As a general observation, *anything* which measures safety as “analyzed by deaths per unit of electricity generated” is a worry – it is a normative expectation of inevitable fatality. The technical solution described is interesting and certainly shifts the hazards and juggles the entropy around a little but nuclear waste is an intractable problem.
Risk analysis must balance dozens if not hundreds (or thousands, millions) of dependent, independent or interdependent vectors and variables. While measuring hazards in regards to “worst case scenarios” seems overly paranoid, my biggest concern is that attention fatigue, software (or hardware failure) and variously “cutting corners” to optimise profits all become inevitable saliences of risk if and when large-scale operations commence.
There is absolutely no doubt that alternative energy solutions are an immediate Global obligation. What logical methods will we use to ensure safety in these systems? Being that cybersecurity (for one) serially demonstrates the endemic and implicit failures of deployed logical control systems, how can we ever assert safety at a level appropriate to the non-trivially complex hazards of nuclear energy generation technologies?