Superlabs and the Expendability of Grad Students

April 19th, 2012

Commenter Eugene has been leaving comments on ChemBark since before it was ChemBark. Last month, he posted some interesting thoughts in the thread about professors who pose for publicity shots doing lab work. I have copied his comment here in its entirety:

Once your group gets to a certain size, you do not really have to worry about the cost of failure anymore. The successes more than make up for it. It’s not a bad thing, it’s just the way it is and not enough chemistry professors who want to be big shots but are at middling chemical departments realize it.

What I mean is someone like Robert Langer takes credit for the successful grad students and the succesful companies, and there are just so many of them, that the ones who fail are forgotten about in light of the success. The better you can manage everything, the bigger your empire will be. Plus it helps to be in a university that attracts students who are very motivated.

If you’re in charge of a small group, then the failure of a graduate student to get any papers will be magnified since you don’t have too many offsetting grad students that are publication machines. In fact, you might have none. I’ve seen a few profs with medium-sized groups that were taking off, but were blinded by their own hubris, forgot that they were not in a top ten school, and decided to ‘fire’ two or three students or postdocs who weren’t performing well enough (not getting a Jackass or Andjewandte or not coming in on the weekend). Not only does this cause potential students in a non top ten school avoid you (since they care more about lifestyle and not monastic scientific pursuit), but it magnifies the failures of your remaining students (should they happen) as your own failures. Automatically, you’re now an average scientist for the rest of your career and not Robert Langer. Not that there is anything wrong with that, as you can do very good science and pay more attention to individual students, but that’s not what some of the more ambitious types wanted before they cannibilized their group. The lesson is get big as fast as possible with as many driven types as possible and then just sit back and manage your success. You have to get rid of someone who creates a bad group dynamic or is lazy or just plain stupid of course, but once it starts being every second person who works for you, then you’re not doing it right (unless you’re in a really crappy department). Once you’re going, you can move to better and better departments and become bigger and bigger and have to worry about failure even less.

The same effect used to work in pharma. If you’re the CEO of a start-up, when your drug fails, your failure is huge. You’re out of business. If you’re the CEO of a big pharma company and a few candidates fail Phase II, it’s no biggie because you’re got good phase III data on that one cholesterol lowering candidate that will pay the cost of failure for the others.

That’s some biting analysis, filled with all of the cynicism and bitterness one expects of a recent product of graduate school in chemistry. Of course, none of the analysis had to do with the topic of that post—pictures of professors working in labs—so just to make sure Eugene’s thoughts didn’t get lost in the shuffle of the other thread, I wanted to unpack them here.

While it’s more than a bit cynical, the point that Eugene makes about large groups’ affording their PIs protection from a degree of failure is absolutely correct. PIs are judged by the magnitude and number of their accomplishments, not by unspectacular failures. In all of the farkakte metrics used to judge research productivity (e.g., paper count, h-index, total citations), I have never seen someone divide the metrics of accomplishment by the number of graduate students and postdocs required to achieve them. The situation is not unique to chemistry. Reggie Jackson is remembered as one of baseball’s greatest hitters, despite the fact that he holds the dubious honor of being the MLB’s all-time leader in career strikeouts. But people don’t remember the 2,597 strikeouts; they remember the handful of home runs he hit in the clutch.

Getting back to chemistry, let me start by stating that I think it’s great when professors establish themselves to the point that they have the flexibility to fail. It is a deserved product of success, and everyone should be so fortunate in their jobs. Getting to this point probably also helps take the heat off of students in the lab. And when students “fail” (or flounder or whatever), I don’t think you can blame the PI solely (or even primarily). A lot of factors influence these outcomes, ranging from things that can be controlled (intelligence, lab skills, motivation, work ethic, design of experiments) to those that can’t (bad luck, family issues, medical issues).

What worries me about “superlabs” is that by their design, one or more students is probably destined to be ignored not because of his or her personal failings in lab, but because the number of available professor-hours in a year is limited. Eugene is making the point that professors in superlabs need only pay attention to a fraction of their students because their accomplishments will be enough to support the machinery of the superlab and ensure its continued existence. In this model, a professor actively makes a decision to ignore some students. That idea is extraordinarily cynical, but I am not at all sure it can be ruled out.

