This paper (pdf) from the National Bureau of Economic Research, titled "Does Science Advance One Funeral at a Time?", and summarized here, got a fair amount of publicity when it came out about a year ago. Aside from the fact it supports everything DM has to say (uh, see here but also see here for other issues of entitlement), there is additional value to understanding the study. I think that the analysis of the results suggests another course of action for PI's.
So what does the NBER paper say? It is a data-driven analysis of publishing rates in life scientists. They looked at folks who died in the saddle at the height of their prowess, what they called super-stars, but I think of as Big Dogs. They measured changes in publication rates of Big Dog collaborators vs. non-collaborators of over the time period that included the death of the big dog. They found that publication rates of the collaborators dropped, and of the non-collaborators increased. They also implied that the papers of the non-collaborators are "disproportionately likely to be highly cited".
From their conclusions:
Overall, these results suggest that outsiders are reluctant to challenge leadership within a field when the star is alive and that a number of barriers may constrain entry even after she is gone.
[note: love the use of "she" in the quote]. So what is your average, run o' the mill PI to make of this, beyond confirming what we have always suspected?
I think it suggests a strategy with respect to trainees. I think it suggests moving them on as soon as you can. Now if you are a Big Dog, the data implies they do better with you and co-authoring. But! You might die! You might get hit by a bus, get a glioblastoma, or pancreatic cancer or all of these. What then? Your poor little trainees will be in terrible shape, and suffer more than the amount that they will from just missing you. If you care about their success (an if, I know), they will do better the sooner they are on their own.
But what, I hear you say, what if you are not a big dog? Ah. I am not a big dog, although I am old. I suppose if I was a serious scientist, if I was a good scientist, a serious scientist, really smart and really good, I would be a big dog by now. But alas, alack, I am not. Any of those things. So, for me, small dog that I am, what does this tell me? That my trainees will not necessarily benefit from my co-authorship. But, and this is an extrapolation, they will be poised to benefit more from the death of the big dogs if they are out there on their own.
Actually, this paper suggests nothing of this kind at all. It does not look at the children of little dogs, nor does it separate the children of little dogs from the older little dogs themselves. And actually, when push comes to shove, I am not so much a little dog as a meerkat or an ermine living in someone else's burrow. But that's an aside.
This is a thought about trainees, in general. In general, I think that trainees benefit from being pushed out of the nest. From being pushed to think of their own stuff and develop their own programs. Of course! It is so damn tempting to keep that incredible student or postdoc. They do well. They make your lab not only run, they make your lab HUMMM. But that is about *you* oh mammal of intermediate size. Think about them. Think about what is best for them. And getting them independent, getting them on their own, finding some new trainees for yourself: everyone will benefit. Whether your trainees will be able to challenge the dominant paradigm, publish more or less, and whether in the end the rate of publishing, beyond a threshold really matters, are all things we have and will continue to debate for the near and distant futures. But meantime, independence is the goal of our training, and we are making more type II errors (not seeing independence when it exists) than type I (seeing false independence when it doesn't).