It's the Black Swan Event problem, inherent in human nature.
The answer is to always be a little bit more conservative than you think necessary.
Absolutely true.
Part of the problem here is that scientists’ ability to generate numbers for the likelihood of various risks is actually quite low.
For example: Let’s say you life in Florida. We all know that Florida is hit by hurricanes from time to time.
But how many of them will make landfall with winds over 150 MPH?
Don’t know.
What is the likelihood that one, just one, will make landfall in *your* area of Florida with wind speeds over 150MPH? We can agree that there is a non-zero probability of there being a 150+MPH hurricane where your house sits in Florida.
What is the likelihood that such a hurricane will hit Florida where you live *while you’re alive and living there*?
Now suddenly the numbers get real fuzzy again.
OK, so let’s say you want to design a house that will survive those winds, just in case. Engineers can do that. We’ll just pour a foundation a foot thick, walls 18” thick, filled with the proper amounts of rebar, keeping rooms relatively small, 6” of ferroconcrete for the roof and we’ll keep it to a single story, with steel pilings that go 50’ into the ground. Doors will have seals on them to prevent water intrusion, and will use retractable bolts to keep from being blown out in the wind. Windows, the few that there are, will have armored plate shutters that close in an lock over the windows.
That oughta do it, yea?
Oh, you can’t afford that? And the wife is screaming about the lack of decor and windows? Something about not wanting to use a powder hammer to nail up pictures of the grandkids? Why? She said she wanted a house that would survive a hurricane!
These are the sorts of problems that we run into when trying to design for ‘black swan’ events. People just don’t and won’t pony up for them.
Everyone wants to go to heaven, no one wants to die first.
Like, for instance, put the reactor pumps up high in the building instead of down low, as they apparently did, just to show proper respect for Murphy’s Law(”Anything that can go wrong, will go wrong.”). Any good engineer has a healthy respect for Murphy’s Law. It’s usually the constraints of time and money and general lack of support from the customer that prevent the engineers from designing a system as fail-safe as they would like.