The Black Swan – Part 1

August 12th, 2008

This is the second of a five-part series reviewing Nassim Nicholas Taleb’s The Black Swan: The Impact of the Highly Improbable. You will find the review navigation at the bottom of this post.

Part 1: Umberto Eco’s Antilibrary, or How We Seek Validation

In the first part of the book, Taleb introduces us to the major concepts that are used throughout the rest of the book. I will discuss some of these here:

The Opaqueness of History: Taleb describes history as opaque, in that we see only the output (or outcome), and not the inner workings. Consider history to be a black box from which we have experiences (or worse, read others’ account of their experiences). The problem with this is that we have natural biases that cause us to form conclusions about causality (perhaps erroneously) that lead us to believe we understand why something happened the way it did. This is further exacerbated when we try to convey this understanding to others, in which case we simplify further into narratives for ease of transmission of the information. When we form our own conclusions and reduce history to narratives, we necessarily eliminate all of the information that we feel is not relevant. Taleb says that this information is important to get a full understanding of causality and so by eliminating it, we fool ourselves into thinking we understand.

Three reasons history is opaque:

Illusion of Understanding: The world is more complicated and random than we realize, yet we convince ourselves that we understand and ignore that which we think is not relevant.
Retrospective Distortion: We assess events after the fact and through the lens of our biases, filtering out important information.
Overvaluation of Factual Information: People overvalue information and believe that the facts they have witnessed in the past lead to general conclusions for the future.

Platonicity: Taleb uses this word to describe our tendency to focus on pure and well-defined “forms” that are constructs existing only in our minds. We prefer these over less easily understood concepts, like randomness, which we ignore. It is easier to accept “the CEO did X, the company was successful, therefore the CEO must be good at his job” than to recognize the amount of randomness involved in the company’s success. This builds upon Taleb’s Fooled By Randomness, to explain that the common tendency to ignore the role of chance in our lives (as discussed in Fooled by Randomness) has the consequence of our focusing on certain clean (and simple) constructs as to how the world operates. By focusing on these constructs, we think we understand more about the world than we really do.

Tunneling: When we do allow ourselves to consider the unknown, we tend to focus on well-defined sources of uncertainty, such as future sales or economic conditions. This gives us a sense of security in that we have the uncertainty “covered” and makes us more prone to the Black Swans which do not easily come to mind.

Confirmation Fallacy: We use our experiences to draw conclusions that experience is evidence of no possible Black Swan. Instead, we have proven there is no evidence of the possibility of Black Swans. The absence of evidence is not the same as evidence of the absence. We have too much confidence in what we think is right and pay too little attention to how quickly we could be disproven.

Narrative Fallacy: We summarize and simplify, reducing the complexity of situations. We do this so that things are easier to remember and convey to others, and we do this to convince ourselves that we understand the environment in which we live. This distorts our understanding of the world. We can’t help it, but it is important to be aware of it and recognize that the world is not nearly as simple as we believe. Favour experimentation over storytelling.

The Ludic Fallacy: Very closely related to Platonicity is the Ludic Fallacy which is the act of confusing a model (or a platonified concept) for reality. By believing our simple, clean concept of reality to be the way the world works, we ignore all of the things that exist outside of our concept, which includes the Black Swan (because it is unknown), and this leaves us open to the Black Swan’s extreme impact.

As discussed in my review of Roger Lowenstein’s When Genius Failed: The Rise and Fall of Long-Term Capital Management, the Ludic Fallacy was a major factor in the fall of LTCM, as the geniuses that ran that hedge run believed that their models accurately showed the risk associated with their trades, ignoring the Black Swan. They believed their risk assessment models were reality and that nothing would happen outside of their models. They were wrong, and they collapsed as a result.

Taleb argues that statistics only work in games with well-defined rules. In reality, we cannot know the odds with accuracy because we don’t have all the information (since it hasn’t happened yet) and that a small variation in any variable could have a huge impact.

Scalable vs. Non-Scalable: Things which are non-scalable require more effort to yield greater impact. Think of lawyers, dentists, and other people paid by the hour: to earn more, you must work more. Things which are scalable require no extra effort to yield greater impact. Consider the investor, who may earn higher returns through the same amount of time spent considering investments. Scalability creates giants and dwarves. Some will be wildly successful, while others will be terribly unsuccessful. Non-scalability create groups of the mediocre.

Extremistan vs. Mediocristan: This is closely linked to the scalable vs. non-scalable discussion.

Mediocristan is a place where events don’t contribute very much individually – only collectively. Events are non-scalable. Black swan events are impossible as a result. Example: The average height of a large group of people. Any single person will not disproportionately alter the average height, because no one is tall enough to do such a thing (e.g. no one exists who is 21 miles tall)

Extremistan is a place where inequalities are great because one single event can disproportionately impact the aggregate (because events are scalable). Black swan events are likely because any single event can have a disproportionate effect. Example: The average wealth of a large group of people. Any single person may have enough m oney to materially affect the average (e.g. Bill Gates would cause a major change to the average, even in massively large groups).

Socially constructed concepts, like wealth, exist in Extremistan, whereas physical characteristics like height exist in Mediocristan. People make a big mistake in believing they live in Mediocristan, when in reality they exist in Extremistan, where Black Swans exist.

Review Nagivation (Links will work once all posts are published)

Introduction | Part 1 | Part 2 | Part 3 | Part 4