If the backup generators had been put on a simple platform, the plants would never have melted down.

I suspect that's not the only SPOF in play.

Nukes, as with rockets, are complex. Many things have to go right and keep going right for them to operate as intended.

(See Rickover's "Paper Reactors" paper for a similar take on this.)

Another factor that contributed not only to Fukushima but other inundations following the Tohoku earthquake was that in response to the megathrust-fault slip, the east coast of Japan sank about a meter lower.[1] Engineers building seawalls failed to take this into account, as I understand.

I'd first realised that such quakes could lower land on the continental plate of megathrust faults when looking over satellite imagery of Banda Aceh after the 2004 Indonesian earthquake and tsunami. Regions along the coast were flooded long after the tsunami had passed, which I realised must have come from a lowering of the land itself.

Seattle and the Pacific Northwest may face similar issue should their Big One occur eventually.

________________________________

Notes:

1. "https://www.sciencedirect.com/science/article/pii/S003808061..." <https://www.sciencedirect.com/science/article/pii/S003808061...>

There are a number of cheap-as-dirt things that could have been done to prevent the disaster. Putting the generators on a platform was one. Venting the hydrogen outside instead of inside, another. Having emergency repair parts on site is a third. Having a gravity-fed way to cool the reactors with external water, a fourth.

And so on.

Every programming problem boils down to a binary choice of having a 1 or 0 in a given memory / binary code address.

Pile up enough binary choices, each individually cheap as dirt, and you've got a large problem.

Even dirt in quantity gets expensive to manage: dams (earthen), mines, levees, landslides, etc.

Nukes are complex with variance conditions that are extreme to the extreme.