My view is that one needn’t adopt as cynical a view to show what is effectively the same outcome. That is, I will assume that professors want to pay attention to all of their students, but sometimes, they just can’t. I assume that professors sleep, eat, commute, attend to personal hygiene, spend time with their families, watch television, and do other “normal” stuff we expect of human beings. These activities require time, and together, probably account for more than half of the day. Even if you are going to spend the entire balance of time performing “work”, there is a lot of work to get done. You’ve got to teach, prepare for class, hold office hours, write grants, write reports, serve on committees, go to departmental meetings, meet with speakers, referee papers, keep up with the literature, travel to conferences, give talks, and write letters of recommendation. I am sure I have missed some things, but that is already a lot of stuff and it doesn’t even begin to address advising students—or ancillary work such as consulting or running start-ups. It is an absolute miracle that people can do all of these things and run groups of 10+ students and postdocs, let alone 30+ or 40+. In these cases, it’s almost inevitable that some students will fall by the wayside.

In an era where we have a surplus of freshly minted chemists and a dearth of jobs, I think our field should consider whether it wants to encourage the model of superlabs run by single professors. I personally like the idea of incorporating senior investigators into these research groups to serve as “minibosses” that can provide hands-on expertise and advising. In defense of Langer’s superlab, which might be the biggest in all of chemistry, it is my understanding that he does employ a group of senior scientists as lieutenants to oversee his various subgroups. But, by and large, I don’t think this idea is very common in chemistry. I think a lot of people feel the money for one senior scientist is better spent on multiple students and postdocs, and the delegation of authority is something many professors loathe to accept. While every lab is different, as time goes on, professors’ increasingly jammed schedules are bound to take a toll on advising. Graduate schools must watch out for their students, because many overextended professors will not.


16 Responses to “Superlabs and the Expendability of Grad Students”

  1. anon Says:

    small groups good, big groups bad. increase the number and diversity of ideas and you increase the number of advances. a higher number of small groups will do this better than one big group. Professors with large groups are selfish and hindering the advance of science and the pursuit of new knowledge.

  2. See Arr Oh Says:

    @Paul – Regarding your last paragraph (the lieutenants / minibosses), doesn’t this closely match the way chemistry research is already done in many other countries?

    Off the top of my head, Japan, the Netherlands, Germany, Russia, India, and several other countries use some sort of “intermediate” professorship – whether they call it a Habilitant, Research Prof, Group Leader, or other title – to achieve goals under a larger umbrella professorship. Even in the US, the larger groups (>30) tend to have more admins, lab managers, coordinators, and sometimes even graphical designers (lookin’ at you, KCN) to help students navigate the waters.

  3. Matt Says:

    @anon I don’t know that large groups are diversity killers. Just look at Paul’s graduate advisor and you will find no shortness of “whacky ideas” 😉 Also, some professors don’t want large groups. It is not in their interest to have a large group and so they don’t. Why limit or tell a PI what they can do especially when they prove that they are capable of it? (We talk a lot about poor graduate education with respect to job training/prospects. But the one thing that I KNEW when I started grad school is that if I joined a particular group I certainly would not receive any help if my project failed in ANY way.)
    Now, large groups do tend to receive more funding. And that is for the reasons that Eugene suggests. You’ve got enough students to eke out enough results for that next R01 … even if some projects are failing. In small groups, if some projects are failing, its likely that all projects are failing. But, that doesn’t necessarily mean that the ideas are bad. Unfortunately the funding situation has been such that if you don’t already have 2/3 of the work finished, you’re not getting the grant. I think that this is the process that needs to be changed to make sure that all PIs are getting some money to help their projects along.

  4. Bunsen Honeydew Says:

    Superlabs can be either good or bad, depending on the PI. When the PI is engaged as much as possible in the individuals and their research projects, it really isn’t an issue. In fact, an assembly of talented students and postdocs can greatly enhance the educational aspects and problem solving potential of the group. There will still be “failures,” but more often than not, it will be attributable to the individual researcher, not the PI.

