Three Types of Swan

Taleb distinguishes between three types of unexpected event:

White Swans result from the ‘Normal’ randomness of a Gaussian Distribution (Bell Curve) circumstance, called Mediocristan by Taleb.  This is the land of the known unknown.  It’s main features are:

• Averages have an empirical reality
• The likelihood of an extreme event reduces at a faster and faster rate as you move away from the average
• This effectively puts an upper and lower limit on the scale of an event (people can only be so tall or so short)
• Outliers are therefore extremely unlikely
• No single event significantly affects the aggregate (you can’t get fat in one day)

Because probabilities are known, the chance or risk of an unexpected event can be calculated precisely, e.g. Weight, height (most anything in nature that grows), coin tossing, IQ scores.

Grey Swans result from circumstances where Power Law (Fractal or Fat Tailed) randomness occurs, called Extremistan by Taleb.  This is the province of the somewhat known unknown. Its main features are:

• Although probability of events follow a pattern this does not give precise predictability (think of forecasting the weather)
• Averages are meaningless because the range of possible events is so large
• The likelihood of an extreme event reduces more and more slowly as you move away from the most frequent event (that’s why they are said to have ‘fat’ or ‘long tails’)
• This means there is no upper limit (Bill Gates could be worth twice, three, four ... times as much)
• Outliers are therefore much more likely than you might think
• One single event can change the whole picture (if Bill Gates is in your sample of wealth, no one else matters)

e.g.    Magnitude of earthquakes, blockbuster books and movies, stock market crashes, distribution of wealth, extinctions (species and companies), population of cities, frequency of words in a text, epidemics, casualties in wars, pages on the WWW.

Black Swans result from all those forms of nonlinear randomness that we do not know about (and possibly never will). We are in the realm of the unknown unknown. We have no means of calculating the likelihood of an event. Everything about Extremistan applies without any probability distributions to guide us. A Black Swan:

• Is an outlier that lies outside the realm of regular expectations (because nothing in the past can convincingly point to its possibility)
• Carries extreme impact (so much so that at the time, if it is a ‘negative’ Black Swan, almost nothing else matters)
• Gets explained after the event, therefore making it look like we could have predicted it, (if only ...).  

This fools us into thinking someone (an ‘expert’) might be able to predict the next Black Swan, which in itself increases the chances of another one catching us unawares in the future.

I see the ideas in Taleb’s book arranging them self into four levels:



LEVEL I — The nature of uncertainty, unpredictability and randomness

Randomness:


The Bell Curve (‘Normal’ or Gaussian distribution):

“We can make good use of the Gaussian approach in variables for which there is a rational reason for the largest not to be too far away from the average. If there is a gravity pulling numbers down, or there are physical limitations preventing very large observations, [or] there are strong forces of equilibrium bringing things back rather rapidly after conditions diverge from equilibrium.
 
Note that I am not telling you that Mediocristan does not allow for some extremes. But it tells you that they are so rare that they do not play a significant role in the total. The effect of such extremes is pitifully small and decreases as your population gets larger.

In Mediocristan, as your sample size increases, the observed average will present itself with less and less dispersion — uncertainty in Mediocristan vanishes. This is the ‘law of large numbers’.” (pp. 236-8, Taleb)
The empirical rule for normal distributions is that:

        68% of the population lie between -1 and +1 standard deviations
        95% of the population lie between -2 and +2 standard deviations
        99.7% of the population lie between -3 and +3 standard deviations
Power Law, Fractal, Fat-Tail distribution:

“Dramatic, large-scale events are [always] far less frequent than small ones. In systems characterized by power-law behaviour, however, they can occur at any time and for no particular reason.  Sometimes, a very small event can have profound consequences, and occasionally a big shock can be contained and be of little import. It is not the power law itself with gives rise to these unexpected features of causality; rather it is the fact that we observe a power law in a system which tells us that causality within it behaves in this way. The conventional way of thinking, which postulates a link between the size of an event and its consequences, is broken.” (p. 170, Ormerod)

Problem of Induction:   


This is not a ‘chicken and egg’ problem, it is a ‘chicken and death’ problem: Everyday of a chicken’s life it has more and more evidence that because humans feed, house and protect it they are benign and look out for its best interests — until the day it gets its neck rung.

Circularity of statistics:


Non-repeatability:   

Black Swan events are always unique and therefore not subject to testing in repeatable experiments.

