it relates to the concept of "probable persistence", wherein the length of time some phenomenon has already existed determines how likely it is to persist for various future timescales
there is a property such that the latter increases non-linearly, proportional to the former
@pee_zombie
Typically hazard functions show decay after a risky settling in period but a relatively smooth curve after that. May be constant hazard, may decay slowly, but there's usually not a discrete drop in the hazard at some arbitrary time far from zero
@pee_zombie
There are exceptions though. For instance if we assume covid takes two weeks to kill you, that'd show up as a bump two weeks after exposure.
some phenomenon posses this property, others do not, & this roughly defines two general classes; the ones which do are referred to as "being Lindy", in reference to some NYC diner (idk the details, something about that diner having been around for an unusually long time)