This book has a really cool innovative style. The first appendix is "a graphical tour of the book" where Taleb graphically explains all the main concepts. It renders the nearly incomprehensible visually explicitly clear. I wish nonfiction writers would use such a graphical appendix. The second appendix focuses on really technical concepts for the quants. That's so Taleb can write the body of the book for the layperson. But, for the mathematicians he is willing to drill down in technical details.
The main point of the book is that the World is really complex and genuinely unpredictable. Black Swans (rare) events will always be Black Swans. Any efforts to forecast such events are counterproductive. But, even though we can't forecast Black Swan events we can manage our exposure to them so they don't hurt us or so we can even benefit from them (antifragility). If we simply remain long the underlying risk by attempting to model Black Swan infested variables, we will be exposed to volatility and fail (fragility).
The main underlying concepts are that the majority of causal relationships are nonlinear. They typically have both a convex section where the curve rises exponentially upward and is associated with a positive effect (antifragile) and a concave section that declines exponentially downward and has a negative effect (fragile). Think of the dose of a prescription drug. At first, as you increase the dose the health benefits improve (convexity). But, beyond a certain dose side effects and toxicity cause harm (concavity). This is shown on the first page of the "graphical tour." The trick is to reduce one's exposure to the concave part of the curve (reduce toxicity, reduce fragility) and increase exposure the the convex part (increase benefit, increase antifragility). And, this is true across all domains. The way to do that is use a barbell strategy that positively captures the optionality of the variable (being long in the convex area and short in the concave area).
Sometimes, this (convex vs concave) metaphor is reversed because a beneficial rising section of a variable's curve can be concave and a declining hurtful section can be convex. And, Taleb does use contradicting examples like that. From a geometric standpoint convexity and concavity do not tell you whether a curve is rising or declining and whether they are associated with positive or negative effect. So, you have to pay attention on a case-by-case basis.
Within this book Taleb offers interesting data insights. The more frequently you look at data, the more noise you get. Assume that you are looking at data with a yearly frequency and that your ratio of noise to signal is 50%/50%. If you look at the same data on a daily basis, the noise to signal ratio will change to 95%/5%. If you look at it on an hourly basis it becomes 99.5% to 0.5%. I don't know how he comes up with those figures (pg. 126). But, they are directionally interesting. Thus, Taleb thinks it is a waste of time to watch the stock market on an hourly or even daily basis. Taleb debunks the merit of Big Data. The more variables you look at the exponentially more spurious correlations you will get (pg. 419). This is a case of a rising convex curve with negative effect. "Modernity provides too many variables... and the spurious relationships grow much, much faster than real information, as noise is convex and information is concave" (pg. 420). Taleb does not vest much in 95% confidence intervals. What matters for him is the consequence when you fall outside the confidence interval. If a plane takes off 95% of the time on time. That's pretty good. If the plane does not crash 99% of the time, that feels like a suicidal mission (1 time out of a 100 you'll be dead). So, the probabilities in absence of their consequences are meaningless (pg 260).
Just as in The Black Swan: Second Edition: The Impact of the Highly Improbable: With a new section: "On Robustness and Fragility" Taleb rejects the entire body of modern finance. In "The Black Swan" he did it by stating that the Normal distribution generates inadequately thin tails and understates the probability of rare events. Now, he adds additional arguments. And, that is that Harry Markowitz comes up with the prerequisite parameters of the Normal distribution namely the standard deviation and the mean. But, ignores an error term in each. "If these parameters need to be estimated, with an error, then the derivations need to be written differently and... we would have no Markowitz paper, no blowups, no modern finance" (pg. 447). Taleb further attacks Markowitz portfolio theory because of its reliance on static correlations between investments. Correlations change all the time, and typically go way up during downturns which eliminates Markowitz diversification benefit just when you need it. Per Taleb, Markowitz portfolio theory causes investors to overallocate to risky asset classes. He further advances that Markowitz does not use his own portfolio theory to manage his own investments. Instead, he uses a simpler but more sophisticated method similar to the one recommended by Mandelbrot and himself (Taleb) (pg. 397). On page 220 and 221, Taleb criticizes Myron Scholes and Robert Merton for getting Nobel prizes for their option formula that others discovered in more sophisticated form before them (but he does not mention who they are). On the same pages, he criticizes Mark Rubinstein for attributing techniques to professors in the 1990s that "we as practicioners used in more sophisticated forms in the 1980s".
Taleb attacks a lot more people than just Markowitz and company. Often his attacks are well grounded; sometimes they are less so. He attacks Thomas Friedman because his influential columns helped cause the Iraq war. He also criticizes his book "The World is flat" for promoting globalization without realizing that globalization increases worldwide systemic risks (pg. 386). Taleb has much intellectual contempt for Paul Krugman because he does not understand the weaknesses in the argument of "comparative advantage" that causes countries to become excessively reliant on the exports of a few commodities (pg. 449). He similarly attacks David Ricardo who came up with this original theory (pg. 212). Taleb roasts Joseph Stiglitz for stating in 2008 that Fannie Mae's probability of failure was effectively zero (Fannie Mae was taken over by the government months shortly after) and for the same Stiglitz to write in 2010 on how he had predicted the 2007-2008 financial crisis (pg. 389). Taleb defines the "Joseph Stiglitz problem"... "Mental cherry-picking, leading to contributing to the cause of a crisis while being convinced of the opposite-and thinking he predicted it" (pg. 432). That's actually Taleb's most convincing attack (the Stiglitz problem).
