Antifragile Antifragile 8.4分

All rise, here comes convexity

2018-04-07 06:25:44

This is not a review. Just a note of what I remember after listening to the Audiobook.

Antifragility is the property of objects/systems/ideas that is opposite to fragility. It is often mistaken as robustness or resilience. But this misunderstanding is missing the majority of its key nature. The fragile get harmed by random events, and has to be heavily protected. The robust is indifferent to random events, therefore it doesn't matter whether it is mishandled. The antifragile, benefits from chaos and randomness. The closest analogy is Hydra. When you cut off one of its heads, two heads grow out.

Sophisticated systems often display certain level of antifragility within a reasonable range. For example, human bodies react to stress to muscle by overcompensating and grows even stronger muscle fabric. Although the antifragility on the system level will sometimes come at the expense of some individuals (what doesn't kill you kills others). A stresser is helpful to the system in the sense that weak and outdated individuals can be filtered out, leaving only the strong individuals. Excessive protection from any kind of stresser is actually harmful to the system in the long run. This is especially reflected in economic policies, where interventionist politicians and academics have tried and failed time and time again to wipe out economic cycles. In Taleb's opinion, the government should nationalize the banks and de-regulate the hedge funds.

Another widespread harm caused by acamedia is that they create the illusion that practical knowledge are produced on paper inside universities and research institutes instead of from trial and errors by practitioners. Birds don't have to learn aerodynamics in order to fly. Option traders don't have to study mathematics to price options. Instead, they both learn from trial and error and actually achieve sophistication much deeper than academics come up with. In fact, the heuristics traders developed accounted for fat tail phenomenon instead of the formal approach, which relies on normality assumption and eventually led to several blow-ups since the black swan events are drastically underestimated. Buildings were built by architects who knew little about geometry and couldn't even do division throughout the history. And in fact, these buildings tend to last much longer than the ones depending on modern calculations, which led to corner cutting (the author said in an interview, 'an economist would probably take away one kidney and one lung from human body, if they were to design it.'). There are surely areas where technology comes from development in formal science, but this is the not the norm, but rather the exception. The narrative that technology is pushed forward only by advancement in centrally sponsored research labs is a narrative pushed by the historians (the author calls this 'history is written by losers').

Antifragility is present in many fields. Antifragility is especially reflected in the the mathematical concept of convexity. If a function is convex, tail events will lead to larger values. Also, a linear combination of two function values will be greater than the function value evaluated at a point between the two points in the previous setting (also known as Jansen's inequality). This also happens to be a strategy adopted by many individuals without any knowledge of it. For example, many women would choose to marry an accountant while cheating with a rock star. The accountant will guarantee a stable bottomline, while the rock star gives a large potential gain (having fun, better genes, etc). Similarly, an investment portfolio of 90% riskfree assets and 10% high risk high return asset tends to perform much better than 100% medium-risk asset, especially since the risk accounting can be problematic and the risk profile might be much worse than people's perception. Fragility, on the other hand, would correspond to concavity. An easy way to test fragility is to check the acceleration of loss with respect to key variables in the model. Say we have a city whose traffic has a capacity of holding 10000 cars at the same time. Suppose 1000 extra cars would lead to a 10 minutes slow-down in the traffic, and another 1000 extra cars lead to a 30 minutes slow-down. We see an acceleration of the loss. Therefore the traffic system is fragile, in the sense if the prediction for peak hour number of cars is off, then the slowdown could easiily blowup.

The core principle of antifragility applied to investment requires us to firstly admit the existence and unpredictability of black swan events. Black swan events have significant impact, yet the probability of black swans are almost always underestimated. An investor has to first limit the downsize exposure to blackswan events. Just as the velocity of a plane doesn't matter if its safety isn't guaranteed. An investor can however benefit from blackswans since there are financial instruments such as options, where there is limited downside, unlimited upside. And since the probability of tail events are always underestimated, these instruments tend to be undervalued. This applies even more in fields outside investment. Since there is no explicit pricing process, people's bias against impactful tail probabilities will not face much effort for correction.

Picking up very similar argument as in the black swan, Taleb emphasized the fragility of predictions and systems relying on accurate predictions. People are not only bad at making predictions. They are actually bad at seeing the center issue to be predicted. However, people are very good at reverse rationalization. Therefore there are people like Stiglitz who gave horrible predictions such as GSEs have 0 risk and years later sell books as the person who foresaw the 08 financial crisis. Academics and policy makers are not in the proper environment to make moral decisions since they have no skin in the game. As a patient, asking the doctor what you should do might get you a very different answer from asking him what he would do in your circumstance. The fact that academics and politicians can make decisions that decide many people's fate without being accountable is the source of countless problems. In fact, Taleb's most recent book skin in the game is largely about putting agents' stake in the game and change the incentives for them.








免费下载 iOS / Android 版客户端