Book review: Antifragile by Nassim Nicholas Taleb

A black swan: just one of these disproves the saying that 'all swans are white'. Creative Commons: Castelli
A black swan: just one of these disproves the hypothesis that ‘all swans are white’. Copyright Castielli, reproduced under Creative Commons

A few years ago I read The Black Swan by Nassim Nicholas Taleb (NNT). I found it, appropriately enough, randomly lying around at home. I remember it being a difficult, technical read. Off the top of my head, if I had to summarise it I would do so like this:

Most of the world’s most significant events are Black Swans – extremely rare, often unforeseeable events that change the world for better or worse. Examples in my lifetime would be the September 11 terrorist attacks, the invention of Facebook and the financial crisis of the late 2000s. Individuals and society should prepare for Black Swans as best they can by becoming…

And that brings us to Antifragile. Written after The Black Swan, it continues Taleb’s line of reasoning. He argues in favour of systems being robust, or even antifragile.

What exactly is antifragility?

“Antifragility is beyond fragility or robustness. The resilient resists shocks and stays the same; the antifragile gets better.”

Taleb points out that in English, and apparently in other languages, ‘fragile’ lacks a word with a precise opposite meaning. ‘Robust’, ‘resilient’, ‘solid’, ‘unbreakable’ and so on aren’t strictly antonyms because they don’t gain from external shocks. If you drop a glass and a ball from a height of two metres the glass will smash and the ball will bounce. The glass is critically weakened by the impact. The ball shrugs off the impact but doesn’t get any stronger from it either. Something antifragile would actually emerge stronger from having been dropped.

The way to become antifragile, according to NNT, is to expose yourself to lots of small risks and learn from them. The opposite (fragile) approach is to place too much faith in new, untested methods and to run large projects where the risks are great and the payoffs frequent and small. This leads him to criticise much of the way Western society is designed, including things as varied as the financial system, modernistic architecture and processed fruit juices.

One of his ethical rules is to have ‘skin in the game’. Taleb argues that we shouldn’t tolerate people who aren’t personally exposed to the risks they discuss. He gets to the heart of what angered people so much about the bank bailouts: the sense that bankers were getting a free option. They kept the upside (high salaries and bonuses during the good years) and everyone else got the downside (the bill for bailing some of them out).

Antifragility and journalism

NNT reserves particular contempt for journalists. He doesn’t read newspapers. He criticises journalists for being too focused on the anecdotal:

“But the media only report the most anecdotal and sensational cases (hurricanes, freak accidents, small plane crashes) giving us a more and more distorted map of real risks”

At least one study has shown that the British public do dramatically under- or overestimate the rates of politically contentious subjects.

benefit spending

An Ipsos MORI poll in 2013 found that respondents estimated that 24 per cent of Britons were Muslim. The true figure was five per cent in England and Wales. Similarly, the public massively overestimated Government spending on foreign aid.

In data journalism we do try to counteract these misconceptions. This story by my boss David Ottewell in the Mirror shows that when you include pensions – which accounts for more than half of Britain’s benefit expenditure – Britain’s ‘benefits capital’ shifts from inner cities to the leafy Isle of Wight.

Data doesn’t escape Taleb’s scorn either. He makes the point that Big Data – the measurement of lots of different variables – can throw up random correlations, obscuring the signal in noise. Nate Silver wrote an entire book on this subject, appropriately titled The Signal and the Noise, which I’d also highly recommend. He is the guy who correctly predicted the result of the 2012 US presidential election in all 50 states and now edits FiveThirtyEight, one of the most influential websites in data journalism.

For example, there was much talk of President Obama struggling to win the 2012 election if he could not get unemployment below 7.2%. No president had been re-elected since Franklin D. Roosevelt (FDR) with an unemployment rate over 7.2%. In the end, he triumphed with unemployment at 8.1%. Those who put faith in this statistic forgot that there hadn’t been many elections since FDR to build up a reliable dataset and that the unemployment rate isn’t the only thing American voters think about when deciding who to vote for president.


To sum up: Antifragile was well worth reading. NNT provides a framework to live in a world we don’t understand, accepting randomness – something I’ve struggled with in the past.

Read more: Antifragile on Amazon