More, more, more

One common feature that seems to characterize many prominent scientists in the prime of their careers, is that they direct large research groups. Loners are rare.
OK, first of all, what does “large” mean, in such a context ?
Well, it varies across disciplines. For someone like me, a condensed matter physicist engaged in theoretical research, a large group is one that includes more than three graduate students and two postdocs — and three graduate students and two postdocs is one which I could call already fairly large.
I have seen theory groups that are (much) larger than that, though, with a number of graduate students hovering around ten, maybe five or six postdocs of different seniority, a few undergraduate students, maybe one or two visiting scientists, and possibly even a secretary and a technician (e.g., a computer system administrator).

A group of that size can make for an exciting and stimulating environment in which to work and grow scientifically; obviously, it takes unusual resources to assemble and run it, resources to which the vast majority of us have no access.
I am not just talking about the money to pay all those salaries, which is obviously quite significant already. The sheer infrastructure needed, in terms of space, furniture, equipment, plus travel and other operating expenses, renders the cost prohibitive to any scientist not enjoying an unusually high level of funding. Not surprisingly, there are only few, lucky individuals worldwide for whom that is a viable proposition, and (again not surprisingly) they tend to be worldwide renowned scientists, who have made seminal contributions to their own fields.
There tends to be a positive feedback too, in that researchers who have greater resources can build a larger group, as a result of which their scientific productivity is enhanced; in turn, that positions them favourably to attract even more funding, with which they can build an even larger group…

In some fields of science, the need for raw manpower (e.g., just about any experimental research effort) justifies putting together sizeable teams, without which the work could not be carried out. Still, one can meaningfully ask: is it really true that, “the bigger the better”, all the time ? More importantly, should an aspiring young scientist always want to be part of something bigger ?
In my experience, there is such a thing as too large a group. There is an “optimal” size (which varies depending on the discipline), beyond which productivity (and more importantly the breadth of the science pursued, and the significance of the results achieved) no longer grows proportionally to the monetary investment, and even the effectiveness as a training and educational venue decreases.
That is one of the reasons I support the general policy of funding agencies such as the NSF or the NSERC, to distribute monies more uniformly, even if it means supporting individual research efforts that are small in size [0].

What’s your problem, man ?
Why would a large group ever be a problem ? Does one not only stand to gain from the presence of many bright colleagues, from whom to learn and with whom to collaborate on bolder and more exciting research projects than one may be able to carry to fruition alone ? And what about exposure ? Is it not clearly better to be at a place “on the map”, rather than being isolated, with only a person to talk to (the Principal Investigator) ? Well, in general yes. But there are also the following potential problems of which someone having to make a choice should be aware (at least based on my personal experience and observation):

1. Limited interaction with the reference person. As I mentioned above, in order to run a large operation one needs to secure the appropriate resources, which in turn means to compete for prestigious funding awards. In order to do that, a Principal Investigator (PI) must spend a huge amount of time on the road, on the phone, and in the office writing grant proposals, which obviously comes at the expense of time spent doing actual research. If on the one hand it may not be necessarily a bad thing for a graduate student or a postdoc to be left alone, free to pursue her own ideas, it can also mean to be stranded, or even worse being de facto supervised primarily or exclusively by a collaborator of the PI, a visitor, another postdoc — even a senior graduate student. Now, that in and of itself is also not necessarily a bad thing [1], but it is typically not what one had in mind by initially choosing to go and work at that place, or with that famous PI.

2. One of the many. In a large group, there are almost always several researchers of comparable seniority, with similar skills and expertise. It may not be easy to compare them based on CVs and publication lists, which are often almost indistinguishable. The PI is the person who, at some point, will be called upon to identify, for hiring or other purposes, a junior member of the group that is especially competent, brilliant, creative. Now, because the PI is human, and therefore fallible, a bad call is always possible. However, the chances that a bad call will be made are greatly enhanced if the PI, for the reasons mentioned above, spends little time in the lab and does not entertain regular, frequent, individual scientific discussions with her postdocs or graduate students.
What tends to happen, is that the PI becomes increasingly detached from the daily grind, prefers to be briefed only sporadically on the status of a project (say, once a month, in a group meeting), and even the smartest person may not be in the best position to tell who is actually doing the work, who is generating ideas, who is leading the effort, who is providing the most important technical contribution. It is easy for a young scientist, especially one that does not have a particularly assertive personality, to be left in the dark, when working within a large group. On the other hand, a PI directing a small group, with enough time to collaborate with and observe more closely her advisees, is in my opinion more likely to speak accurately and fairly about each one of them.

3. My project, dammit. One would think that, precisely because the group is large, members would enjoy relative freedom. After all, what difference does it make if one out of N persons (with N much greater than one) branches off and works on her own projects ? Well, yes and no. What I have seen happening, especially in Europe, is that group members are often expected to be disciplined “team players”, contributing (essentially full time) to the one, ambitious project or initiative generating the massive funding that in turn makes it possible for the group to exist to begin with. I have seen cases of rigid “division of labor” implemented among group members, with unspoken but very real rules (e.g., for article authorship) and tacitly implied roles, reminiscent of those which one would see in a division in the corporate sector.

