27

I am a machine learning researcher, and my field is beginning to embrace the massive, multi-institution collaborations that have become common in other sciences. For example, I am a coauthor on this paper, which has 52 authors. I am also involved in on-going large collaborations organized by EleutherAI, Google, and HuggingFace.

It was recently pointed out to me that the US National Science Foundation requires disclosing all “Co-authors on any book, article, report, abstract or paper with collaboration in the last 48 months (publication date may be later); and [c]ollaborators on projects, such as funded grants, graduate research or others in the last 48 months.” (II-5e) The document also implies that people who are listed as collaborators are unable to review grant applications.

It seems very possible that I will have hundreds of collaborators in the field of natural language processing over the next year. However, I worry that this will make it very difficult for me to get NSF grants, if I collaborate with a significant portion of the community that has the same interests as I do.

How do researchers handle Conflict of Interest disclosures when they have hundreds of collaborators? I assume fields like physics and astronomy, which can have several hundred collaborators on a single paper, must have a solution, but I’ve been unable to figure out what it is. I am particularly interested in the NSF, but am also interested in hearing general answers about dealing with conflict of interest policies that require disclosing all co-authors.

17
  • 3
    Hire a manager. Include them in the grant(s). Or, depend on the research office at the university of one of the authors. – Buffy Jun 2 at 15:09
  • 2
    Someone at CERN can probably help with ideas. – Buffy Jun 2 at 15:15
  • 6
    @Buffy I thought about cold emailing someone at CERN, but figured it made sense to ask here first as I don't know anyone at programs like CERN. – Stella Biderman Jun 2 at 15:33
  • 3
    I'm not sure the title reflects the question well, maybe something like "Who can review my grant if I collaborate with everyone in the field?" – Azor Ahai -him- Jun 2 at 21:40
  • 5
    Don't email to CERN, and @Buffy I wouldn't suggest to do so: CERN, being a European joint-venture, has probably institutional sources of funding, or different funding agencies, and they are probably not familiar with NSF (in general, asking European people about US funding agencies is not a good idea). Contact NSF instead. – Massimo Ortolano Jun 3 at 6:39
32

As a mostly theoretical physicist with some peripheral involvement in very large experimental collaborations, I asked essentially this question of an NSF program officer a couple years ago. The program officer said that, for now, it was not necessary to list all the members of huge experimental collaboration, only those that I had actually worked with myself. However, he also warned me that this might be different from program to program, and it could change from year to year. So the only way to know for sure is to contact the program officer.

3
  • 18
    That's a great answer because it is based on real first hand experience! And it highlights that one should reach out to "the other side" when anything is unclear... Asking random people online is really only supplemental. – user2705196 Jun 3 at 12:33
  • 6
    Just as a side comment for the OP, if I was the NSF I'd treat an experimental collaboration between ~5000 physicist very different from a paper with ~50 authors... So definitely don't jump to the conclusion that you can omit your co-authors from the list. – user2705196 Jun 3 at 12:35
  • Yes and while it's entirely possible you'd know nothing of 36 students working in each of 25 collaborating labs, how could you not know at least the names of every co-author? – Robbie Goodwin Jun 7 at 23:22
21

The document does not imply what you say. In a section "Potentially Disqualifying Conflicts of Interest" you can see that the NSF can issue waivers.

This is typical. Conflicts of interests need to be reported when they reach a certain level, but they are not always automatically disqualifying.

You might contact a program officer at a relevant NSF program and ask for how this works in practice.

4
  • I am aware that II-2 allows for the existence of waivers, but the very existence of waivers confirms the assertion that collaborators are typically disqualified from grant review. "You still need to report the co-authors, but it's typically easy to get a waiver for review if you have only collaborated a couple of times on massive projects" could very be an answer to this question. – Stella Biderman Jun 2 at 15:42
  • 9
    I hope my answer is generically useful, but really you need to talk to an NSF program officer. When I email them with questions like this, on technical issues on proposal writing and reviewing, they have always replied in a friendly and useful manner. – Terry Loring Jun 2 at 16:29
  • I should add that I have experience on how other governments deal with conflicts of interests in science funding, but not so much with US agencies. I hope someone else can give a better answer. – Terry Loring Jun 2 at 16:33
  • 6
    The Population Biology panel of NSF granted a more-or-less blanket waiver to ecologists participating in workshops at the National Center for Ecological Analysis and Synthesis – Ben Bolker Jun 2 at 19:04
13

Your question seems misguided, it is not the research group that "handles" this conflict of interest. It's the funding body that will have to decide how to deal with potential conflicts of interest of their reviewers.

To answer your question: the way large research groups "handle" this is by listing all their collaborators and co-authors as requested. And that's it.

As to what happens on the NSF side, I don't know. But I'd safely speculate that no grant has ever not been reviewed just because there are not enough reviewers who haven't co-authored a paper with you!

4

You're overthinking this, and in particular your concern that a lack of eligible reviewers would result in your application being tossed I think has the process the wrong way round.

I don't work for NSF but I do work for a US funding agency. I don't know how NSF does it, but we are obligated to review compliant applications that are submitted. While the call for applications are out on a program I manage, I am recruiting the review panel in parallel. After the review panel is assembled and applications are in, I assign reviews to panelists and ask them to check for COIs. If COIs or potential COIs arise I notify our legal counsel and ask whether this is disqualifying or whether they can still review; if it's disqualifying, I assign to a different member of the panel.

If a situation arose where I could not identify sufficient panelists to complete a review of your application, I would go find someone who could. At the end of our merit review process I have to provide documentation explaining our funding selection decisions. I don't think my manager would appreciate it if one of my justifications was "this was too hard to find reviewers for so I gave up :("

2

This is easy. You list your hundreds of collaborators as required. You can potentially go to jail for issues around this -- example: https://www.sciencemag.org/news/2020/01/us-charges-prominent-harvard-chemist-failing-disclose-china-ties -- and you need to avoid that.

Be extra careful about making sure you get your foreign collaborators right. People are finding themselves in very tricky legal situations over this in recent times.

3
  • 1
    -1 This seems overly alarmist. The professor went to jail for making false statements about his collaboration with Chinese nationals while holding a US security clearance. That’s miles away from anything I’ve suggested. – Stella Biderman Jun 5 at 3:02
  • 2
    As our VP of Research keeps pointing out, failure to disclose is a false statement. The case I'm pointing to resulted in many communications from the Feds to research orgs to the effect of "get your disclosures right", which is why I'm pointing to it. That case opened a xan of worms. – Scott Seidman Jun 5 at 11:21
  • 2
    While it's true that collaborators must be listed, I'm not sure the question was whether disclosure is optional, but rather something like: are there exceptions, how do grant agencies deal with a huge number of co-authors, are many co-authors really a problem when it comes to finding reviewers etc.? – henning Jun 5 at 18:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.