    Conversely, there are some PIs who have figured out how to exploit the system to grow superlabs where one cannot readily justify the size/output ratio. Certainly, those PIs have and have had significant scientific (and funding) success; however, if they may parlay that success into the ability to grow beyond a output-justifiable size. For example, using a disproportionate amount of student support mechanisms (e.g. TAs) simply because they can attract an seemingly unlimited supply of students. This can allow a group to grow beyond what the research funding levels would dictate. These superlabs may exist at the expense of the health of a department as a whole. Anecdotal evidence suggests this is more prevalent at 2nd tier and lower institutions, where a higher profile researchers may have disproportionate political power compared to their colleagues.

    Once a group grows, there are significant disincentives to contract. Maintaining lab space, TA positions, etc. may motivate a PI to allow poor or marginal students to continue in a PhD program even after it becomes apparent the student lacks the necessary aptitude. Moreover, this can even lead to degrees being conferred to undeserving individuals. It is probably still a minority of the current unemployed chemists, but I suspect these marginal degree holders are often the ones who have trouble finding post-graduate work.

    Another anecdotal observation is that some students are attracted to superlabs for the wrong reasons. Rather than interest in the research, some prospective students will correlate obtaining a PhD from a superlab with desirable employment options. Future prospects are important, but ultimately that’s a hard way to maintain motivation for 4+ years of grad school. A more tenuous suspicion is that some marginal students are actually attracted to superlabs because they observe group members having a more comfortable graduate school experience than students in smaller groups, who are subject to more scrutiny from their advisor (while still obtaining the same degree). If this is the case, it perpetuates a decrease in average productivity of the superlab and the production of sub-par PhD recipients.

  5. Wavefunction Says:

    I think having Research Professor positions (UNC Chapel Hill has some for instance) can help in alleviating the lack of contact one might feel in a superlab. In the UNC biochemistry department where I worked, many labs had one research professor, typically someone who had about 5+ years of post-PhD experience. Most of these people were quite knowledgeable and pretty good in partially substituting for the PI. The position of course lasts only until the funding lasts but I can’t imagine that being a problem in a lab like Langer’s.

    Ultimately it depends on the individual PI. A very prominent superlab PI on the West Coast who I am familiar with is known to meet up at least twice a week with every member of his large group, and he is incredibly familiar with every individual’s project. I don’t know how available Langer is but a friend who did his postdoc in his group met up with him twice during his entire stint (I am not saying that my friend had an unproductive experience there, just that these are contrasting styles). Most of my friends who worked in superlabs have done it for the prestige and the job prospects, but I am not sure they have really been mentored in a satisfactory way.

    One thing seems clear; to run a superlab really well takes something of a superhuman.

  6. Special Guest Lecturer Says:

    I find Eugene’s opinion to be a dramatic oversimplification. Graduate education is not dominated by individual PI – graduate student interactions, even in moderate-sized groups. Performing outstanding science is learned by immersion – designing, performing, and interpreting experiments, interacting with peers and more senior classmates and postdocs (by far the most important in my opinion), attending seminars and conferences, speaking in public, writing and revising one’s papers. A PI provides an environment that encourages this professional growth, but that can have varying amounts of direct PI-student contact. I think students and postdocs who join supergroups know (or should know) that they are signing up for this type of environment. In my own training, the largest group of which I was a member had the least interaction with the PI but was the most educational.

    One way in which a supergroup can “fail” is when a student realizes (or needs to realize) that pursuing a PhD was not the best personal decision. It takes a lot of courage to realize this reality and choose to leave – most students don’t. There is plenty of non-science administrative work to be done, especially in a large group, and a student can appear quite busy with these tasks without progressing in their education. A supergroup structure allows students this refuge, at least for a lot longer period of time, when it would have been better for the PI to confront the student and help them find alternative career options.

    Eugene’s cynicism on fame in science is pretty tiresome (and I’m not famous by any means). Every competitive undertaking has superstars who collect a disproportionate amount of attention and resources. The difference between a star and a regular professional baseball player (who themselves are statistical outliers) is about 2 hits per week – you’d be hard-pressed to notice the difference without keeping track of the statistics. Same for science or anything else – there are good people everywhere doing good work – but it is human nature to look for the outliers. These supergroups became “super” by being a little bit more productive than their competitors over a long period of time. The same is true of the best doctors, lawyers, businesspeople, etc.

    We all need to make peace with this reality and choose whether to pursue the superstar level of achievement – it does not come without significant personal sacrifice and there is certainly no guarantee of success.