4. “White collar” graduate student and postdoc. I am personally not in the least fond of the above-mentioned “corporate-like” approach to research and training, but, hey, it may just be my problem — after all, I am no big shot, so, what do I know ? However, it is my sense that many graduate students and postdocs who find themselves working in such a setting, may not have fully realized that that is what they were signing up for, when they made their decision.
The one problem that I see with running a “scholarly enterprise”, is that it has the effect of blurring the differences among the members of the group, whose individual, original contributions end up buried in the greater effort of the team as a whole. Eventually, however, all of these people will be competing for the same position(s). How is one to carve a niche, make a name for oneself, render oneself appealing to a potential employer, who will be presumably looking for someone “sticking out of the pack”, if one is not given the opportunity to pursue, at least in part, one’s own individual research ?
Here too, my personal experience is that there are tangible benefits deriving from operating within a smaller group, interacting with a PI who may not be running equally grandiose an effort, but has more time, is more personally involved in the research, and with whom an advisee can enjoy a more regular, deeper and meaningful interaction. I think it makes it easier to strike a proper balance between original work, and contribution to an existing research effort.


[0] As an aside, this is also one of the reasons I am wary of adopting a nationwide policy (advocated by some) whereby graduate students and postdocs would be supported by transferring funds directly to them, in turn allowing them to “shop around” for a suitable investigator willing to serve as their mentoring host. One of the (several) undesirable consequences would be, in my view, that of funnelling all the most promising young talents to a handful of prestigious places, where they would be operating within large groups and not necessarily find the optimal environment to express their potential.

[1] But it often is, because of the hierarchy that is implied. Seldom does a postdoc appreciate having to report to another postdoc, for example.

Tags: , , ,

5 Responses to “More, more, more”

  1. GMP Says:

    Very nice post! I generally agree with you that after a certain size the productivity of the group does not necessarily go up with increasing group size: there may be a negative differential productivity (NDP) feature in a graph presenting papers per year versus group size, where the number of papers at some point drops with increasing group size because a threshold for poor supervision has been crossed. I would say you may want to increase the group size to no larger than the size at which the NDP feature occurs. 😉

    I have a colleague (an experimentalist) who at a time had 25+ students and maybe 2 postdocs, and the group didn’t get more than 4 or 5 papers per year (they were so-so, not that you could say they were holding out for high impact stuff). The issue was that most of the students were completely unsupervised and were wasting time and money. The group imploded when the funding didn’t get continued, 2/3 of the students just got sacked. It’s really sad — all the time and money spent and nothing.

    No doubt some (very organized and successful!) people are able to be crazy productive with large groups. But you hit the nail on the head with your insight that in such an entreprise the success of the PI and the group far outweigh the success (present or future) of any student or postdoc.

    As for an optimal group size, it takes a bit to determine one’s own. It depends on one’s workstyle and environment (my tenure home is an engineering dept, so postdocs are not ubiquitous). For me, I am a bit on the high side right now in terms of personnel (1 postdoc, 8 grad students, 1 undergrad), and, since I want to meet each one of them every week individually and also have a weekly group meeting, this is a lot of meeting time… But is absolutely necessary to keep everyone productive. I had a time where all I had was 3 or 4 students and that was really not enough, but on the upside I still had the time to do the nitty-gritty techical work on my own (e.g. publish single author papers). I can’t do that with this group size, just the bandwidth require to handle the group’s output (not to mention raising funds!) is considerable.

  2. Cath@VWXYNot? Says:

    I’ve always been in what you would define as large groups (although I think group sizes are larger on average in biological sciences than in your field), and I agree totally with your first and second points. However, the severity of those disadvantages has varied wildly between my various labs, depending mostly on the personality and supervision style of the PI.

    I haven’t been aware of your third and fourth points being a major problem in any of my positions, although I agree that those are very real dangers.

  3. transientreporter Says:

    At the institution where I did my postdoc, there was a Big Shot (Biomedical) Scientist who had forty postdocs in his lab. That’s not an exaggeration. And that’s not including ancillary personnel (technicians, visiting scholars etc). His “lab” took up a whole building and had their own clerical staff. I’m not sure in what sense they were a “lab” and in what sense they were a “research institute” in their own right.

    How much individual attention did each postdoc get with Big Shot PI? The answer is zero. They communicated with him via email only (he was off-campus most of the time, busy running his biotech company).

  4. S. Pelech - Kinexus Says:

    Massimo Boninsegni is very correct about training in very large lab groups. In addition to the points raised above, lab politics can be very problematic as some individuals will perform very well, but many will do poorly. In these sink or swim lab groups, the principal investigator is just too busy to attend to the needs of each trainee. Jealousies and perceived favoritism can create a poisonous atmosphere that is counter productive.

    At the same time, however, if the lab group is too small and the research program underfunded, the training experience can also be a disaster. This is especially challenging for trainees working with new investigators. It is extremely important to do one’s homework when considering where to pursue one’s graduate research training, which can take half a decade or more.

  5. Schlupp Says:

    I could see arguments in favor of joining a large groups in North America, after all the boss of a large group there has a high chance of being really good and/or important. Like you, I happen to think that there are a few disadvantages as well, but at least, I see some point.

    The issue that keeps amazing me is why entire academic systems are set up to have few large hierarchical groups as opposed to more smaller ones, even though I do not often hear arguments why large groups are better, not even in Europe. Usually, the only arguments are (a) it has always been like this, (b) why should it change? and (c) oh, but it is very nice to have a big group. The only more decent attempt was that a large group might favor risk taking, because the person in charge does not have much to lose if something fails. First, I’d like to see at least some attempt at data before I believe this effect, because the the most hierarchical setups do *not* appear to be the most creative ones. And second, this is a first-rate argument against joining a large group.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: