There has to be result oriented money distribution system. Close to what soviets had.
Huge specialized r&d facilities pounding on a very specific task like creating new type of advanced weapon. There also was artificial competition created between them. Good example SU vs MiG jet design bureaus.
No result, no money.
The system was also feeding significant number of theorist do they can catch up. Schools were suffering a bit but looking at totally opposite situation is US maybe it was a good thing. Nobody should spend millions chasing gravitational waves and proving for yet another time that 'Einstein was right' without even remote goal of practical application.
Now we have science which is more like high fashion industry with skinny models walking down the podium where the real closes are steadily moving to the plus size category. Good thing that high fashion industry exists without public funding.
Most of the pains are in the actual r&d labs. Those people oftentimes lack academic credentials and time to go and confront theorists during conferences. if they did, the obstacles are in place to prevent them from that. Teams in NASA have to fudge their own theories to explain sudden acceleration of deep space probes, correction to GPS satellites offsets and finally emdrive.
The answer from mainstream is typically R.Fyenman style: you are certainly wrong but I can't explain you why since it requires super smarts to understand which you obviously don't have