Moses Charikar and I organized a grant writing panel at STOC 2013. The panel was well attended and I think (or at least hope) provided useful information to the audience, thanks to our great panelists: Joan Feigenbaum, Piotr Indyk, Bala Kalyanasundaram, and Salil Vadhan.
Panelists had many insights on aspects of achieving funding, some highlights that stuck in my memory:
1) There are many funding opportunities out there apart from the “obvious” sources (i.e., NSF theoretical foundations) – check those out. Don’t be afraid to contact people with lots of funding experience for advice or even to ask (if appropriate) to join future large projects.
2) That said, you should always make sure that you actually want to do the proposed research, and shouldn’t join a project you don’t care about just for the funding. This is particularly true for large projects or funding from “mission oriented” (as opposed to “basic science oriented”) agencies, where the funding often comes with plenty of strings attached.
3) If you work in theoretical CS in the U.S., you should definitely consider applying for grants from the Computing and Communication Foundations division (and in particular the Algorithmic Foundations program). Grants have gotten bigger and competition smaller than people may remember from past years, and it’s definitely worth the trouble.
4) NSF program officers are your friends, they are more than happy to talk to you, and can often help you find the right program for your proposal.
5) Especially (but not only) in NSF CCF grants, people understand that theory research is unpredictable. If you’ve done good research in the area, nobody will blame you for not solving the particular problems you set out in your proposal. That said, the proposal should have a mix of problems of various difficulty and “speculativeness”. If the proposal just lists some important open problem without suggesting that you have any new approach to solve them, it’s unlikely to get funded.
6) Collaborative projects are much easier to manage when it’s 2-3 PI’s (primary investigators) who are close collaborators than project involving a large number (say 7 or more) PI’s.
7) Writing a proposal involves making choices – how big the team, one area or interdisciplinary, contain everything you want to work on or leave topics for future proposals, apply to NSF or to more “mission oriented” sources of funding. All those choices are non-trivial and have pros and cons, which you’d do well to consider carefully and consult with people who’s had experience.
8) Proposals (especially but not only large multidisciplinary ones) often don’t get funded in the first attempt. People said they found that the proposal became better and the team more cohesive in future iterations.
While some of the advice is universal, we did not have a panelist with specific expertise in funding outside the U.S. We hope to have some guest posts on this blog about this topic in the future, though in the meantime we would welcome tips about resources on grant writing in the comments.
One such resource that could be very useful was collected by Moses Charikar and Piotr Indyk – this is a collection of successful NSF CAREER proposals along with retrospective comments by the authors. We are truly grateful to these researchers for making available this valuable resource to anyone that is considering submitting a grant proposal. Salil Vadhan also made available his multi-disciplinary grant proposal.
People might also find useful Joan Feigenbaum’s presentation from the event.