I understand "one point safety" to refer to the "detonation of the high explosives by means other than the firing system" as the firing system is designed to fi

I understand "one point safety" to refer to the "detonation of the high explosives by means other than the firing system" as the firing system is designed to fire at the multiple points at once. Can anybody honestly claim that it's completely impossible that the firing system accidentally does the task that it is its main purpose and... fires? Speaking just from the point of the computer engineer, bit flips or even more bit flips actually occur in practice. One specific "if" branch wrongly taken... one single specific bus line true instead of false... etc... Only for the first few bombs, the human had to actually add the last piece of uranium or plutonium for the bomb to have enough radioactive mass inside to make the nuclear explosion physically possible. In all the more modern designs everything needed in already inside of the bomb or the warhead. The claimed safety seems to be "everything's there, but the electronics (or the computer) controls everything." How calming. Regarding one in the million, the people who made the PBS documentary say: For the accident in Damascus, Arkansas, US: "the air force claimed there's one in the million chance that the fallen socket would hit the missile. During our shooting of the recreation we dropped 12 sockets and 6 actually hit the missile": https://www.youtube.com/watch?v=EoxB88rL_2M&t=2m15s https://www.youtube.com/watch?v=EoxB88rL_2M&t=2m15s The bomb designer, writing the paper a few years ago: http://psam12.org/proceedings/paper/paper_497_1.pdf http://psam12.org/proceedings/paper/paper_497_1.pdf "It is recognized that it is very difficult to provide assurance that the < 1E-03 numerical requirements for a safety subsystem have been met, let alone the < 1E-04 or 1E-05 assignments given to elements or components within a safety subsystem. In fact, it is not possible to amass quantitative data that supports such assertions with a high degree of statistical confidence across all relevant environmental conditions. In other words, it is not possible to conceive of all possible abnormal environments, nor can all environments which can be conceived be tested exhaustively in a repeated fashion to generate overwhelming statistical certainty of weapon response."