  7. eugene Says:

    SGL,

    I didn’t say that Langer’s lab provides a bad educational experience. Like you say, it could be a good educational experience, with senior members helping out, etc… However from my personal experience when you’ve got a bunch of people of equal rank, some of them take offence when you tell them that: looking at it logically, their experiment was useless and a waste of money for the group and you would advise them not to waste their time or the labs money on similar things in the future. What I learned is that you have to be the PI to say that, even if you are correct and are saying it gently or in a joking manner. If your PI is gone most of the time, that doesn’t tend to happen quite as often except for a general statement at group meetings every few months. Saying this to someone of equal rank half the time makes people very defensive. (Fine what do I care; waste your time and the group’s money. After all, the nickname for the boss is Mr. Moneybags, isn’t it?)

    I was just commenting on the dynamics of having such a huge group at such a prestigious university as MIT. It’s a lot easier to be ‘famous’ as the leader of a huge group, then if everything depends on you yourself when you’re a baseball player isn’t it? There are definitely good people doing good work everywhere, but the system is set up to attract a lot of the ones who want to be really famous. I was just giving them a leg up, by telling them which pitfalls to avoid, and how it really works, in my original comment. It doesn’t matter if you don’t know that much more chemistry than the average postdoc everywhere. Once you’re chosen to become faculty member there is a clear formula to follow if you want to be a big shot. If I may plagiarise myself for a bit…

    “If you’re in charge of a small group, then the failure of a graduate student to get any papers will be magnified since you don’t have too many offsetting grad students that are publication machines. In fact, you might have none. I’ve seen a few profs with medium-sized groups that were taking off, but were blinded by their own hubris, forgot that they were not in a top ten school, and decided to ‘fire’ two or three students or postdocs who weren’t performing well enough (not getting a Jackass or Andjewandte or not coming in on the weekend). Not only does this cause potential students in a non top ten school avoid you (since they care more about lifestyle and not monastic scientific pursuit), but it magnifies the failures of your remaining students (should they happen) as your own failures. Automatically, you’re now an average scientist for the rest of your career and not Robert Langer. Not that there is anything wrong with that, as you can do very good science and pay more attention to individual students, but that’s not what some of the more ambitious types wanted before they cannibilized their group. The lesson is get big as fast as possible with as many driven types as possible and then just sit back and manage your success. You have to get rid of someone who creates a bad group dynamic or is lazy or just plain stupid of course, but once it starts being every second person who works for you, then you’re not doing it right (unless you’re in a really crappy department). Once you’re going, you can move to better and better departments and become bigger and bigger and have to worry about failure even less.”

    You see, I’m only helping aspiring professors here. If they choose to have a small group and lots of interactions with their students, they’ll still be working a lot, but that superstar is not operating at a level that much higher, as you say in your baseball analogy. They do not even have to be more productive per capita if they follow my rules on how to build a supergroup. And if they are a young professor at a top notch place like Northwestern (for example), technically you don’t even have to be sane, as talented students and postdocs will keep joining your group and publishing those papers. The trick is to get that Assistant professor job in as high a profile institution as possible. I promise you that someone without pedigree (on the off-chance that they are hired at MIT) will build a supergroup just as fast as someone with, as long as they follow my simple plan. Hordes of students doing something is good enough for five to ten papers a year by the time you’re up for tenure. A few Jackasses in there and you’ve got it made. As a bonus, your senior students and postdocs can also write grants and some of those highly citable reviews. When you’re a professor, you’re judged by the performance of your group and very often the bigger the group is, the more performance. Of course, the individual glory goes to you, plus a little bit of glory crumbs to the graduates who get a reference letter from a famous group. That’s the way the system is set up; as long as there is a huge overproduction of PhDs, you should take advantage of it. If you’re a smart person in a top five place and you have a small group, you may be a very excellent scientist, but you’re not playing the game right or you’re choosing not to take advantage of it.

    P.S. I don’t think I like this whole limelight thing. Pretty soon someone will figure out my real identity and I’ll never be able to work in chemistry again, or something… But I guess I was asking for it with comments like that.

  8. eugene Says:

    “My view is that one needn’t adopt as cynical a view to show what is effectively the same outcome. That is, I will assume that professors want to pay attention to all of their students, but sometimes, they just can’t.”

    Yes, and they also really want to meet with that one student who gets on their nerves by not having any results, but the just ignore them for another two weeks and tell them to repeat the experiment but in different solvent. In other news: human nature is basically good in vast majority of people; global communism is now realized as an ideal form of government.

    Well, I really do agree with your last paragraph. Having some research professors, who would have secure careers from the university and have an option to move between groups (so that they could contradict their main boss from time to time without risking job loss), would solve a lot of problems. This way maybe teaching courses could be passed off onto actual professors (even if they are only ‘research professors’) and not on sessional lecturers that are paid hunger wages. Class sizes would also benefit.

  9. Lila Says:

    Paul, I’m curious how your grad adviser ran his group — it was huge, no? I know Harry has Jay Winkler (and maybe still some super-postdocs like Angelo DiBilio? Is he still there?) but from experience, I also know that the organic synthesis groups at Harvard had no such minibosses, at least in the early-mid 90s.

    And can any of your commenters attest to firsthand experience of how Langer runs his group? I asked him once, and he said he checks in with each of his grad students every week.

    Not exactly related, but I once reported and wrote a big story on undergraduate research experiences and one of the things I tried to look at was whether it mattered if you were at a small liberal arts college (where people are always saying how great it is to get so much attention from profs) or at a big research university. I contrasted two excellent schools. Totally non-scientific, but I did interview something like 40 or 50 undergrads doing research at the two schools (and read a bunch of actual scientific reports on the topic from NSF and elsewhere). I only found one student who was foundering and practically ignored, and she was at the large school. But other students even in the huge supergroups at the large school were doing well — largely because they had approachable grad students as minibosses.

  10. Superlabs and the Expendability of Grad Students [Link] « Macdrifter Says:

    […] than go on a rant about how much agree with this link, I’ll just leave this […]

  11. Matt Says:

    @eugene,
    I’m giving a test right now. Upon reading your comment “And if they are a young professor at a top notch place like Northwestern (for example), technically you don’t even have to be sane” almost laughed out loud. (I am an NU alum, btw)

  12. wolfie Says:

    My wife has been doing research on Alzheimer’s disease for 15 years at some of the most prestigious institutions in the world, including the University of Heidelberg, and the outcome for patients, not only of her own research in this time, has been about zero.

    So, why do we do all of this ? To run into insolvable questions of untreatable problems ?

  13. Mrs._Beakley Says:

    I really think this kind of introspection is important, and it’s a shame none of us can spend enough time away from the bench to do a quantitative analysis of what makes some labs more productive than others on a per-member basis. I want to add to the complexity by bringing up the issue of the unequal contributions of graduate students and postdocs. I worked in a superlab as a graduate student, which had three graduate students and about thirty-eight postdocs by the time I graduated. I now work in another superlab, also with forty members, but it is about half graduate students and half postdocs. My graduate lab generated more than twice as many publications per year as my postdoc lab does. I attribute the bulk of the difference to the number of postdocs in my graduate lab, who are, on average, more productive per year than are graduate students. My unscientific opinion is that a disproportionate number of breakthroughs and landmark papers, especially in superlabs, are the work of postdocs. So, by all means get as big as possible as fast as possible, if that is your game, but make sure to negotiate as much postdoc support as possible into your startup package.

  14. Special Guest Lecturer Says:

    Wolfie, you’ve convinced me. Time to pack it in and go to astrology school.

    The more interesting aspect of this issue of supergroups: if you are the US government or a private foundation with research dollars to invest, who do you give it to and why? What is the marginal impact per dollar? Do you invest in a bunch of assistant professors with demonstrated promise, spread the money around to solid mid-career investigators, or make large investments in superstars?

    Does the answer change if the money is for research infrastructure (i.e., big, expensive instruments) rather than personnel and consumables?

  15. wolfie Says:

    For a single scientist, supergroups are not interesting, because he or she will not be able to achieve anything reasonable him- or herselves, although in somma, great things may or may be not achieved this way.

    Could Trommsdorff have invented Plexiglass in a supergroup, and made his money afterwards ? Probably, no. Why then, should he have been interested in any ?

  16. wolfie Says:

    As I said, it was Röhm who fondled Röhm and Haas, but is it really important ?


Leave a Reply

*