Let's start with the important information that until last month I, like anyone who needs to read this post, had never applied for, let alone received, US National Science Foundation research funding. Having been out of the US for most of my time since earning my Ph.D., and other extenuating circumstances, kept me from applying to this extremely important source of funds. As with my posts on applying for NIH and ERC funding (which I didn't get), I'm writing this not as an expert, but because most people who write advice on applying for NSF funding are to varying degrees experts, and have been doing it for so long that they have no idea what us newbies might not know. I've never been on an NSF panel, I've never been to NSF, and all my attempts to talk to NSF employees have been unsuccessful. I am, like you, an outsider. So I learned a lot along the way, much of it later than I should have, and I'd like to share some key points. Before we begin, let's take a moment to contemplate this bodacious caterpillar I found in the University of Wisconsin Arboretum last summer:
1. NSF, we can agree, does not have the funds to support more than a small fraction of the grant applications they receive. Other huge governmental funders of science, like NIH, are in a similar situation. (I am assuming here, perhaps in vain, that the US government under El Presidentisimo continues to invest in funding basic research at some meaningful level.) This is a problem not only in terms of almost-everybody-doesn't-get-funded, but in terms of the phenomenal amount of time that scientists squander writing unsuccessful grant applications. In many cases, months of each year are spent applying for funding that is not available. On the flip side, NSF has to harness huge amounts of scientists' time to serve on committees, wading through the reams of applications finding reasons to reject as many of them as possible.
NSF has attempted a partial solution to this problem: the preliminary application. Basically, one starts by writing a short (four pages of text, plus many ancillary documents, including a one page Project Summary) version that has to be approved by a committee before one is invited to submit a full application. Most applicants (about 75%, in the program I'm applying to) only have to write the short version before being rejected, and the committees mostly have to read piles of short applications, with relatively few longer ones coming after that.
2. Writing the preliminary application was honestly not that bad. Several reasons: No budget is required, and many of the ancillary documents that NSF needs before they can fund anyone don't get submitted until your preliminary proposal is approved. More importantly, I have good collaborators, who are in practice writing these things. There is a huge amount to know about NSF specific 'grantsmanship.' When the committee gets to our application, after having already scanned scores of others, they will be both eager to find something really interesting that keeps them awake, and eager to find some reason that the thing can be rejected, so that they can get it over with. They will be looking for key phrases that everyone should have, and possible pitfalls indicated by things only people who have served on these committees (or maybe only that particular committee) know about. In short, it would be tremendously surprising if someone who just had phenominal scientific ideas but no insider guidance as to the evaluation process ever got funding. I am lucky to have had that guidance; while it means I did a lot of rewriting to try to conform to a culture I've never encountered, it also, hopefully, means we have some chance of being invited to submit a full proposal. If so, that's what I'll be doing this summer, again with about a 20-25% chance of success.
3. As with any funding application, reading past successful applications to the same program is important. Notice not only the language used, the level of methodological detail given, and the structure of the proposal, but also the scope and scale of the proposed science.
4. Write, and revise, the one page Project Summary, and make sure everyone in the project agrees on it, before bothering with the longer Project Description. I made the mistake of drafting the Project Summary, receiving only minimal feedback on it from one of my collaborators, then writing the rest of the grant. By the time I got more extensive feedback from this collaborator, I had only a few days to reconsider the scope of the work being proposed and extensively rewrite. The proposal ended up much better for it, but I could have gotten a lot more sleep if I'd pushed for more feedback after writing just the summary.
5. The whole thing really isn't that much writing. Given that one knows what one is doing, and has good communication with collaborators, a good draft can be banged out in a couple of solid days.
6. In most cases, the one page Summary has to be uploaded as unformatted text, and it takes up more space the way NSF automatically formats it than it would if you or I formatted it according to their rules. I, and a few people I've talked to, ended up hacking down the one page summary very shortly before the deadline when we figured this out.
7. In order to apply, you need an "NSF ID." Looking on the NSF web page, you will find lots of information on retrieving your NSF ID, what to do if you have two NSF IDs, whether NSF ID is the same thing as various other identifiers NSF has used in the past, and so on. You will not find any information about how to get an NSF ID if you don't already have one. If you call NSF to ask how to get one, they will be so confused by you that they won't be able to help you, as if you had called to ask how to breath in before speaking. You just do it, and you must already know how. So I will tell you the secret: somewhere in your university or other approved research organization, there is some individual with the official power to communicate with NSF to get you your very own NSF ID. You will have to find out who that person is and request the NSF ID several days before the application deadline. My collaborator at another university didn't get a reply to his request until a few hours before we submitted, and we were actively discussing what would happen if we had to leave him off the PI list. Don't let that happen to you.
8. Which reminds me of another important point. I was advised to contact the NSF Program Officer for my application when I had a nearly final Project Summary worked up. As mentioned above, that didn't happen until less than a week before the grant deadline, at which point the Project Officers were all swamped by others trying to meet the same deadline. I emailed the Program Officer, and got back a very short email, but never got to talk with her. Finishing your Project Summary early will allow you to get input from your Project Officer.
9. Be smart, work very hard, and have extremely good luck.