Insiderstan and Crankistan; Extremistan and Mediocristan

My experience of Talib wasn’t that different from Kevins -but then we both have a physical science/math background. I did think it was useful because if gives a wide audience a common set of examples/volcabulary to discuss such isuues.

My guess was that during hunter gatherer times (a few tribes still live that way, so I’d rather talk living arrangement than geological time), the black swan consequences were far more limited (-cutting short a life that wasn’t all that long anyway) to have a significant effect on evolution. Unfortunately the emotional style of thinking that evolved is not so well suited to making decisions in the modern world, particularly when it comes to stuff like economics and politics. Without applying at least semi-rigorous epistemology the chances of being led off the tracks in the modern world are very high.

If our ancient ancestors had lived in Extremistan, what more adaptive risk-predicting cognitive mechanisms would evolution have selected for, if not relatively straightforward induction from past experience?

Having spent many years working on this very problem, I can say that the question “how do you reason about black swans, when you have no reason to suspect the existence of a black swan?” is about as hard as they come. Seemingly promising solutions collapse into dust when you poke them. I am close to finishing a paper (with Simon Grant) that seems pretty good to me, but previous efforts have seemed good and have, in retrospect, been quite inadequate.

That’s why the general tone of Taleb’s book “these silly academics never even considered this until The Great Taleb came along to set them straight” puts my teeth on edge, even when I broadly agree with a lot of what is in the book.

It could be that navigating Extremistan with only a map of Mediocristan to go by is better than no map at all.

The map of Mediocristan is the periplum for Extremistan

And I see from the comments at Kevin Drum’s place that this is far from an original observation. White swans all round.

I think Taleb would say (after adequate reflection) that the difference between before and now is that we’re trying to make these models explicit, and allocate risk based on them. Until recently we didn’t really quantify the risks, and in fact Taleb claims that traders were better at respecting the risks before they started following the explicit models.

Isn’t the solution then to create a meta-model which is the old model plus an extra x‘s worth of free-floating riskiness added on top? And isn’t the only reasonable way to determine the size of that extra x of risk —— induction from past experience (and namely that of past black swan experiences)?

So you might think, Bryan, and I’ve tried it as have lots of others, but it turns out to be much more difficult than that. Once you’ve put a number on the x, you’re back in white swan territory. And if you don’t put a number on the x you can’t apply explicit models.

You need to work from induction, but (in my view) the appropriate role of induction is to tell you when you should apply quantified risk analysis and when you should resort to heuristics like the precautionary principle.

Bryan, there is no grand historical model, people did not trade on a certain model and will not do so. Models can explain to some extent what happened in one part of the universe but that’s all. The claims made here are vulgarizing scientistic claims from people with some dangerously naïve conception of science.

(by the way: the human context is for the most part non-genetic, human behaviour has freedom to adapt to it in non-genetic ways (by education for instance) – reference to hunter-gatherers to explain behaviour in what’s an almost purely verbalized context has most minimal relevance, if any)

I refute it thus. The problem with “black swan theory”, if it can be described as a theory, is that its foundational event, the great financial crisis, is not explained by it. It was not impossible to predict or to quantify; in fact it was predicted with some exactitude, using fairly basic heuristics. In fact, there is a considerable literature on financial crises, and we know that: they happen (so they are white swans), they happen frequently in the right conditions, they tend to happen in the same way (the second stage of the rocket), and the consequences are usually much the same.

The fact that a lot of people didn’t predict it, or didn’t pay attention, only goes to show that the chief symptom of an approaching financial crisis is just that. Example:

The biggest concern, many economists say, is that the new mortgages have come onto the market at a time when low interest rates and rapidly rising home prices are the only reality many people can imagine. Families might be making decisions assuming that combination will last forever.
In a speech to bankers in New York, Donald L. Kohn, a Federal Reserve governor, said yesterday that he expected a strong economy over the next few years.
“My message this morning, however, is that this is not a time for complacency,” he said. There is “significant uncertainty,” he added, because some recent financial innovations “have not yet been rigorously stress-tested.”

