary for all sorts of cutesy neologisms that already had definitions. Maybe--for all my critiques--Taleb is a legitimate genius, and not just a genius at seeming legitimate, and can genuinely flesh out any of the details in his tour d'horizon. And maybe they're the ones that encouraged Taleb to write (or wrote for him) the most egregiously idiotic footnote in the history of the printing press. (We'll get there.) But, whatever his intentions, his book remains as such. And it's that selfsame, rage-inducing book that people are referring to when they praise him. So we can safely focus on the book itself, and its flaws. And that terrible, terrible footnote that I'm getting to.

I'd like to note here that Taleb has no qualms throwing a genius like Gauss to the curb just because it fits Taleb's pet theory and makes it easier for him to write, and has no qualms about then picking up a lesser light like Zipf or Mandelbrot in a bus and driving over 'ol C.F. Gauss, recently toss'd aside. (details in the "Appendix" of the 2011 review). Look: I have no special reason to care about Gauss, but if you're someone with any statistical inclination or training whatsoever, you'll likely recoil at Taleb's reductive, metaphor-heavy interpretations of the world at large. You'll likely get hot and bothered over Taleb's repeated claim that the normal distribution is a "Great Intellectual Fraud". And you'll likely be unsettled by the unstated premise that economists, academic, and statisticians are more impactful in the world than the kinds of Hank-Scorpio-esque venture capitalists Taleb seems to be implicitly preaching to. And you'll likely either: 1) laugh your head off or 2) drink more alcohol than you should when you read that footnote.

3a. That Footnote

About that footnote: The chapter for "Chapter Fifteen" is "The Bell Curve, That Great Intellectual Fraud*". And it's footnoted - that's right, the chapter title is footnoted. That footnote reads as follows:
"*The nontechnical (or intuitive) reader can skip this chapter, as it goes into some details about the bell curve. Also, you can skip it if you belong to the category of fortunate people who do not know about the bell curve."
Now, unlike Taleb, I make no such requirement or excuse for the people reading this essay. If you're a nontechnical reader, I will try to be as accommodating of you as I can; and if I can't explain it, that's on me, not on you. But I have to say: Whatever your qualifications or mine, I happen to find this footnote remarkably condescending to both of us. There's nothing fortunate in not knowing about the bell curve. According to Taleb's own arguments, even if you accept the absolute audacity of his "Great Intellectual Fraud" claim, then it's probably just a bit helpful to know what is apparently the underlying intellectual basis for many of the power-brokers of this country and world. Knowing about the bell curve, in Taleb's world, in mine, and in yours, is really helpful. But Taleb would rather score a cheap point making nontechnical readers feel good about what they never learned. And there's a dangerous xenophobia and anti-intellectualism implicit in that idea, a line of bullshit something "nontechnical readers" may just be best poised to sniff out, in addition to the fact that he is telling you not to read the chapter: Taleb is saying, in essence (pretend that footnote has a double asterisk at the end):
"**If you don't know about the bell curve, I actually prefer it that way, because just hearing about it is deleterious to your processes of judgment, in my opinion. It's such a convincing fraud, that you can't even be trusted to read it. What's more, those that have been inculcated into the Church of the Bell Curve are so tainted by it that I have to quarantine them and fix their terrible ideologies, and I'm only writing this chapter to accomplish that." 
Just to be clear, that double-asterisk part is a thing I wrote just now, but read the actual footnote above again, and tell me - with albeit perhaps a knowing wink - that Taleb is saying something substantially different. I don't think he is. I think he's trying to encourage a radical, almost nihilistic rejection not only of intellectual authority, but to see even the process of educating one's self in the existing order's tenets as harmful. He's giving you investment advice and telling you to ignore well-established math (con or not, the normal distribution is well-studied and universally respected). What is your first response to that? Some would laugh, some would ignore it, some like me would fly off the handle. But maybe someone is able to see it in good humor as a clever jab and roll with the punches. Intelligent people can disagree. But I come away from it as the worst footnote in the history of the printing press, or at least that I've come across. And my old combinatorics textbook had some bad puns in the footnotes.

To me, it's a footnote that speaks - like mathematician Pierre Fermat's awesomely misguided claim of a proof of FLT in the margins - to centuries of accumulated bullshit, and to a book that popularizes concepts by first totally distorting their place in the world and then oversimplifying on that simplistic premise. Metaphors upon metaphors. I remember once speaking to a guy - in the military, if I recall - that mentioned that his first "tell" for simplistic, wrongheaded presentations was the use of metaphors. Which is a stretch, but there's a grain of truth there (sorry): If you're using a metaphor and never connecting it back to objective reality or established academic concepts? Then you're deceiving your audience with good-sounding metaphors, not communicating in a down-to-earth manner. I realize that some books go beyond the framework of what academics have covered or use the ideas in a different way. Still, Black Swan will often make a specious, easy-to-refute claim like, say, that the bell curve isn't just a distribution but an ideology, and that the two notions (ideology and distribution) are literally interchangeable. The simplest clear distinction between "Gaussian thought/thinking" (Taleb's term) and a Gaussian distribution (an extremely well-understood and well-recognized statistical concept that arises naturally out of the Central Limit Theorem and the simplest of probabilistic assumptions) would have sufficed; there is none, and what's left is that Taleb has just baked a metaphor and an argument into his terms without clearly delineating what he's doing.