Complex systems:


LEVEL II — How we get fooled by uncertainty, unpredictability and randomness

Black Swan blindness:


This means that the effects of extreme events are even greater because we think they are more unexpected than they actually are. As an example, we do not distinguish between ‘risk’, which technically, can be calculated, and ‘uncertainty’ which cannot. Taleb maintains that the ‘high’, ‘medium’ and ‘low’ risk options offered to you by your financial adviser are a fiction because they are all based on ‘normal’ distributions that ignore the effect of Black Swans.

Confirmation bias:


Part of the bias comes from our overvaluing what we find that confirms our model and undervaluing or ignoring anything that challenges our model.

Fallacy of silent evidence:


By “the cemetery” Taleb means: The losers of wars don’t get to write history; Only survivors tell their story; Prevented problems are ignored; Only fossils that are found contribute to theories of evolution; Undiscovered places do not get onto the map; Unread books do not form part of our knowledge.

Apparently 80 per cent of epidemiological studies fail to replicate, but most of these studies don’t get published so we think causal relationships are more common than they are.

Fooled by Randomness:


We have a hard time seeing animals as randomly selected, but even when we can accept that, we are even less likely to accept that the design of a car or a computer could be the product of a random process.  Also, with so many people in the financial markets, there will be lots of “spurious winners.” Trying to become a winner in the stock market is hard, because even if you are smart, you are competing with all the spurious winners.

Future Blindness:


Our inability to conceive of a future that contains events that have never happened in the past — just think how difficult it can be to help some people associate into a desired outcome of their ownchoosing!

We can use Robert Dilts’ Jungle Gym model to postulate that Future Blindness is equivalent to Perceptual Position Blindness (e.g. the inability to take into account the mind of someone else — one of the features of autism)  and Logical Level Blindness — the inability to notice the qualitative differences occurring at different levels (e.g. using sub-atomic theory to ‘explain’ human behaviour). Gregory Bateson frequently griped about Logical Level Blindness, and it is a characteristic of people Ken Wilber calls “flatlanders”.

Lottery-ticket fallacy:

The probability of winning the lottery is known precisely and is therefore not a Black Swan — just unlikely.

Locke’s madman:


As examples, both Taleb and Ormerod pick out a couple of Nobel prize-winning economists (Robert Merton Jr. and Myron Scholes) who did much to promote the Modern Portfolio Theory. In the summer of 1998, as a result of having applied their theories, their company, LTCM, could not cope with the Black Swan of the Russian financial crisis and “one of the largest trading losses ever in history took place in almost the blink of an eye ... LTCM went bust and almost took down the entire financial system with it.” (p. 44 & p. 282, Taleb) However that’s not the real story. The more important question is, why ten years after the failure of LTCM, is the Modern Portfolio Theory still the dominant model taught at universities and business schools?

 Ludic fallacy:


Narrative fallacy:


i.e. to explain past events. Note, the plausibility of the story is not a factor. The fallacy is forgetting any narrative is only one of many possible narratives that could be made to fit ‘the facts’ and therefore believing we understand ‘the cause’.

Nonlinearity:


Platonic fallacy:

Has several manifestations:

1. “the focus on those pure, well-defined, and easily discernible objects like triangles” and believing those models apply to “more social notions like friendship or love”;
2. focusing on categories “like friendship and love, at the cost of ignoring those objects of seemingly messier and less tractable structures”; and
3. the belief that “what cannot be Platonized [put into well-defined ‘forms’] and studied does not exist at all, or is not worth considering.” (p. 309, Taleb)

Good examples are the importance we attach to numbers (did you know that 76.27% of all statistics are made up?), and how much better we feel when we have a label for an illness.

Retrospective distortion (Hindsight bias):

“Examining past events without adjusting for the forward passage of time leads to the illusion of posterior predictability.” (p. 310, Taleb)

Commonly known as ‘hindsight’, we think we could have predicted events, if only we known what we know now.  Most of what people look for, they do not find; most of what they find, they did not look for. But hindsight bias means discoveries and inventions appear to be more planned and systematic than they really are. As a result we underestimate serendipity and the accidental.

Reverse-engineering problem:

Round-trip fallacy:

 “The confusion of absence of evidence of Black Swans (or something else) for evidence of absence of Black Swans (or something else).” (p. 310, Taleb) Just because you have never seen a Black Swan doesn’t mean they don’t exist.