Note that Kohn clearly knew how and why the crash would come, but wouldn’t believe it. Before some imbecile like Worstall asks why I’m not rich, yes, I did trade on it – I didn’t buy a house in 2006.

@jedharris

I think “no better map than Mediocristan” is roughly right and that Taleb basically says it is the best we can do except we need to remember there are all these black swans just off the borders of the map (however much we improve the map).

Here be dragons, here blemmyes, and here black swans

I didn’t read the book, and what follows is just generic hand-waving. You say

“The Pleistocene wasn’t Mediocristan. Our ancestors were just as bad at estimating risks in their environments as we are in ours.”

but one interpretation would be that these fat-tailed, power-law kinds of distributions which lead to rare events occurring more frequently than expected by expectations based on gaussians are generated by rich get richer or network-effects as described by Price for citations for example. So inasmuch modern people live in these kinds of networked environments where these kinds of effects are more important you could speculate that we are not wired by evolution to take into account these rare events. The hypothesis would be that we are now more “social” or “connected” than on the savannas.

Taleb is a charlatan and his ideas or banal at best. In some cases he is just flat out wrong. It seems like I’m preaching to the choir here, so I’ll leave it at that. (arguments available upon request)

I still haven’t read the book, but I shared a cab with the guy once, and while he didn’t seem unduly modest about his intellectual accomplishments, nor did he seem like the complete jerk that he comes across as being in public …

I’ll skip the evolutionary biology part, except to point out that if evolution doesn’t apply to short-lived species, there’s a Nobel Prize in biology awaiting somebody.

Two comments on: “I recall someone observing that one of the reasons [one of…] for African economic underdevelopment was that the relative extremity of threats regularly posed by the environment [weather, disease, etc] had provoked a culture of extreme economic conservatism. Their swans are white, they’re just vicious, and there are a lot of them. All depends what you mean by ‘Extremistan’, I suppose.”

1) From casual reading, this conservatism is common to peasant cultures everywhere, because there’s a lot which can go wrong, and they’re close to the edge in decent years.
2) This really means that our ancestors lived in Extremistan – when you just getting by, it doesn’t take much going wrong to kill you. In the developed world, the financial meltdown will not result in famine or plagues; in a subsitance society, a partial crop failure would, let alone a general crop failure.

From what I’ve gathered about this financial crisis, the only distinguishing characteristic is that it’s got large elements of a pre-New Deal financial panic, specifically because we rolled back huge chunks of the New Deal. In the end, major financial firms wrote insurance policies for which they didn’t have reserves (AIGFP). They made bets which could be painted as paying off in the short term, while hiding the long term losses, under models which were clearly ridiculous (Fed rates staying ahistorically low indefinitely, housing prices going up rapidly indefinitely) for the simple reason that the top people in these firms could collect vast sums of money on a yearly basis. The rating firms rated anything as good, because they were paid to do so by the issuers of the products they rated.

In short, the financial elites looted the system until it crashed, because it paid so very much in the short and medium terms.

So where does Ethniclashistan fit into this model?

But it seems to me that if a black swan is defined as a risk of unpredictable magnitude and likelihood, then by its very definition it defies modeling. It is in fact defined as something that can’t be modeled. So attempting to create a model that incorporates black swans is a doomed enterprise. Pretty much all you can do is try to turn as many black swans white as you can while recognizing that yes, maybe tomorrow we will be engulfed in the shockwave of a nearby supernova which is currently moving towards us at the speed of light and your carefully planned retirement portfolio will be wiped out with the rest of the planet. So how much should we care about things like that? Well, what’s the sum of an unknowable quantity of unknowable quantities?

