> "Collapse" isn't within the statistical distribution though

Uh? Maybe you could explain what you mean by this a bit more.

1. It's not a rigorously defined term.

2. "System collapse" would be unexplored territory, so how would statistical analysis be able to infer when it occurs?