A ton of blog posts and a few papers have been written on the issue of the optimal size of a biomedical research lab. If you would like a nice digest, see this Youtube video by Jon Lorsch, the director of one of the NIH institutes. Jon makes a very good case for smaller labs. They allow for more diversity, better mentoring, are more efficient with their use of funds. Unfortunately, almost all incentives in academia are for group leaders to have bigger labs. The PI of a larger group has more security in terms of funding and employment, more prestige, more negotiating power in almost any situation. Almost no one in academia looks at how efficient in terms of output per dollar a group is – they invariably look at the total output of the lab, ie. the total number/prestige/citations of pubs where the PI is senior author. Similarly, no one looks at how well the PI’s trainees do on average – they only look at the few superstars that have come out of the lab.
All in all, for the group leader, there are essentially no drawbacks to having a larger lab – bigger is almost always better for them personally and professionally. It’s sort of like monopoly power – it’s better for the monopolist, but worse for the market as a whole. And like for monopolies, we need systemic solutions to disincentivise having larger labs. The main Polish public funder of science, the National Science Centre, has for instance put a cap of 3 grants per PI, which is reasonable, given that postdocs and grad students can apply for their own fellowships. NIGMS, whose head is Jon Lorsch, put out a MIRA program (maximizing investigators’ research award), which is limited to one grant per investigator.
When designing these systemic solution, one should really think about what is optimal in terms of lab size. There is definitely a lower limit below which the higher efficiency of a smaller lab drops off quickly. A lab below that limit will lose out on economies of scale, for instance when negotiating consumable prices with vendors or buying large equipment. They will lose the synergy of sharing reagents and know-how. They will also be very inflexible – unable to expand to new territories and unable to take any risks. If all you have is one or two projects, they’d better be safe bets.
My intuition is that in my field the optimal lab size is probably 6-10 people, including the PI. Jon Lorsch cites data suggesting that ~$300 000 per PI is about the limit where the productivity per dollar drops off. This is definitely less than what you can support a group of 10 or even 6 people on in the US, but perhaps the different labor costs in different countries will determine where the cutoff is. It would really be interesting to do these studies in Poland to guide the funders’ decisions of what size groups to favor.
More importantly, there are probably cases where larger labs really are better. There are specific types of large projects where large hierarchical structures do better. Other than that, I am sure we will be able to find examples of labs that work efficiently despite their large sizes. Perhaps the PI does no teaching, so they can devote more time to science, perhaps they are an amazing mentor and so each and every trainee becomes more independent more quickly, perhaps they are very good at picking the right people, and so everyone works more efficiently. As much as I dislike metrics, I think that reviewers and policymakers should have access to some sort of “productivity per dollar” number that they could use to gauge lab efficiency. We should study examples of labs that score highly on that metric to figure out what makes them special and let that guide our policy.
Finally, I think that rather than artificially putting a cap on lab size, perhaps we should consider ways to mitigate the disadvantages of leading a smaller group. Better shared administrative support could help PIs who cannot afford their own lab admin. More comprehensive common facilities would make owning expensive equipment unnecessary. Better support for bridge funding would make it easier for smaller labs to pursue risky new directions. Same with funding mechanisms specifically designed for exploratory research with lighter requirements for preliminary data. I am sure many PIs would be happy to lead smaller groups if the advantages of expanding lab size were less pronounced. In other words, perhaps it’s the carrot that is the better solution, not the stick.