The roughly right approach to ‘robustness’ is to recognize that many significant events in history could have happened in lots of different ways. If Einstein hadn’t discovered relativity, others were really close. If Attila hadn’t sacked Rome, someone else would have.

I suspect this is the premise of psychohistory in Asimov’s Foundation series, and the reason the character of the ‘Mule’ had to be a mutant to break the robustness of Hari Seldon’s history. Sometimes those old, bad-writing sci-fi dinosaurs get at some things more clearly than Gibson and his progeny

There is a subtle but difficult problem here though, which becomes starkly clear when you try to define the problem mathematically. When you attempt to model a history path/possible world as the trajectory of some giant equation and define robustness the usual way (in terms of sensitivities and closeness of different paths and initial conditions.… this is the calculus of variations) you realize you have to decide a priori what are the essential (necessary) vs. peripheral bits of any causal relationship between A and B. Locally it is easy [Any of 100 scientists with Einstein’s level of competence between 1899 — 1920] –> [relativity would have been discovered] kinda stuff.

Oh well. Back to work.