"Transformative" research can't come from milquetoast

Philip Ball writes in The Guardian about another new initiative from NSF to fund “potentially transformative” research (“Science funding tends to favor mediocrity over grand ideas”.

He begins his essay with this:

The kind of idle pastime that might amuse physicists is to imagine drafting Einstein's grant applications in 1905. "I propose to investigate the idea that light travels in little bits," one might say. "I will explore the possibility that time slows down as things speed up," goes another. Imagine what comments these would have elicited from reviewers for the German Science Funding Agency, had such a thing existed. Instead, Einstein just did the work anyway while drawing his wages as a technical expert third-class at the Bern patent office. And that is how he invented quantum physics and relativity.
The moral seems to be that really innovative ideas don't get funded that the system is set up to exclude them.

The system is set up to exclude really innovative ideas. But Einstein is a really misleading example.

For one thing, Einstein didn’t need much grant funding for his research. Yes, if somebody had given the poor guy a postdoc, he might have had an easier time being productive in physics. But his theoretical work didn’t need expensive lab equipment, RA and postdoc salaries, and institutional overhead to fund secretarial support, building maintenance and research opportunities for undergraduates.

It is a better question whether we would have wanted Einstein to spend 1905 applying for grants instead of publishing. But even this is terribly misleading. Most scientists who are denied grants are not Einstein. Most ideas that appear to be transformative in the end turn out to be bunk. Someone who compares himself to Einstein is overwhelmingly likely to be a charlatan. There should probably be a “No Einsteins need apply” clause in every federal grant program.

Setting aside the misleading Einstein comparison, our current grant system still has some severe problems. Is it selecting against “transformative” research, or big breakthroughs? I would put the problem differently. “Transformative” is in the eye of the beholder. Our grant system does what it has been designed for: it picks winners and losers, with a minimum of accountability for the people who set funding priorities.

We might be perfectly happy if the winners were scientists who all go on to make important breakthroughs. But in reality our system picks winners in a way that often selects against creativity and significance, and selects for established networks of institutions and senior scientists, and above all “grantsmanship”. What is the difference between a 35-year-old assistant professor who becomes the manager of 1.5 million dollars of federal money, and his 37-year-old colleague who has been denied twice for the same grant before applying for tenure? In a system where fewer than 20% of grants are funded, the difference may be luck. The “transformative” value of either person’s ideas hardly comes into this calculation. Yet this is the system we are currently using to staff the next generation of senior science positions.

Anyone who has submitted grant applications in multiple years can see this in action. The reviews in one year often completely contradict those of the previous year, even for the same project. An application’s chances of being funded are based on the luck of who reviews the application and who is in the room. The way to ensure bad luck is to be an outlier. I have applied for federal grants several times, and have often had strong reviews but have never been funded. A good application can take weeks of effort to prepare, as much as a research paper. Yet only a small fraction of applications are funded. For me, each time has been a costly training for writing the next unsuccessful grant application.

I don’t want there to be a pool of money set aside for “high-risk” or “transformative” work. I want the agencies to set transformative objectives and to fund projects in accordance with them. If my scientific objectives don’t match those of the agency for my research area, I want to know that so I don’t spend time on fruitless applications. If a grant agency has milquetoast objectives, I want a transparent process by which ordinary scientists can participate in changing those objectives.

Anthropology has its own unique grantwriting challenges, and we can’t easily generalize across fields. Some of the very large grants available from NSF for human evolution research are strongly interdisciplinary, and much of the budget of funded projects is spent outside of anthropology. “Transformative” research in this context may be conceived as the collaboration of scientists from different areas, even when the results themselves may be quite conventional. In my opinion, funding productive field projects is the most effective use of federal money in paleoanthropology. Every new fossil discovery might be the one that transforms our understanding of the important events in human evolution. But those field projects may actually inhibit “transformative” research if they do not make their results available to other scientists in a timely manner, if they do not openly archive and document their activities in the open, and if they do not contribute to the education of future scientists.

Many have pointed to the problem of transitional funding for early career scientists. Few have noticed that this is a decay curve. As researchers get older, some of them get their first grant, but many others stop trying. Personally, I’m still quite a ways from the age when the average grant recipient receives his first federal grant. But I’m very glad that I’ve chosen a research area where I can do great research without that federal money.