“Our ancestors were just as bad at estimating risks in their environments as we are in ours. Their black swans, the stuff that gave them fairly short lives, were just different. Mediocristan isn’t a time or a place, or a primitive development stage, it’s a projection of a confused form of cognition. It’s a place everyone acts as if they live in, but no one has lived there. Mediocristan is only a state of mind. This is the important thing, for Taleb, so I expect he would respond by saying – yeah, whatever. I didn’t write a history of the Pleistocene. But he does repeatedly imply that it’s a state of mind because, once upon a time, we really lived there. He talks as though it’s an adaptive trait that has become mal-adaptive. But that’s wrong, right?”

I don’t think this is quite right. Our ancestors experienced black swan events, but they couldn’t plan around or do anything about them. Also many of the black swan events were randomly fatal, so they didn’t apply selective pressure. If you got hit by a meteorite, it was definitely a black swan event. But your death wasn’t really preventable and it didn’t apply a selective evolutionary pressure that would have changed evolutionary psychology.

Shit happens.

even his paradigm case, the political collapse of Lebanon, was actually pretty widely foreseen

Among other things, because it had actually happened, or rather, it had come very close to happening, in 1958.

Anyway, Taleb argues that primitive man lived in Mediocristan; but we know that they also experienced very high rates of death by violence. We also know that murder rates in Europe have fallen dramatically on several occasions since the early modern period and each time they stayed that way.

Surely nothing can be as extreme a tail risk to the individual than a knife in the back? If you can be murdered at any moment without much certainty of consequences for the murderer, you’re in Extremistan. After all, absolutely nothing would be better conserved in evolutionary terms than traits – whatever they might be – that would reduce the risk of being murdered.

The whole question of an evolutionary basis for our risk attitudes and how we estimate risk is pretty interesting. But one thing to note is that more risk back in the Pleistocene d0esn’t translate into … well, anything in particular. A (not so) straight forward example is risk aversion vs. risk neutrality. Over the long run, the risk neutral actually do better than risk averse (in financial or evolutionary terms).
Changing the frequency of random shocks doesn’t alter the relative payoffs so risk neutral folks still do better, although there’s a lot more volatility. So why are humans risk averse rather than risk neutral? (Assuming we are – it’s probably more complicated than that but at least getting a modicum of an answer to that question is a first step here). Particularly since on an evolutionary time scale only averages should matter?
The closest I can come up with here is that risk averse folks’ (i.e. our) ancestors were free riders. The risk neutral and the risk lovers would adopt new methods and techniques. If these turned out to be better than old ones and non-deadly then the risk adverse got almost the same evolutionary advantage that the initial, risk neutral/loving, adopters got. If these turned out to be a waste of time or deadly, the risk neutral/lovers would die off leaving more risk averse genes in the gene pool.
I’ve tried formalizing this idea and it’s a huge mess.

It’s a reference to Popper, though.

The fact that a lot of people didn’t predict it, or didn’t pay attention, only goes to show that the chief symptom of an approaching financial crisis is just that

Since *every* crisis can be predicted after the fact, it’s *always* true that the crisis occurred because people didn’t listen to the nay-sayers. In other words, Black Swans are events that are unpredictable enough in people’s minds that they *can’t* pay attention to them.

Take a crisis that hasn’t happened yet: Agriculture based on monoculture. Lots of people, whom we are all ignoring, are warning that we’re vulnerable to some disease that could substantially drop world yields leading to world-wide starvation and anarchy.

If it occurs, is this a Black Swan?

The answer, despite it’s theoretical predictability, is yes, because as humans, we’re simply incapable of taking that sort of risk seriously and suffering the competitive disadvantage against the rest of the world that continues to reap the productive benefits of monoculture.

In other words, there are going to be whole classes of theoretically predictable events that are going to be Black Swans in any realistic use of the term.

I think the first and only black swan I ever saw was in Canada in 1962. I was impressed because I had always been fascinated by the ship named after it.

Since my background and preconceptions are rather different from most of the other commentators, I wonder why the opposition to the present age, be it moderate or extreme, is some generalization about peasant cultures or the paleolithic. Looking over the the Mediterranean world from 500 BC to 500 AD (sorry about the abbreviations, old habits die hard) with the perhaps tendentious exception of Gibbon’s beloved Antonine Age I can think of no time where rapid, and for many disastrous, change, what I understand as a black swan in this discussion, wasn’t the norm.

