I don't know, I think you have to consider the nature of the system you are dealing with and its design life. I did an audit for a now-decommissioned research reactor and looked at their operational records. It never failed to SCRAM when called upon. Never. Not. One. Time. Ever. Granted, it is a simple system, but that is the key to its reliability. Turn off the current, and gravity effects the shutdown. Gravity doesn't fail. Would the system have ever failed? I don't know, I guess, maybe if they ran it long enough so things were crumbling into dust. Certainly everything has that kind of endpoint failure. But for the practical lifetime of a system, I think it is possible to have essentially a perfect safety record. LWR technology, at least from the viewpoint of public safety, comes as close to that as anything we have out there.
Ah, but there’s the rub, you have qualified a time frame and thus have to say that there is a statistical probability of failure in a certain period of time... MTBF.
Even Gravity will eventually fail but not in a time frame we should be concerned with. An asteroid may hit, the Sun will Nova, the magnetic field will shift (it already has more than once) but will any of these things happen in a time frame important to us now? Probably not.
You said:
Gravity doesn’t fail. Would the system have ever failed? I don’t know, I guess, maybe if they ran it long enough so things were crumbling into dust. Certainly everything has that kind of endpoint failure. But for the practical lifetime of a system