As we've already discussed on Mihai's blog a month ago, this year FOCS has a strange new procedure: a week after the real submission deadline, one has to submit a two-page brief description. Now, via Muthu, we get a little more information about what this is supposed to mean: Umesh Vazirani's proposal for including them in the submission process and several examples of brief descriptions for past FOCS/STOC papers.
While it's helpful to have greater clarity, I have to admit that I still don't see why one would want to make such a thing significantly different than the first two pages of one's paper. All of the purposes it supposedly serves, of attracting readers to the paper and providing a brief hint at the more technical material to follow, are purposes that it would be good to fulfil in one's real introduction. And from my reading the proposal, specifically the part about “reluctance of senior researchers in the field to serve on STOC/FOCS committees,” implies that these senior researchers are lazy and would only judge papers by the first two pages anyway, and explicitly encourages them to do just that. Having just been one of the more senior members of a STOC committee, I find that a little insulting.
Regardless, it's the procedure so we follow it, I suppose. Even if “follow it” means “cut off the first two pages and submit it again”. If doing it that way leads people to put more care into the first two pages of their submissions, I suppose that would be a good thing.
I see the point as being the one week delay after submission. Many people do a terrible job of writing in their FOCS submissions because they are still furiously editing and rewriting two hours before the deadline. With an extra week to work on the brief description, I hope it will be better than the introduction to the actual submitted paper. It's an anti-procrastination device, which freezes the technical content and then says you have one more week to explain and motivate it better.
That makes some sense. But the same people who procrastinate on the writing to the point where they need to furiously edit and rewrite two hours before the deadline are likely to procrastinate on doing any revisions to their introduction until the week-later deadline at which point there will be little to do but use the first two pages of the paper anyway.
I think I am a good example of someone who would write an informal description very differently than I would write the first two pages of my paper. I typically write my full paper (and post on ArXiv) before I write my conference submission. So, my conference submission looks like a redacted version of a full paper, with all the technicalities.
In contrast, I would write my brief description much more loosely, and perhaps more helpfully, as it could be very different from anything that I would ever publish.
Writing in full detail and cutting down to conference submission length if necessary is a good idea, and one I try to follow as well. The part I don't understand is, why would you want your full technical paper to avoid having this sort of motivation and loose description in its introduction?
I used to write papers with loose high level description of strategy and implications first, actual proof later. In Math, where this style of writing is commonly used, people would praise my writing. In computer science, the typical comment I got was: "why are you repeating yourself and writing such informal verbiage before the actual proof?".
After a set of jarring rejections a decade ago, I rewrote those papers in an all dry formula and they were readily accepted. I haven't written in that style since.
I also often receive negative feedback from the theoretical computer science community when I try to give some loose high level intuition of the overall proof strategy before launching into the formal proof. My feeling is that if you write this way, you take a risk. If you skip the intuition, and just give a dry definition, lemma, proof, theorem, proof, corollary write-up, there is less risk.