[Amygaz]

You are OverBallooning what he just said (sorry couldn't resist). There is no implication that the system is absolutely deterministic or will ever be.

Imagine a normal symmetric binomial distribution, along a X-axis of events and severity. Now imagine that the binomial distribution isn't normal and symmetric anymore, it is skewed toward the X-axis of more severe events (i.e. a negative skew). In this model, we still have the same number of events, but their distribution along the severity axis is skewed towards more severe events, and as such they have a higher probability of happening.