Ecologists will be very familiar with the idea of r Vs. K selection, coined by MacArthur and Wilson in the 1970's to describe reproductive strategies. In a nutshell, r selection is the production of many offspring without investing a huge amount in any one. K selection, on the other hand, is a heavy investment of resources in relatively few offspring.
A K selection strategy is one that is optimal in a predictable and stable environment, where offspring will be given a leg up by a heavy investment in, say, egg yolk or lengthy parental care. An unpredictable environment selects for an r strategy, where many offspring will perish, but the few that land in the right environment will have the opportunity to thrive.
Obviously these are two ends of a spectrum with a lot of gray in between, but the point I'm trying to illustrate is the strategic decisions junior PIs have to make to achieve some level of survival in the grant game. Whereas it does no good to spam agencies with poorly thought out proposals, I would also argue that repeatedly banging one's head against the walls of NSF or NIH with a single proposal or idea is just as flawed. Both may result in funding (with the latter probably more likely), but both could just as easily result in nothing.
My advice to junior people is to get more than one idea in the system. Yes, you'll have you favorite proposal, but you need to be floating more than that at all times. Two is better. Three is even better. You will get the best value if they can go to different directorates/panels/ICs/agencies. Diversify your portfolio so you are not completely dependent on one source of funding now, and for your career. Being too far on the K selection side of the slider leaves you incredibly vulnerable to changes in your particular corner and just the general stochasticity that comes with single digit funding rates. There have been times when a K strategy was viable and you will still hear that advice from some senior people, but that time is not now, nor will it be for some time.