Yes, that's very important but it is also important to remember that this was in a particular context and that the "pick a design and just build it" approach on

Yes, that's very important but it is also important to remember that this was in a particular context and that the "pick a design and just build it" approach only works when you have a design (however imperfect) that is going to work. First, the context. Rickover needed to get a reactor designed, series built, and operated right away. In that context, he was absolutely right. Second, the PWR design works and it works very well for submarines. I don't know that there has been a better design concept for submarines (the USSR built a few liquid metal cooled reactors which caused them lots of problems). When you have a basic design that does work, then going ahead and building it is the right thing to do. We can see the consequences of what that approach does to earlier stage research programmes as well because some of Rickover's people were put in charge of Argonne's reactor research programme which they did not understand and where they caused an immense amount of trouble by trying to duplicate this approach in a context where the basic approach had not been definitively decided. This lead, for instance, to spending an enormous amount of money and time pursuing an oxide fuel fast reactor concept despite being warned that there were fundamental problems with that design and that spending more time up front on metal fuel design work would be a better idea. They didn't want to hear it - oxide fuel works (which it does - in thermal PWR and BWR reactors) - so go with that and just build it. Rickover was right of course about the principle of understanding technological predictions in the context of what we would now call Technological Readiness Levels and optimism bias but in understanding this we also have to keep in mind the other side of the coin which is not to entirely commit to premature optimisation.