So why did they choose to not use proven designs? Honestly I don't have a problem using riskier designs if it could reduce costs. And increased complexity is a risk in itself.
It's difficult to say, but it's likely to be a factor that an official expressing a risk tolerance >0.00 would have a very short and unglamorous career ahead of him in the hostile climate surrounding anything nuclear, despite us casually accepting much, much higher risks in all sorts of areas.