#note/sourcereview/book | #source/book📚/ingested ## What is the thesis? Things that have never happened before happen all the time. ![[Pasted image 20240615070641.jpg]] ![[Don’t be a turkey .jpg]] ## Am I convinced and why? Yes: - Empirical evidence in the world from COVID-19, Black Friday, 9/11, from the assassination of Arch Duke Ferdinand to the invasion of Ukraine, Brexit, etc. shows that events in complex systems are unpredictable. They sometimes, actually often, add up to be "rogue waves." - [[Aggregation theory]] is tightly linked, and his story of Yevgenia Krasenova in Chapter 2 is a perfect example. "Beware the scalable" names the next section. [[Aggregation theory]] is the "speculator and the prostitute." Prostitutes don't scale, there is a limit to the services they can provide. Aggregation theory shows scalable processes (ie digital) are black swan generators. - The key error is believing the problem set is "Mediocristan" when it is actually "Extremisitan." - "The Problem of Inductive Knowledge" is the idea that the transition from specific observations to generalizations is limited in that it is vulnerable to black swans. It is not just that (in the turkey case in the image above) the inductive knowledge that the farmer is friendly based on 1000 observations of a farmer coming to feed the turkey is unhelpful. The [[Mental Model]] created is the _opposite_ of reality, it is net harmful to the reality the turkey faces. The turkey becomes blinded by what they think they know, what they have learned, because they become more confident that the farmer is friendly as the days grow. The turkey lacks [[epistemic humility]], and thus becomes vulnerable to risk because of their growing "expertness" whereas the reality is that they are developing a more significant [[Blind Spot]]. The riskiest day is when it seems to be the safest. - The rest of the book is really difficult to follow with a multi-pronged several chapter rant against the use of Gaussian analysis in financial analysis. He spends quite a bit of time arguing that financial analysis is more of a fractal "Mandelbrotian" sequence with patterns that are self-similar but not limited by Gaussian curves. Naturally occurring distributions of things like people's height follow (roughly) Gaussian curves with SD distributions that exponentially become less likely (a 9 foot tall person is one in billions) but financial and digital worlds are not limited by nature and multi-SD outliers are actually common comparatively. ## What is the other side of the argument? - This is one lens to use to look at the world, but there are so many other useful lenses. While some systems likely do not follow Gaussian distributions, others may, and there may be times when each model is useful. ## What else do I wonder about? - I'm quite unsure how to apply this book to my life, with the exception of one idea. As a reader, the rant was hard to follow, it lacked structure. Other books of this type tend to be more formal in their argumentative structure which makes them easier to follow. If I ever wrote anything like this, I would try to have general readers give feedback on structure and analysis. ## When do I want to stumble across this? #on/risk| #on/perception | #on/uncertainty | #on/systems ## Source: Taleb, N. N. (2007). _The black swan: The impact of the highly improbable_. Random House.