Is it easier to start a research lab when you are competing for a small pool of researchers with a bunch of already established labs, or when there are 3 underemployed PhDs at every coffee shop in town?
If a world war lasts 4 years, and a typical PhD takes 5 years, which country can more easily spin up research: the one with an excess number of PhDs at the start, or the one which waits for the outbreak of war to fund more PhDs who will graduate 1 year after the war is done?
Yes, some resources are spent on training a PhD who doesn't use it, but there is also a cost to not having enough PhDs available when you need them. If the latter is greater than the former, excess PhDs are optimal.
Not sure how useful the PhD who's been serving coffee for 10 years will be to the next Manhattan project.
Perhaps we should initiate some moonshot projects or something to keep the skills sharp. Low pay but you get to do (perhaps self-directed) work on the coolest problems. You would need an army just to process the applications I believe.
If you only produce just enough PhDs, then when you start the next Manhattan project you need to scrape the bottom of the barrel and grab someone who's been out of the field for 10 years, or wait several years for the rate of PhDs being completed to increase.
If you produce a substantial excess, on the other hand, you can instantly scale up to meet increased demand. It's all about that pipeline.
But yes, moonshot projects can also be utilized so that past PhDs can be kept sharp. Indeed, a huge number of science programs currently do exactly that - society may see little direct benefit to discovering methane in the space around Uranus, but the people optimizing those radio telescopes are pretty nice to have around if you think you'll ever need people to optimize radar systems.
