This is the sort of situation that arguably cannot be a single point of failure; because allowing it to be one is itself (at least) one other failure to keep it company.
It is certainly quite possible that the proximate cause was ‘dumb human error’; but the sheer inevitability if periodic instances of dumb human error, and the fact that it isn’t exactly a secret, means that anyone who wants to build a robust system of nontrivial size has to treat it as a fact rather than an excuse and design with it in mind.
Indeed, that’s sort of what the bigger-picture-stuff types are supposed to be doing to earn their premium over button pushers: implement systems(in the broad sense encompassing both tech bits, procedures, and organizational structure) that are more reliable than their parts, rather than merely as reliable as their weak links or tending to produce exciting failure cascades that make them more brittle and/or capable of dramatic error synergy between multiple erring parts.
If you can get hollowed out because one guy forgot a patch, you are certainly no better than ‘as reliable as weakest link’; and very likely in full horrific resonance Cascade of failure territory.
