Of late, I have noted, among a subset of scientists, a disposition that the quality of the PhD candidate and/or PhD training declined.
Of course, this is based on purely subjective measures – usually centered on the degree of independence and engagement.
I posit that there is, always has been, and always will be a distribution in these areas, and that perceptions of decline are either: (a) just that, perceptions colored by general negative feelings about the academic system today, or (b) a result of the increased volume of PhD candidates and recipients (i.e. the raw numbers have increased, but the distribution remains the same).
But maybe I’m just an optimist …
What say you?
Ever on & on 
The number of candidates has grown, but I wonder if the growth is different from that of the general population? Depending on their relative rates, there should be more of the very smartest around too.
The rosy retrospection bias might also be part of it. I notice myself getting more prone to this as I get older!
It is cane shaking about kids these days.
I posit that the mentorship vacuum extends up and down the chain and across lab types, but there’s a non-trivial number of oases of normalcy (or semi-normalcy), though not enough to shift the mean if it could actually be measured. The lack of mentorship and mis-mentorship takes a multitude of forms as do trainee rationalizations (as you point out), which contributes to the perception gap.
I do think the “relative numbers have increased and therefore the number of the ‘low tier’ will increase too” is showing a lot. It’s also that the pool of PhD students is larger and more diverse than “back in the good old days” when only the small group could get admitted into the lab and then be groomed by their mentor. I mean after all, how is this “quality” defined? ^^