Flip your NSF Rejection

I spent the last few days of March in 2015 refreshing TheGradCafe and my email inbox. I get push notifications for my email, so I’m not sure why I bothered refreshing it. I guess I was hoping it’d make things go faster. It didn’t.

One night, FastLane went down for maintenance. Posters on TheGradCafe noted that this had happened last year shortly before notifications had come out. I stayed up and waited for a few hours, but nothing happened, so I went to sleep. When I woke up, I saw this:

2015 NSF Graduate Research Fellowship Program Application Notification

Please be an award, please be an award, please be a-

Your application for the 2015 National Science Foundation (NSF) Graduate Research Fellowship Program has been evaluated. We regret to inform you that you were not selected to receive a fellowship in this year’s competition.

I’d gone into the application cycle pretty optimistic. Maybe I just wasn’t cut out for this research thing.

I got my first taste of academic rejection. Mostly reasonable and constructive academic rejection, but back then, it read like this:

This applicant has a very good academic record and rich relevant work experiences which are helpful for the proposed research. But all reference letters are from the same institution. … Also this application hasn’t any publication.

The proposed is an interesting idea but is needing more development. The questions seem a little naive. It is a research direction worth pursuing.

The research proposal is a bit rough and would be improved by better placing within the context of more existing work. While the letter writers are enthusiastic about the student, they collectively do not add many comments about their support for her specific research plan.

Just kidding, it actually read like this:




On TheGradCafe, the other rejected folk were posting their reviews. E / E / VG / VG / VG / VG. E / E / E / VG / E / VG. E / E / VG / VG / G / G. And there was a double poor just staring at me. Clearly, grad school wasn’t my thing.

This year, I got the NSF. There was no magical transformation in between the two years. I worked just as hard both years. But this year, I did better work. For those of you looking to flip your computer science NSF GRFP application, I’d like to share some advice.

Generate Lots of Bad Ideas

I remember emailing my undergrad professor the first time I applied and asking how people think of ideas. We chatted a bit and he sent me some papers. I didn’t really understand any of them. I thought of something on my own that sounded kind of interesting, chatted with a friend, and decided to take that direction because it sounded cool and I didn’t feel like generating more ideas.

I had one idea. It wasn’t really a great idea. Most of my ideas are bad. Most of most people’s ideas are bad. But bad ideas aren’t actually bad! Bad ideas are great. They’re ideas. Don’t filter them out too early. In your long list of ideas you think suck, there’s probably one that’s actually good. So write down the whole list.

This of course doesn’t help much if your list is empty. And it’s totally fine if that’s true. So I’m going to take this classic advice a step further and share some ways to actually think of ideas.

How to Generate Ideas

If you’re in industry, look around you for problems. There are plenty of problems. There are plenty of things that are unnecessarily hard. Builds take forever and always break. Developers spend as much time writing unit tests as code. Then they wait 24 hours for code reviews. Bugs regress. Developers get paged. Everything is broken, always.

Every time you find something that’s hard that you think would be easier in an ideal world, write it down. Send it to yourself in an email. Whatever. Anything to keep a list running. Later on, think about ways to solve these problems with computer science. These are research problems!

If you’re in academia, read current work. Go to talks. Talk to people about their ideas. Talk to people about the almost-ideas you have that sound really stupid that you don’t think should actually be ideas. Eat lunch with your research group. Eat lunch with other research groups. Absorb as much as you can.

Write down all of the questions you have after reading papers or attending talks. Ask those questions. Write down the answers. Ask questions about those answers. Always ask questions. I personally hate asking questions during a talk, but find that most people are willing to talk about the same questions either immediately after the talk or later in the day. Look for existing work that tackles those questions, read that work, and then (you guessed it) ask more questions about that work, too.

Oh, and all of this still applies if you’re an undergrad. Grad students don’t bite. We love undergrads who are interested in our research areas. They’re free labor impressionable get ’em hooked while they’re young! bright young minds.

Why not mix the two? If you’re in industry, hang out with your local university’s research group. Attend their talks. Show up to their reading groups. This might feel weird, but you’re probably welcome. If you’re in academia, talk to your friends in industry (make some if you don’t have any) about the problems they have. You’ll probably end up with something high in both broader impacts and intellectual merit.

Get Peer Feedback

Generating ideas is the fun part. You don’t have to worry about how stupid your ideas actually are, because they’re allowed to be stupid. You just have to have them. The scary part is actually pursuing an idea. If you think the idea is dumb, it feels awful to bounce it off of your peers, let alone your professors.

