Sunday 18 August 2013

Nassim Nicholas Taleb's Antifragile in 1000 Words (not 500 Pages)

What is the opposite of ‘fragile’? Most of us would naturally give the answer ‘robust’ or ‘resilient.’ Nassim Nicholas Taleb, however, points out that this is incorrect. An object is fragile if it suffers negative consequences from volatility or stress, he reasons. The direct opposite, then, is not a lack of consequences, as robustness implies, but rather positive consequences from volatility or stress. 

Thus Taleb introduces the concept of antifragility, the central idea to which everything else in his book is tied back. He argues that antifragility is a very general concept, but one that has been pretty much overlooked in science and philosophy up until…his book. He provides examples of things-that-are-antifragile in a wide range of domains. In physiology, muscles strengthen after being stressed; in finance options become more valuable when markets are volatile; in the media an attempt to supress information leads to its wider dissemination (the Streisand effect). Evolutionary processes lead to antifragility at the system level as environmental volatility kills off weaker members of the population.

Once you get beyond the enticing prologue, the rest of Taleb’s tome is a rather painful read. It contains many interesting ideas but they are wrapped into an exposition that seems designed to alienate. Taleb expresses a disdain for most other human beings, and comes across as an arrogant know-it-all. To give an example, even though Taleb lauds the efforts of entrepreneurs in taking risks and pushing forward technological progress, he gives obnoxiously unflattering descriptions of the few technology practitioners he has actually met:

“Technothinkers tend to have an ‘engineering mind’ – to put it less politely, they have autistic tendencies. While they don’t usually wear ties, these types tend, of course, to exhibit all the textbook characteristics of nerdiness – mostly lack of charm, interest in objects instead of persons, causing them to neglect their looks. They love precision at the expense of applicability. And they typically share an absence of literary culture.” [p.314]

As an ardent fan of Taleb’s earlier books, “Fooled by Randomness” and “The Black Swan” I made the effort of seeing this one through to the end1. As such, I shall try to summarize three of its core concepts, which in addition to the notion of antifragility, make up the meat of the book.

1.) Non-linear Effects
The human mind deals quite easily with linear effects. Double an input and, as a first approximation, we tend to expect output to roughly double. If you drive twice as fast you get to the destination in half the time. If you put in 10% more effort you expect a result 10% better, and perhaps to end up 10% more tired afterwards. 

We find non-linear effects less intuitive. For example if you drive twice as fast you use up four times as much energy, so your fuel costs will be four times higher. And in some circumstances, being 10% better than average at something can earn a payoff 100x the average. In reality, non-linear effects abound.

In Taleb’s earlier books he looks at our poor understanding of probability, and so in this book he combines this with the concept of non-linear effects. Fragility, for example is a non-linear response to a shock, where small shocks have no effect but large shocks have a catastrophic effect. This changes our interpretation of variance2: an increase in variance makes large shocks, and thus catastrophic impacts, more likely. A high variance regime is therefore qualitatively different from a low one. We observed this in the financial markets in 2007-08 when the above-normal volatility brought us close to financial Armageddon.

2.) Causal Opacity
Uncertainty is one of Taleb’s favourite topics, and in this book he focuses a lot on the uncertainty that surrounds causal relationships. He is extremely sceptical of theories about causal mechanisms. Many systems in the world are highly complex, so our attempts to rationalize them are often futile. It is wiser to be honest about our ignorance and focus instead on empirical regularities and phenomena than abstract theories.

A key field in which this philosophy has practical implications is medicine. Taleb believes we should rely on medical intervention a lot less than we presently do. His key insight here is that the negative side-effects of interventions are potentially large, and intrinsically opaque. Some side effects take decades, or generations, to manifest. The risks inherent in any medical treatment are impossible to quantify, so the wise approach is to be conservative and only intervene for very serious conditions.

Taleb further points out that doctors and pharmaceutical firms have skewed incentives in favour of providing treatments. Doctors want to be seen to be doing something – and Pharma companies want to make a profit. Taleb urges readers to be resilient to minor problems, letting the immune system do its job; after all, the immune system is the result of millennia of evolution which has given us a degree of antifragility.

3.) The Ethics of Transfers of Fragility
It is clear throughout the book the Taleb harbours a visceral anger towards the Western elite, firstly for promoting a culture of competitive materialism, and secondly for fixing the system so the dominance of the elite is perpetuated.

Taleb identifies that transfers of fragility (and antifragility) are a key method by which those with power prosper. A transfer of fragility means we make ourselves antifragile (i.e. susceptible to large gains) by making others more fragile (i.e. vulnerable to large losses). Many examples of this exist in the asymmetries of the financial system. For example fund managers take a large proportion of their fund’s profits but do not pay up if the fund make a loss. A bond trader who makes a large profit gets a bonus, but the worst-case if he or she makes a large loss is to get fired, while the loss is borne by the bank. Even the banking system as a whole has an asymmetric incentive to take risks as it can be bailed out by the government if it loses.

The more fundamental insight is that while we tend to notice first order monetary transfers (e.g. taxes and welfare payments), we also need to think probabilistically about such transfers. Any conditional cash transfer can been viewed as a probability distribution, so we must consider the higher order moments, for example the variance and skewness, of these distributions when we consider the ethics of such transfers. A transfer that seems equitable and ethical if we look at the ‘expected’ payoff may be quite biased once we take the whole probability distribution into account.

Taleb is clearly an intelligent guy, so I can only conclude that the abrasive presentation in much of his book is a result of his disdain for copy editors and an attempt to cause controversy. I could easily have written an entire blog post about the things I dislike about the book. However it does contain many ideas of value, of which I hope I have given you a flavour here. 

1  If you do decide to pick up a copy of Antifragile, my suggested reading plan would be the prologue and Book 1, then skip directly to Books 5, 6 and 7.

2 And not just variance: the interpretations of the other higher moments change, with skewness and kurtosis also important.