Toxic knowledge:


My summary of Level II - How we get fooled:



LEVEL III — Why we don’t seem to learn that we are fooled by uncertainty, unpredictability and randomness

Black Swan ethical problem:


The former are ignored; the latter highly decorated. Would we have rewarded the person who made aircraft manufacturers fit security locks to cockpit doors before 9/11? Hence it pays to be a problem solver.  Similarly we pay forecasters who use complicated mathematics much more than forecasters who say “We don’t know”.  Therefore it pays to forecast.  Both these asymmetries increase the likelihood of negative Black Swans.

Expert problem:

Scandal of prediction:


Common ways we blame others for our poor predictive abilities:

• I was just doing what everyone else does. (The “herd”, or maybe the “nerd”, effect.)
• I was almost right.
• I was playing a different game.
• No one could have forecast that.
• Other than that, it was okay.
• They didn’t do what they should have.
• If only I’d known X.
• But the figures said ...

Epistemic arrogance:


Arrogance is the result of


And I would add to Taleb’s list:

Comfort of certainty:

We’d rather have a story that makes us comfortable than face the discomfort of uncertainty.

Uncertainty quickly becomes existential:

I have observed that, generally, when we consider:

• what we know in relation to what we don’t know;
• our ability to predict in the face of what’s unpredictable;
• how much we can control as opposed to what’s uncontrollable;
• how much effect we can have compared to a Black Swan event;

we are faced with the nature of our individual existence in relation to the Cosmos ... Oh my God!


LEVEL IV —  And how we can learn from how we are fooled by uncertainty, unpredictability and randomness

Maximize serendipity:


Taleb calls this an “Apelles-style strategy”. Apelles the Painter was a Greek who, try as he might, could not depict the foam from a horse’s mouth. In irritation he gave up and threw the sponge he used to clean his brush at the picture. Where the sponge hit, it left a beautiful representation of foam.

Despite the billions of dollars spent on cancer research, the single most valuable cancer drug discovered to date, chemotherapy, was a by-product of mustard gas used in the First World War.

Taleb recommends living in a city and going to parties to increase the possibility of serendipity. [Other examples: speed dating, holding liquidity, write articles, newspapers, sowing seeds]

Use a barbell strategy:


Taleb suggests that rather than putting your money in ‘medium risk’ investments (see Black Swan Blindness) “you need to put a portion, say 85-90 percent, in extremely safe instruments, like Treasury Bills.  The remaining 10-15 percent you put in extremely speculative bets, preferably venture capital-style portfolios. (Have as many of these small bets as you can conceivably have; avoid being blinded by the vividness of one single Black Swan.)”  (p. 205, Taleb)

Taleb lists a number of “modest tricks”.  He says to “note the more modest they are the more effective they will be”:


Accept random causes:

Avoid the silent evidence fallacy, “Whenever your survival is in play, don’t immediately look for causes and effects. The main reason for our survival might simply be inaccessible to us. We think it is smarter to say because than to accept randomness.” (p. 120, Taleb)

By “survival” Taleb means not only your physical survival from, say, disease, but also survival of your company, survival as an employee, survival of your marriage, survival of your success or status, etc.

De-narrate:

“Shut down the television set, minimize time spent reading newspapers, ignore blogs” because the more news you listen to and newspapers you read, the more your view will converge on the general view — and the greater the tendency to underestimate extreme events.

Account for emotions:


Taleb’s point is that our emotional responses will unconsciously dominate unless we take that into account when we make decisions in the nonlinear world of Extremistan.

Avoid tunneling:   


Question the error rate:


Recognize self-delusion:


Also see an article by Penny Tompkins and myself on Self-deception, delusion and denial.

Use Stochastic tinkering:

Use a lot of trial and error.  To do that you have to learn to love to lose, to be wrong, to be mistaken — and keep trialling.

Personally, I prefer the term trial-and-feedback because it presupposes the noticing of ‘errors’ and acting on that awareness.  Wikipedia is a living example of stochastic tinkering.

Focus on Consequences:


Foster cognitive diversity:


My conclusion?

David Grove was himself a Black Swan ...

---

The next Developing Group will continue to focus on Level IV — Maximising Serendipity:The art of recognising and fostering potential.