Accident one (Chernobyl): The plant manager disabled multiple safety overrides (including the one that would provide power to drop the rods if the turbines went into emergency shutdown), then ran above rated capacity to see how far he could push it. This caused the turbines to stop generating power. Without backup power, rod control shut down.

I argue this is not the fault of the plant designer or technology. With enough effort, almost anything can be turned into a bomb.

Accident 2 (Fukushima): Plant designers ignored tsunami warning signs in the area, and also put the backup generators underground in a costal area. This was a screw up in the risk model's input.

Ignoring that, 250 * 50 = 12,500, so the prediction is within an order of magnitude.

The choice is between running 6 plants for a year or two with those odds, or actively funding ongoing genocide. This isn't a particularly difficult instance of the trolley problem.

Judging by your numbers, is if we had 12,500 plants in the world we will have a disaster every year.

Also considering human error, I don't see more plants equaling less human errors, needing more workers my drop the quality of workers. No one wants Homer Simpson running their local plant.

We know how to build plants that are physically incapable of melting down. The numbers only work looking backwards at this point.