Since I started interviewing for faculty jobs, I had to seriously ponder on how I would run my potential future lab. One question in particular has been bothering me quite a lot. How many postdocs (if any) should I hire? This is what I would like to discuss here today. I will be deliberately provocative, but also probably quite naïve. So please, do share your thoughts and contradict me if you think I am wrong.
There are three principal dimensions to this question. The first dimension is of course the money: if you have money, you hire people; if you don’t, you can’t. Real decisions are made along the other dimensions.
The second dimension is productivity. Research labor comes primarily in two flavors: PhD students and postdocs. (I will discuss permanent researchers below; let’s focus on students and postdocs first.) Postdocs are more productive and more independent, but are also necessarily subject to less of principal investigator’s (PI) intellectual influence. Graduate students are less productive and less independent, but are subject to more influence and can be substantially cheaper. PhDs come in batches once a year and they are a pig in a poke: some may turn out to be brilliant, others not so much. Postdocs can be hired essentially at any time and are much more predictable. As a pragmatic PI, I would certainly hire more postdocs, if money were available. After all, a postdoc is the fastest, surest and the most cost-effective way to convert grant money into papers. So, is there any reason NOT to hire postdocs?
Unfortunately, there is the third dimension to this question. Is it moral to do so? The majority of postdocs (at least in my and closely related fields of biology) are people who themselves strive to become PIs after they leave their postdoctoral lab. So, when they join the lab, they implicitly or explicitly assume that the main payoff for their work will be their future faculty job. (Most postdocs certainly don’t work for their immediate compensation, which is rather small.) Here is the problem: Their assumption is wrong.
Let’s look at some data.
This is a figure from this paper published last year. There are two observations here. First, it shows that the number of annually awarded PhDs today is about 10 times higher than the number of annually opened faculty positions. In other words, the chance for freshly minted PhD to end up as a professor is about 10%. The infographic below made by a fellow Harvard postdoc Jessica Polka illustrates the same point very nicely. And here is another estimate that puts this chance at about 6%. These pieces of data should already be enough to conclude that that for a freshly minted PhD investing time into an postdoc is quite similar to “investing” money into gambling.
The second observation is that over the past 30 years the total number faculty positions stayed almost constant. This is very interesting if we think of academia as a reproducing population : one generation of PIs “produces” the next generation of PIs. In population genetics people often model such reproducing population with the so-called Wright-Fisher model. One key feature of this model is that the total number of individuals in the population stays constant (for example because the available resources are limited). Every generation all individuals attempt to reproduce, but because the total population size cannot grow, each individual on average leaves only one surviving offspring. And this is almost exactly what happens in academia. The size of academia stays roughly constant, which means that the average PI will “produce” over their entire career only a single postdoc or graduate student who would become a PI him/herself.
Importantly, the distribution of offspring numbers in academia is far from being narrowly concentrated around 1. The majority of PIs do not leave any academic descendants at all, while a few especially fit PIs produce dozens (my PhD advisor Simon Levin is a spectacularly fit PI, for example). Leaving the causes for this variation aside, let’s consider a reasonably successful PI whose fitness is 5, which is well above the average and certainly something to strive for. If this PI plans to have, say, a 30-year career and to take, say, 1 PhD student and 1 postdoc per year on average, that’s 60 “attempted reproductions”. So, how likely is this PI’s next postdoc to be lucky enough to get a job? That’s about 8% chance. So, even for a postdoc who joins the lab of this quite successful PI, the a priori chances to achieve their goal of becoming a PI themselves are rather slim.
When I think about my potential future lab in this light, I can’t help but see my potential postdocs as people walking into a casino. And this would make me, the PI, very much like the owner of that casino, i.e., an institutions that in most cases collects the reward and sends the person home minus their cash. So, is this morally acceptable?
I have heard four types of answers to this question.
Answer #1. There is no problem. There are actually plenty of jobs out there. Good people always get jobs.
Answer #2. There is no problem. Postdocs are adults, they well know it’s a gamble and choose to do it anyway. Ultimately, it’s not my problem what happens to them after they leave my lab.
Answer #3. Yes, it’s a moral problem. But it’s a tough world where everyone has to fend for themselves. I need to make my career, so let the chips fall where they may.
