Ankur Moitra comments

Some remarks/advice:

1. Organization. You can’t see it from just the final version, but something that changed drastically from my first draft was the order of the subsections. At first, I had all of the prior work together, then followed by all of the proposed new directions. This has the effect of making it feel disjointed. Ideally, a proposal should integrate the two so that the new directions seem like a natural extension of the projects you’ve already worked on. Sometimes this is non-obvious. You pick questions because they are technically or conceptually interesting. But if you think hard about how your future work connects with your past work, I think you’ll find that there are more commonalities than you may have realized.

2. Technical details. There’s something a bit unfamiliar and strange about writing your first grant proposal. Most of what you’ve written so far has been about things you’ve done. But the most exciting parts about a proposal are what you’re going to try to do. I think there’s a temptation and a tendency to talk about the background, state an open problem and stop there. What makes a grant proposal fun to read and all the more compelling is some concrete ideas. What tools can you bring to bear on your problem? What special cases do you plan to investigate en route to solving larger problems? Are there any papers in the literature that prove related things, whose techniques can be thought of as a blueprint? You don’t have to have a fully developed plan for all of the projects you propose, but having at least one or two that are very concrete (in this case, I think the sparse coding section is quite concrete in terms of the techniques it suggests) is great. A proposal shouldn’t just be a white paper, or a checklist of problems.

3. Broader impact. I always find these types of sections to be the hardest to write. It’s hard to think of what to say, so that there’s enough text and you don’t sound too braggy. I got a lot of value in reading other people’s proposals and seeing what they wrote, and how they phrased it. Something else to think about: Proposing a new graduate course is not great broader impact. Sure, by all means talk about it, especially if it’s something you’re excited about. But realistically, everyone will talk about planning a graduate course in their area. The more you can show having broader impact *not just at the graduate level* the better. Anything you can talk about that actually has an impact, not just for the technically elite, will really stand out and be a major plus in your application.

%d bloggers like this: