Can someone explain this to me like I'm seven? My maths just isn't good enough. I've read Pinker's Better Angels and I've read the unsatisfying letters that Taleb wrote in criticism. Taleb seems to have a chip on his shoulder and while he's undoubtedly an incredibly smart man, his thinking is often sloppy when he steps outside his field.
I work as an actuary and are somewhat used to seeing calculations like these. They often point to very large tail risks, for example in catastrophe modelling like windstorm, or in this case violent conflict. What makes me somewhat doubt the applicability of these models is that they often statistically encapsulate what you already know:
Life is fragile (to borrow Talebs word). Weather, environment (think supervulcanoes) and man itself (violent conflict, nuclear events) all have the capacity to do enormous damage and are a threat to society.
I know it's a variation on the age old question "What's the use of #$this_science?", but even Talebs own narrative goes into making systems 'antifragile'. You can only make systems antifragile for so far as the causes of the fragility are understood and manageable. So for an insurer we can do something with a 1-in-200 chance of total ruin: don't insure. A society can take measures to build earthquake-resistant buildings. Perhaps even move away cities from supervolcanoes.
But in case of violent conflict: What can we do to prevent the US - China war? (To name another now trending topic [1]) Or the next black swan conflict?
Just a side remark, I always wonder when reading about Taleb whether he was truly the first to discover the balance between fragility and robustness or anti-fragility. I remember some really inspiring work by John Doyle [1] at Caltech in the early 2000s on the concept of Highly Optimized Tolerance [2] and complexity and robustness [3], stating that natural and engineered networks exhibit a high tolerance towards common perturbations, yet can show catastrophic failures when encountering extremely rare events (as described by power law distributions). Never read much about Nassim Taleb's hypotheses, yet sounds similar at first. Has Taleb popularized a concept that got stuck in the unpopulated wasteland between theoretical physics and control theory?
I think the most interesting here is this line:
"All in all, among the different classes of data (raw and rescaled), we observe that 1) casualties are power law distributed."
If I understand the math correctly, power law distributed distributions arise when there is some cascade effect, for example cities where the biggest cities also grow the fastest. Investment income would be another example - the more money you have, the faster it grows.
That to me suggests that what happens is that small conflicts can grow into larger conflicts, and more importantly the larger the conflict the more likely it is to grow bigger.
I don't know what that says about how we ought to minimize the risk of big wars - except maybe to resolve them peacefully while they are still small conflicts.
Although - now I'm going to contradict myself. Things like skyscraper heights also follow a power law distribution, and those don't grow in any meaningful sense. So I guess I don't know what that means. But if anyone else does, please enlighten me.
