Dalio's "by and large" indicates that he feels it's generally true, not universally true. I.e. a statistical trend, not a hard and fast rule. How do your uncited counterexamples show that a statistical trend is not true in any useful sense?

Moreover, why isn't it reasonable to treat it as a hypothesis that is true in many specific contexts? Should we also discuss the personal awfulness of people who believe in normal distributions as a useful principle since not all data is normally distributed? "Racists and child molesters sometimes use normal distributions to model the world!"