But here’s the secret: Nobody is judging you. Nobody’s going to go, “oh man, did you hear Talia’s idea today? It was so dumb.” People are mostly reasonable.

Talk to your peers about your ideas. Talk to your professors about your ideas. Go back to your peers and talk to them about your ideas some more. Then go back to your professors. Write a really shitty first draft. Bounce it off of everyone who’s willing to read it. Get feedback. Refine your draft. Get more feedback. Send it to your professors. Get more feedback. Exchange essays with other people from your year who are applying. Exchange feedback.

But Talia, what if my peers give me negative feedback?

Better them than your reviewers. Don’t turn in a draft.

My first year, I was too scared to get any feedback ahead of time, so I sent in what I had. This year, I got feedback from several people in my lab, including two students who had won the NSF the year before. Some of it was really bad! But nobody decided I was stupid because the first draft of my research proposal wasn’t good.

Two students in my lab won the NSF this year; the two of us had exchanged essays with each other and given each other feedback. I believe that one of the reasons UW is so successful at winning NSF fellowships is that feedback among peers is highly encouraged.

Don’t be Scared

Once you think an idea is good, there’s a terrifying moment when you search through existing work and you’re like oh god, what if someone already had this idea that I really love, and they’ve already done it better than I ever could? 

Well, what are your options?

If you don’t look for existing work, one of your reviewers will know about it already. It will show up in your reviews. Ignorance doesn’t help.

If you find existing work that does exactly what you want, that’s great! That means that you had an idea that landed a publication. It was a good idea. Read the papers. Ask questions about the papers. Maybe you can do it better, or maybe you can find a new idea from those questions.

If you only find related work, you should still read all of those papers and mention them. They’ll help you refine your idea and make a good case that you’ve actually put a lot of thought into your proposal.

Looks like there’s only one way to lose.

Write for the Busy Reader

Your reviewer is probably juggling three commitments on the job and also has a family and friends and a life outside of reviewing NSF applications. So when you’re writing you should strategically use bold and italics. Get the reviewers to notice exactly what you want them to notice.

Tell your reader as clearly as possible how you’re addressing broader impacts and intellectual merit

My first year, I wrote a really nice personal statement that read beautifully if you had like thirty minutes to sit down and actually read it. Nobody did. I was profoundly surprised when my reviewers missed the broader impacts in my personal statement. This year, I added some nice headers: MENTORSHIP – BROADER IMPACTSTEACHING – BROADER IMPACTSRESEARCH – INTELLECTUAL MERIT. Nobody missed it this time.

Tell a Consistent Story

One of my professors emailed me the first year and asked me to send my materials so that he could give me a recommendation that was consistent with my story. I was too scared to actually do this. I think at first I gave him a high-level overview, then maybe eventually I caved and sent it but I think I sent it with a disclaimer that was longer than my actual research proposal. Whoops.

I don’t know why, but for some reason I thought that the professor who had helped me get that far would suddenly decide I wasn’t worth it anymore after reading my bad proposal. This was silly. It bit me in my reviews:

The research proposal seems to be in an early stage would benefit from being better placed within the context of existing work. The application could be strengthened by having a consistent message between the student and advisor. The collective letters of support do not provide strong support of the specific research goals. A more comprehensive package that reinforces the intellectual merit and shows more faculty support on the research contributions would make this proposal more competitive with others.

Even if my proposal had sucked, if I’d at least kept my professors on the same page, I’d have had a much stronger application!

This year, I sent my full application materials to everyone as soon as they were in reasonable shape. Then I sent the final materials before I submitted them. My story was consistent. Reviewers noticed. Your professors aren’t psychic.

Learn from your Mistakes

Which brings me to my final point. You’re going to get rejected a lot in academia. That’s what everyone tells me, at least. It’s good practice. Sometimes you’ll get rejected for stupid reasons. But most of the time you can learn from it.

The NSF does you a huge favor by releasing the reviews to you. Some of them probably aren’t useful, but I’m willing to bet there’s some useful information in there. You can sit there and beat yourself up over negative comments, or you can honestly look at them and figure out how you can strengthen your application next year.

Or, if you’re like me, you can beat yourself up a little, save them to a document, buy a pint of ice cream, eat it, and come back a few days later with a fresh mind. Then you can read them and figure out exactly how you’re going to win the NSF next year.

Go get it!