Science magazine recently interviewed me for an in-depth/opinion piece they ran last week here (paywalled). The gist of it is that the NSF is worried that acceptance rates are getting too low (15% now) and that this is putting a big burden on reviewers:
Later this summer, NSF’s astronomy division intends to announce a new policy that will “strongly encourage” scientists to submit just a single proposal for each annual funding cycle. The voluntary cap is designed to boost success rates, which would please applicants. It’s also meant to ease the workload and frustration levels of peer reviewers poring over proposals that they know have little chance of getting funded.
A strict limit could be “a disaster,” worries Jason Wright…“having a ton of co-PIs with different skills is what makes an application so strong.”… [He] says a ceiling would also change his approach to grant writing: “If you only get one shot, I’d put in more sprawling proposals, and ask for more money.”
I encourage the AAAC to ensure that its survey includes plenty of input from low-effort co-I’s, science-PI’s, admin-PI’s, and soft-money researchers.
Let me elaborate on my concerns regarding why a “strict limit” would be a “disaster” (apologies for the hyperbole, which was made in an oral conversation):
1) This will hurt soft-money folks and young researchers most — running dry can end their careers; tenured faculty can try again next time.
2) Many proposers add co-I’s (not necessarily “a ton” of them) who provide specialized service at low effort that significantly strengthens the proposal. Many researchers will have to decline to join collaborations as co-I for, say, 2 weeks’ effort if it means they can’t submit a proposal to fund their own group. This effect will not help the submission rate problem but it will do structural damage to collaborations and make proposals weaker.
3) Acting as administrative-PI is an important role for tenure-line researchers to play for postdocs and adjuncts of various sorts who, for purely bureaucratic reasons at their host institutions, cannot PI proposals themselves. This rule would discourage this source of collaboration and professional development.
4) If fewer than 15% proposals feature duplicate proposers, as this letter suggests, then this proposal won’t actually have a big effect on success rates, but it will hurt the quality of the science in those proposals.
Finally, Ulvested neglected two more ways to boost success rates (perhaps beyond his control, but not to be neglected): more funding, and, to a lesser extent, more support and less stigma for those thinking about leaving the field. The 1-proposal plan feels a lot like treating a minor symptom (reviewer burnout) of a much bigger disease. As someone who has served as an NSF reviewer, I do not support this “1 proposal” plan.So, I would ask that this request be modified to limiting only full PI-ships to 1 per year, and allowing multiple low-effort co-I and admin-PI roles. But even with this less strict rule, I still worry about the effects on the careers of soft-money and untenured researchers, and suspect that it will cause us to unnecessarily lose a lot of really good people.