Friday, November 9, 2007

Branding and STOC/FOCS

Kamal Jain gives some advice on resubmitting papers that got rejected from STOC/FOCS. That reminds me of a question I've had for while: can you build up a name by only submitting papers in the right places (i.e. where they get accepted)?

Personally, I feel very strongly about self-selection. In my (short) career, I have had a total of 2 rejects:

  • a paper rejected from ICALP, which got accepted to the subsequent SODA ;)
  • a "short paper" rejected from SODA, which essentially disappeared (we came up with new ideas and obtained much better results). Incidentally, I'm glad the short track at SODA was stopped. It was a temptation to do mediocre work -- if you write a really good result in 2 pages, you will always get accepted even without the short track.
I self-select due to my own ethical beliefs, independent of the environment in the community. However, I'm often told that this is not career-optimal from a selfish perspective: if you shoot for a reasonable reject rate, say under 50%, you can often push papers up a notch, making for a better CV.

Is this advice short-sighted? Does having a low reject rate help you in getting more attention from reviewers? If a reviewer knows you're not a "deceiver", he will feel uneasy about rejecting, and will think about it for another minute -- at the very least, he knows one person in the world (you, the author) finds the paper interesting. If a reveiwer has already rejected a couple of your papers, he won't have so many moral problems -- after all, it's possible that you yourself don't think the paper is good enough.

What do you think? Can you get reviewer sympathy by active self-selection?


Anonymous said...

My feeling is that if you're not getting rejections, you're not trying hard enough to make your work visible to others by sending it to the best conferences it can get into. My own rejection rate is, I would guess, around 33%, which I'm quite comfortable with. I tend to submit a lot more to SODA and SoCG, and to less highly ranked conferences such as Graph Drawing, than to STOC and FOCS, but that has nothing to do with self-selection based on probability of acceptance and everything to do with how much attention the other people who are most likely to be interested in my work pay to those conferences.

For less selfish reasons, sending papers that might be rejected is good, too: outsiders to the field, not knowing its standards, judge us in part by looking at acceptance rates of conferences. We can educate them to use less superficial criteria, of course, but we can also help make our conferences look good to them by sending borderline papers that can be rejected.

Anonymous said...

Getting papers into better conferences than they deserve to be in helps only up to a point. It will impress people outside your area (which helps for getting hired), but people in the area know which papers are good and which are not. Researchers can also develop a reputation for pushing weak papers, or (on the other hand) for letting their results speak for themselves (regardless of where they get published).

Anonymous said...

The problem is that very few results speak for themselves.

I doubt if there is *one* paper every year that is considered important by a large fraction of the community.