I’ve never liked the concept. It just seems to have so little predictive value to say that “Unexpected Things Happen.” The evidence it highlights is interesting wrt the semantics / syntax dichotomy though. Think of semantics as searching in sparse spaces and syntax as searching in dense spaces. In that language one could think of Black Swans as semantics breaking events, events which confuse the formalism and force one to abandon a nice dense space for some unbounded sparse search.
We limit ourselves to certain semantic spaces depending on the problem we are solving, from the inside of which many observations may look representable but are in fact not.