8 Comments
Suppose a funding agency happens to have some extra money and needs to decide how to invest it. Should it invest that extra money in a large, highly productive laboratory, so that laboratory can expand a bit more? Or should it invest that extra money in a small to moderate size laboratory? Given our inability to predict the future with great certainty, which approach represents the smarter investment strategy?
Jon Lorsch, the director of the National Institute for General Medical Sciences (NIGMS), has posted an interesting video on just this question. Lorsch invokes the well-known economic principle of “diminishing marginal returns,” that is, each additional input on the margin yields less and less additional output. Thus, for a small to medium sized laboratory, an additional 20% of funding may yield 15% additional output; on the other hand, for a large laboratory an additional 20% of funding may yield only 10% additional output. Lorsch cites published reports suggesting that this happens in biomedical research – marginal returns decline with increasing investments.1, 2, 3, 4
Lorsch notes that there is another potential advantage of using marginal funds to support small-medium laboratories rather than providing additional support to large laboratories. Funding agencies have limited resources. Giving extra money to small-medium laboratories means that agencies can fund more laboratories, offering the potential for a more diverse portfolio. Science is inherently unpredictable – we simply don’t know where the next breakthrough will come.5 So in the interest of maximizing the likelihood of breakthroughs, we might follow the maxim of many smart investors – spread the risk across a wide spectrum of projects.
We encourage you to watch and send us your thoughts about Dr. Lorsch’s thoughtful video, as we continue our dialogue about accountability and stewardship.
The portfolio ideas makes a lot of sense of distributing the limited funds among a large number of laboratories. But there is another paradigm that should be evaluated that today’s research requires a more sophisticated and expensive infrastructure. It will be interesting to investigate the relationship between the type of infrastructure and groundbreaking research results. Usually the example of fusion research is presented as a good example but is this true in the majority of groundbreaking research results?
I think some large research groups must be allowed to grow to a size that’s larger than median. Clearly, there is a distribution of lab productivity (per capita). While it may be true that there are diminishing returns on subsequent R01’s, the fact of the matter is that a superstar’s 3rd best idea may be still significantly better than a more average scientist’s best idea.
If we artificially cap the number of R01’s or total funding per year to a lab, we end up losing incentives for great PI’s to continue being great PI’s. Some great PIs who are more ambitious will leave academia to direct a bigger, better-funded group in industry. Although industry does great research and contributes meaningfully to technology, I think we all agree that industry’s investment horizon is shorter than NIH’s, and that short-term profit-seeking is not ideal for long-term research progress. Other PIs may mentally clock out after hitting their cap of 2 R01’s, because what’s the point of working harder to publish your 5th Nature paper if you’re already secure in renewing your existing one and don’t have the possibility to expand your research?
Also, while it’s certainly true that “breakthrough science” can come from any lab, large productive labs that I know of tend to produce *multiple* breakthrough science projects. It’s not clear to me that distributing funds more widely increases total number of scientific breakthroughs. If anything, I feel that the “breakthroughs vs. dollars” curve may be concave to the origin. As more brilliant graduate students and postdocs from different fields are gathered in one group, there’s a super-linear return on investment, because cross-disciplinary possibility grow combinatorially.
I want to mention that I’m an ESI PI in my early 30’s who does not have an R01 yet. But I think it would be a great tragedy for NIH to pursue a model of “socialist science” where funds are uniformly distributed to everyone claiming to do science. Science is not a bell curve; science is a power law distribution. What motivates me (and likely every other ambitious scientist) is the possibility to become the tail end of the distribution.
Dear Dr. Lorsch,
Thank you for your video is very interesting video. A number of friends in Brazil have listen to it and discussed the differences we might have on using the terms “bigger lab” and “PI” over here.
Based on our own experience, having more staff (PI’s, lecturers, …) congregated on a single group or lab has been a way to survive to the scarcity of support (infrastructure and money).
One again congratulations for your comments.
Edmundo Grisard
It’s nice to see some thought going into ‘productivity per dollar’. I think that absolutely should be a criterion for funding. Probably even an explicit main criterion — more important than the highly subjective and ultimately questionable ‘innovation’ or ‘environment’.
It’s also about intellectual diversity. Sure you can dig deeper and deeper in the same mines. And yes they might keep yielding. But it’s good to keep prospecting for new gold too. Haven’t a lot of historical breakthroughs come from unexpected directions?
I support the concept of investing limited resources on more small-moderate size abs, than on a few large labs. This is based on the following reasons: (1) I believe this approach better diversifies the research portfolio, enhancing the chance of ‘breakthroughs’. (2) History teaches us that often the initial stage of ‘breakthrough’ came from one or a small group of individuals who were willing to think outside-the-box and willing to take chances. ‘Breakthroughs’ can happen in small-moderate size lab. (3) On the other hand, the usual top-down management style of many large labs has the risk of stifling junior scientists/trainees in their desire to explore high-risk yet innovative ideas. (4) From my own experience and observations over three decades of academic research, I believe small-mid size labs provide a better training environment for students than large labs. Students in small-mid size are better guided and supervised, because their contributions to the productivity are vital to the success of the lab. On the other hand, students in some large labs may work in a data-assembly line.
I completely agree with all these comments. I particularly agree with the comments about the quality of training in the smaller or midsized groups, which is, in my experience, much better than that in large groups.
I would rather disagree. “Data-assembly line” is now, unfortunately, the norm in science, but especially so in smaller labs where obtaining or not obtaining a second R01 (or renewing the only one) makes a survival difference, and all life is a daily single-minded struggle for more preliminary data, and younger and ambitious PIs are not eager to hear input from their students and postdocs. In bigger labs, however, there may be more leeway.
As everywhere, there should be only one criterion: merit. Introduction of any other considerations, be it expectations, equality, diversity, whatever, only results in unintended consequences and overall degradation.