Answer #4. Yes, it’s a moral problem, and I am trying to deal with it.
Although, I did not expect to hear answer number 1, it may be not so surprising in retrospect. It mostly comes from very successful PIs that (a) are superstars who have not had problems getting a job and (b) have succeeded, one way or another, in pushing their postdocs into jobs. So, from their perspective, the system works fine, and people who fail, fail for a reason. This answer also shows that this moral problem is substantially more severe for junior faculty whose ability to deliver on implicitly promised jobs for their postdocs is generally much weaker than that of senior big-name faculty.
Answers number 2 and 3 appear to be most common. Both of them amount to the same action, or rather lack of action, and support the current imbalanced state. The only difference is that the person in position 3 feels guilty whereas the person in position 2 does not. Either way, postdocs get hired and mostly fail to get faculty jobs themselves. The main problem with this position is how to look the prospective postdoc in the eye and convince them to work for you, despite your inability to land them into a job. This problem is similar to an apparent problem of how one would convince a person to go into the casino and gamble away their money? Naively, it seems that nobody should ever do this, knowing the odds. But a typical gambler does not know the odds! Likewise, most newly minted PhDs do not know their chances of getting a faculty job (info here and here, see also the infographic). And 90% of those who do know the odds honestly think that they belong to the successful 10%. So, the PI rarely even needs convincing, prospective postdocs come knocking on doors themselves. But for a PI who knows the odds, taking a naive postdoc is, in my view, immoral and amounts to little more than exploitation: the hired postdoc produces a benefit to the PI mostly for the promise of a future job that will likely not materialize.
Most people in category 4 that I met are themselves postdocs or very junior PIs just starting their own labs. So, what are the options for such people? I know of 4 possible options, maybe there are more.
1. Hire postdocs jointly with big-name senior PIs. This arrangement doesn’t really solve the main problem, but at least relieves the junior PI of the moral burden of overpromising to their prospective postdocs. With a big-name senior PI in the back, the chances for that postdoc to get a faculty job are somewhat higher.
2. Hire only brilliant postdocs who have a priori high chances to get a faculty job. This seems to be a reasonable but cruel strategy which favors people who are already on the “A train” and discriminates against people who were late to figure out what they want to do, even if they would have turned out to be brilliant given the chance. Moreover, it’s hard to recruit the A-trainers to a junior PI’s lab, especially if it is not in a top school.
3. Do not hire postdocs at all, run lab with graduate students only. This seems quite reasonable, although at the expense of productivity. One might object to this: Isn’t there the same problem with PhDs as with postdocs? In my view, the problem is not the same. I believe that entering a PhD program in natural sciences is not a commitment to an academic track, whereas entering a postdoc is, in most cases. Most jobs outside of academia do not require a postdoc experience, so a postdoc definitely narrows down one’s options. In contrast, a PhD generally widens the options. So, in my view, most PhDs should not go onto the academic track. But in general having more educated people in the non-academic world is good, especially given how many people do not believe in evolution or what idiots oversee science in Congress. A more detailed discussion of this subject is a topic for another day.
4. Hire permanent researchers instead of postdocs. This I think is closer to a fundamental resolution of the problem. Rather than hiring a short-term postdoc by dangling a future faculty job in front of them, it is far more fair to hire a researcher permanently with a salary and benefits adequate to their experience. Although the current funding system is not particularly suitable for this – obviously, permanent researchers should be paid by the university not by grants – it can be done. A permanent researcher also becomes a great asset for the lab as they accumulate valuable skills.
To conclude this long post, I would like to leave you, the reader, with some questions. If you are a graduate student, ask yourself whether you really want to enter the academic track. Unless you are an A-trainer, are you willing to gamble on 3+ years of your life? If you are a postdoc aspiring to be a PI, how would you run your lab so that it is fair to people you hire? If you are a PI, do your postdocs work for the promise of a future faculty job? If so, do you think you are being fair to them? And given an estimate of your academic fitness from past experience, how many postdocs should you hire in the future to maintain fairness?
 I thank my colleague Wolfram Möbius for drawing my attention to this analogy