Perhaps connected is the notion that over that millenium it may not ever make sense to speak of “policy” as we might use the word; and only in the time of effective Roman emperors was there a real decision making process.

Henri@38 pithy statement is what the old, “just in case” system of business supply and logistics wherein goods were stored in vast wear-houses which were technically inefficient in a world of white swans but vital to effective survival if a black swan stumbled along was all about. By contrast, “just in time” modern economic logistics is HIGHLY efficient (and effective at the same time), but assumes a world of white swans all the time. Comes a black swan, the lack of a “just in case” system–either primary or fallback–is often fatal. The moral is that wide-ranging, highly rationalized connected systems are, in effect, also very brittle and subject to catastrophic shocks. Disconnected, “inefficient,” stand alone systems that, while connected, may be easily disconnected without transmitting the shock wave of a black swan universally throughout the entire system are much more survivable in the long run, even if inefficient in the short run. People all too often confuse “efficiency” with “effectiveness.” Nirvana obviously is both, but Obama wants to build a “smart” interconnected, highly rationalized national grid that, while it would be both efficient and effective in a world of white swans, might spell utter disaster in the presence of a black swan.. Texas, by contrast has it’s own power grid totally separate from the rest of the nation’s–and therefore, I would contend, much more survivable to external shocks elsewhere on the continent. Somewhat inefficient in a world of white swans to be sure, but highly effective if survivability is the key. “Inefficient” systems are perhaps the key to survival when to perform “effectively” is to merely survive.

To avoid repeating myself, I’ll just point you at my review of The Black Swan.

I think I remember someone else saying this before, but I wonder, who does Taleb think he’s writing for? If he’s writing for academics who might work out something better than Gaussians (or finally realize that the ubiquity of bell curves is an emergent, not an ontological, property of the world), why the splashy popular book? Am I totally wrong to believe outlets like Newsweek are “consumed” primarily by those who have little influence on either power or policy?

bianca steele/

The sort of people already aware of the “ubiquity of bell-curves in an emergent world”
quit reading Newsweek after High School.

Cranky@45 “Michael Sullivan’s 35 is interesting and worth discussing, although as with all evolutionary biology hypotheses certainly subject to counterargument.”

To be fair, not only is it subject to plenty of counterargument, it is, like most evo-bio or evo-psych talk on the internet, little more than a plausible just-so story. The only evidence behind my story is the well-documented inadequacy of current human brains that it purports to explain. That certainly exists. How it came to be so, I honestly have no idea. My story is merely one, possibly testable hypothesis that represents my internal composite of stories I’ve heard around this particular brain misfeature from various writers I respect.

I haven’t read _TBS_, only bits and pieces of Taleb’s articles and blog posts, but my initial presumption is that he, like others, is thinking of an ancestral environment like I describe, one in which “black swans” aren’t worth devoting resources to analyzing, rather than one in which they do not exist or are dramatically rarer than today. For evolutionary purposes, they are equivalent.

I read Taleb’s other book, fooled by randomness. Nobody seems to have a problem when he is railing against the wall street mob.

The structural issue with the academy is that progression is achieved through novel ideas that are judged to be clever ideas. Those are not necessarily the best or most useful ideas when attempting to model reality. The first paper written on the World Wide Web was rejected because the ideas were not considered novel and the system was too simple. The academy was trying to write papers, not solve an engineering problem.

The moderating force on the academy is the need to teach. Unnecessary complexity is harder to teach than stuff that is simple and elegant.

That said, I really wish that we did not have to watch that idiot from Harvard who almost sunk the world financial system with his ‘Long Term Capital Management’ hedge fund lecturing us all on why the bailout is unnecessary. You can win the Nobel prize and still be a dangerous crank. At the very least I would like to see equal mention given to his connection to LTCM as his Nobel prize. And as Taleb points out, there was nothing Long Term about that particular hedge fund.