The increase in short-lived labs
I was reading up on an unfamiliar field today, and many of the references came from a young, female PI at a nearby university.
Hooray! I thought, maybe I could over there and talk to her in person.
But no. Her lab is gone. She did not get tenure, despite publishing quite a few high-impact papers, and Google has her at any number of companies in any number of locations over the last few years.
It would be interesting to find out the average life-span of the average lab now vs. 20 years ago. My latest (some will probably say ridiculous) hypothesis is that losing these people as resources after they've reached the PI stage is worse than losing them, say, after grad school, because by the time they've had their own lab for a few years, they've published more.
Bad enough, as has happened to me quite often, when famous older scientists who published landmark papers are already dead by the time you think of the ultimate question you'd love to discuss with them (along the lines of, if you could invite anyone living or dead to dinner, who would it be).
I've found that interviewing former academicians is often painful and rarely productive, since these people usually don't remember, and would prefer not to bother trying to remember, much less discuss, the details of their past published work. This is also true, unfortunately, for much older papers of much older academicians, even ones who still have functioning labs.
Unfortunately the scientific method doesn't work unless published papers can truly stand on their own, but that is rarely the case.