By Karl Arnold Belser
04 September 2013

I just watched a Stanford seminar by Nassim Taleb on anti-fragility. These are my notes and I am making this post to explain my previous statements that central control can cause fragility and that a distributed system can result in robustness (which Taleb calls anti-fragility). 

What does anti-fragile mean? It is not the same as robust, and I did not understand this concept, even after reading Taleb’s book The Black SwanI will in the future be more careful about my use of the word “robust”

ROBUST means that the object or process can tolerate large variations of use without failing or causing a large loss. Consider a simple screwdriver. It can take a lot of abuse by unusual usage and still be functional.

FRAGILE means that for many instances of use the object or process provides a useful or modestly valuable service, and in rare instances the object might be destroyed or cause a huge loss. Consider a fancy tea cup that can be enjoyed on many occasions but if accidentally dropped it becomes a total loss.

 ANTI-FRAGILE means that for many instances of use the object or process requires a modest effort and little reward for this effort, and in rare instances the object or process might result in more objects, a more valuable object or a large gain.  Consider a new business that makes pasta. It requires a lot of work initially to get a small profit but at some point the product becomes viral and everybody wants the product, which in turn results in a large gain.

Taleb gives examples of  robustness,  fragility and anti-fragility in real society, namely in finance.

A ROBUST situation is one in which one has cash or gold whose value is intrinsic. Hence perturbations don’t affect the value much.

A FRAGILE situation is like the American banking system in which the bankers or investment companies take large but hidden risks, pay themselves large bonuses in the several years before a collapse occurs and then gets a government bailout to cover the incredibly large losses. The central action was an inappropriate bailout. 

Taleb points out that more money was lost in the 2008 financial collapse than was made in all of the history of the United States by all banks, and by the way no one has been prosecuted for this fraud that every taxpayer will have to pay for. Since there were so many individuals and companies that were at fault, only bank failure would give those responsible for the disaster the consequences of their actions.

The hallmark of fragility is central control. If a bad result occurs it will likely be an enormous loss or gain. The difficulty is that it is hugely probable that the experiment will fail and result in a loss because there was no small-scale trial and error test to find out what might happen. It is also true that luck will determine the outcome because prediction by human beings is unlikely, even though those predictors will assert they know what is right. The truth is that no one can know what is right in most cases. Hence one needs the experiments to ferret out the truth.

An ANTI-FRAGIL situation is like the many start-up companies in Silicon Valley. Each start-up is an experiment that, if it fails, will result in a small loss, but if it succeeds like for Google, Facebook, Ebay, and Apple will result in huge wealth creation. The new businesses from Silicon Valley have been a major engine keeping the United States competitive.

The hallmark of anti-fragility is distributed control that allows many experiments to occur, some of which will be very successful. The people who create these experiments are usually scientists, engineers or business people, rather than politicians. Scientists, engineers or business people are educated and trained to ferret out the truth about reality. Politicians and lawyers are generally not.

Last updated October 16, 2013
HTML 4.01