So, who does Taleb like? Steve Jobs. He is the one person that Taleb adulates. He refers to him at length four times throughout the book always in anthological fashion. He likes Job because he was anti-establishment, without academic credentials, autodidact, visionary, aesthete and artisan in temperament. It is easy to agree with Taleb on those counts.
Taleb is at his best when criticizing modern medicine (chapters 21 and 22). Somehow his convex-concave framework allows him to analyze well where medicine overtreats and overdiagnozes. As mentioned earlier, treatment benefits are nonlinear. It is all in the dosage. Medicine has an intervention bias. Doing something (vs nothing) is almost a requirement for defensive purposes (preventing malpractice suits). Medical interventions are also a response to powerful lucrative economic incentives. This medical mindset has created the medicalization of many normal conditions. For instance, the threshold for hypertension and high cholesterol levels have been chronically reduced so a rising portion of the population can be counted as prospective patients for related prescription drugs. Yet, the benefits of those drugs are convex to the severity of those conditions. This means that patients with near normal conditions will not be helped by those drugs and may be hurt by their long term side effects. It is only for patients with more severe conditions that such drugs may be beneficial. Although, Taleb mentions that reducing markers metrics (cholesterol, high blood pressure) does not always correspond
Taleb has much scorn for any monetary and fiscal policy interventions. Even though Taleb starts from the same place as John Maynard Keynes that the future is unpredictable. They tackle this uncertainty completely differently. Keynes propose expansive government policies (both fiscal and monetary) to shore up economies during downturns. Taleb recommends the government to do very little and for individuals and companies to deal with uncertainty by managing their exposure to it (convex vs concave framework). Taleb acknowledge he has an obsessive stance against any government debt at all (pg. 53) and wishes governments never borrowed and always balanced their Budget (pg. 286). On page 101, he lauds George Cooper's The Origin of Financial Crises who damned the Fed for all the wrong reasons and did not understand that its role is to manage inflation and unemployment rate levels. Taleb also criticizes Bernanke for his "Great Moderation" statement less than a year before the onset of the financial crisis. But, he does not give credit to Bernanke for very quickly changing course and coming up with creative expansive monetary policies to prevent the Great Recession from turning into the Great Depression II. On page 303, Taleb indicates that government interventions almost always end in disaster. He got his timing wrong. Disastrous economic shocks occur first, and government interventions to mitigate those disasters kick in second. Additionally, if governments truly never borrowed and central banks never conducted expansive monetary policies the world's capital creation over the past couple of centuries would be a small fraction of what it is today. Economic growth for centuries prior to this recent modern era was minimal (<0.5% per year). With the advent of modern government policies economies have grown far faster.
Many of Taleb's other arguments are weak.
His entire section on city-states being stronger than larger countries is less than convincing. On page 87, he lauds Switzerland for having a decentralized government with much power vested in its Cantons. But, this government model is not that different from Spain's regional governments set up that has currently turned into a fiscal disaster. The Swiss successes has to do with a lot more than its Canton governments. Later, he advances that Switzerland has been very successful with a very low level of formal education. That's wrong. Switzerland has a high level of formal education with 31.3% of its population having an associate college degree or higher. That is much higher than the OECD average of 27.5%, the European Union average of 24.5%, Germany's 24.3%, and Italy's 13.6% (source OECD 2009).
Along the same line, (pg. 90) he states "thankfully, the European Union is legally protected from overcentralization." That makes little sense since the European Union is a uniquely centralized supranational government by definition. On page 97, Taleb somewhat contradicts himself by stating "until recent history, the central state represented about 5% of the economy compared to about 10 times that share in Modern Europe."
As an example of fiscal austerity, he mentions Sweden (pg. 131) "which responded admirably with a policy of fiscal toughness" in response to a severe recession in the early 1990s. But that is the opposite of what occurred. Sweden undertook unprecedented expansive fiscal policies associated with huge Budget deficits (12% of GDP in 1993, 9% in 1994). And, it has been praised as a model of drastic fiscal intervention to fend off major economic downturns and avoid the nearly three decade long Japanese malaise. Granted, Sweden shored up its fiscal position later much after it had mitigated its recession.
When it comes to investment strategy Taleb struggles to make an implementable recommendation. He recommends investing 90% in very safe assets such as Treasuries and 10% in highly risky assets with massive upside (convex antifragile) (pg. 161). Here, I understand he is talking of Puts on stock indices that are way out of the money. Those would become very lucrative in a market crash. However, he recognizes that such financial options are now very expensive (pg. 175). He reiterates that such options are very expensive like insurance contracts (pg. 183). Therefore, it does seem more like an insurance strategy than a viable long term investment strategy to create wealth. That's because your portfolio will lose purchasing power every single year until a market crash. Let's say if you live in California, it may be good to buy earthquake insurance to protect your home from a related disaster. But, this is not an attractive investment strategy.