How do you build mean relationships, like
here? As I understand, the logic is "structure is distrubuted like X, we are looking for its mean. If experiment fails with P=q, the average size is incremented by 1. Alternatively, with P=p, (average) structure size is incremented by 1+E(X), so E(X) = q * 1+ p (1 + E(X)). It seems to give right answers but not quite intuitive to me.