Comment by ggm

The time to certify life-or-death changes to systems exceeds the time to train more people, or pay better to retain existing staff.

"do both" is actually a good answer. The Manhatten project did this. Thermal diffusion and other forms of concentration were initially put head-to-head in competition. It took a while for people to realise both worked, and should be run in parallel or even enriching feedstock. A competitive A or B not both position would not have worked out better.

So yes. research tech replacement, but expect it to be a 15-20 year project with the same costs as other 15-20 year projects. At the same time, don't assume tech will solve social issues, and pay ATC better and increase richness of training programmes by cloning the schools.