Antifragility sounds exciting, but there's really nothing there. In the context of economics, it simply means relying on competition instead of central planning (by government or by TBTF banks). Outside of economics Taleb has no real examples of antifargility. Engineering is always about robustness, you have 0 examples of engineering systems that are "antifragile".
The idea is that the evolution of engineering systems is antifragile because they benefit from trial and error. Computers keep getting better over time because of how the industry identifies and fixes issues. An obvious (but not the best) example would be the Macbook power connector. A given system cannot be antifragile (unless it learns on its own), but the evolution of technology can be. Contrast with economics, where (he argues) policy makers learn little or nothing after each boom-bust cycle.
Let me play devil's advocate. From what she's saying I'm drawing the conclusion that mental illness correlates with drug use which correlates with violence. The prevalence of mental illness in the population is small, so the proportion of the mentally ill among violent criminals is also small, but if mental illness is a predictor of violence then being more aggressive about institutionalizing the mentally ill would provide society a better return on investment with respect to preventing violence than policies aimed at the general population (such as gun control).
We're not just interested in the proportion of people with mental illness in the population of violent criminals. We are also interested in the proportion of violent criminals in the population of people with a mental illness. We find that very few people with a mental illness are violent. We find that people with mental illness are far more likely t be the victim of violent crime than the perpetrators of violent crime.