And, as much as the financial crisis may force a genuine reevaluation towards academia and the finance industry in some quarters, the retreat, if so, cannot be an escape to the people like Taleb giving you easy answers, asking you to skip the hard parts and telling you that all the hard parts are just elaborate lies, as if academia itself is one giant Sokal affair and all the bankers are frauds. No; the truth is more complicated, and the way Taleb creates simplistic, hostile dichotomies and complicated anti-intellectual constructs and rhetoric seems to me far more dangerous than whatever over-narrative intellectual climate in our universities that Taleb perceives.

The first two sentences of Taleb's Chapter 15 feed my interpretation as well:
 "Forget everything you heard in college statistics or probability theory. If you never took such a class, even better. Let us start from the very beginning"
And thus we have a suitably representative trifecta of chapter title, footnote, and disclaimer. You'd have to be paying almost no attention not to see at least one of these three things as a reader, and even less to have missed the dozens of unexplained references to this "Great Intellectual Fraud: See Chapter 15" thing that kept popping up, seemingly once per page* in the previous 230-some pages. Now, in fairness - because there are about three or four sensible responses to this trifecta, only one or two of which involve fits of rage - I should note that the rest of the chapter is fairly unobjectionable and reasonably well-written, even for someone prone to such fits of rage! 

*Well, it would be Poisson-distributed, which distribution Taleb also condemns as a 'non-scalable sibling' of the Gaussian. I'll permit myself a single chuckle here. Heh.

3b. That Chapter With That Footnote Is Actually Alright

Yes, believe it or not, but - after that crazy, rambling Section 3a., we're going into Section 3b. with a positive attitude. For Taleb's Chapter 15 is actually not as bad as it may seem to those of you with statistical know-how that have followed me this far. Really. Taleb goes into the differences between Gaussian and long-tail distributions with a number of decent examples - with a clarity that makes me wish Taleb'd shoved aside the metaphors and persona sooner and more frequently. If you genuinely didn't know much about statistics, there are worse, much more difficult high-level "nontechnical" introductions to fat-tailed distributions, often using obscure prerequisites and definitions. The metaphor of height for Gaussians and wealth for power law/fat tails is well executed and reasonably down to Earth, even hinting at the relative median/mean differences. Taleb has a cute example along these lines: the book industry is so lopsided towards outliers that -if you knew two books had combined to earn $1 million in sales- you'd expect one book to account for 99.3% of the sales! And, to me anyway, that's pretty fascinating. I hadn't studied these distributions enough to know that off-hand, and the counterexample - say two people's height combines to 14 feet; by far the most likely outcome is for the two people to be 7-footers, because of the extremely skinny-tail of a normal distribution making even a person of height 7'5'' extraordinarily unlikely. Excellent example, and using almost no technical jargon.

Unfortunately, even with the increase in clarity and the legitimately engaging parts, the hegemony of the "Black Swan" remains in Chapter 15. After all, Gaussian distributions, Pareto distributions, Bayesian priors, random walks, exponential distributions, and Poisson processes can't simply coexist. No; Taleb insists on finding reasons why certain types of things are almost never useful; and why others are almost always useful. It can't just be that we developed tools for different toolkits for different problems as necessity and understanding arose: 

It can't be that Gaussian distributions are only there to deal with the presumed independent-and-identically-distributed random error terms in, like, regressions and point spreads. It can't be that these Gaussian problems still arise staggeringly often very naturally, and in part because our society is built in part on precise assembly-line engineering with huge amounts of repeated work in which the rare trillion-µ service time Google request can often be dealt with (and, heck, must be) on its own terms by well-trained data analysts as it arises. No, it can't be. After all, regressions are only applicable in Gaussian thought, and its non-scalable siblings! 

And it can't be that rare-but-catastrophic system shocks can coexist with a meaningful understanding of a system in normal times. 

It can't be that the Poisson process helped us understand e.g. bombing patterns and computer systems, that some distributions evolved for quality control, that even the highest-risk investments implicitly require things like computers and cars and transistors to work almost perfectly* to deliver the goods, and that the struggle for a world with those consistent computers and cars and basic human rights has sometimes required individual- and country-sized risks, many of which have failed in history.

*I mean, he paid lip service to Mediocristan; don't get me wrong.

It can't be that the advances in thought by people like Gauss are important too, and that if I'm modelling a computer system, I'd probably start with a Poisson process for arrivals, assume a fat-tailed distribution for service counters, and use normality to smooth out errors that don't fall within the purview of my model in the short term, and use ad hoc analysis if there are larger systemic or unforeseeable fat-tail problems.

No, no, never ye mind; it's all a Great Intellectual Fraud, so we have to chase it and tease out the "fraud" over and over to prove Taleb isn't just giving us something interesting to know that could help us understand the world a little better. No; God forbid a book just do that without creating notions of macronutrients that are always bad or introverts that are always underutilized or project management techniques that are always great or entrepreneurs that are always... etc.

No; God forbid we get a story about the 99% of things that don't benefit from the book being written or even about the 5% of things that, you know, even with the best of intentions, might go seriously wrong if you change the world unpredictably because it turns out your half-baked dichotomies about stats that totally eschew mathematics until the last possible moment don't actually help anyone solve real-world problems except to decide which speaker to go and see next month, and might actually harm someone because of the usually-careless exposition. How do I know something might go wrong if this book actually acquires some traction in the real world? I don't know; it just sounds right. I read it in a book somewhere! 

No comments:

Post a Comment