Tuesday, March 31, 2009

Submit and Forget

The FOCS deadline is fast approaching, this Thursday at 6:59 PM Eastern. 

I have a tradition of submit and forget. After I submit a paper to a conference I put it aside and don't think about the paper again until we get back the decision. In recent years I have broken that strict rule by sending the submission to an archive site like ECCC but in general I don't work on or even think about a paper while the program committee does its thing. No use fretting about the paper while the committee decides so best to keep busy with other research and activities.

No such luck for this year's FOCS. Umesh explains why we have to write a 2-page follow-up a week later. Lots of bloggers have weighed in, check out Jeff and everyone he links to. My main take from Umesh's letter: Authors cannot be trusted to give the proper motivations, explanations and examples in the initial write-up. And that's a shame.

The 2-pager makes bad writing self-fulfilling. Why bother with proper motivation at all in the first stage if you get to work on it later?

6 comments:

  1. Authors cannot be trusted to give the proper motivations, explanations and examples in the initial write-up.

    I think it has more to do with the 10 page
    limit and the added onus on the author to provide full proofs. The latter in my opinion is a good thing. Unfortunately, this also means that there is lack of space for informal discussions, motivations and miscellaneous other fluff. The extra two pages can serve as a repository of this excess stuff which might enable a non-specialist to appreciate the paper better, but which would not make it past the referees when the paper is finally submitted for journal publication.

    I also agree with a suggestion made by Jeff Erickson in his blog to publish only these two pages in the proceedings. Doing so would force all authors to submit papers to archival journals as in most scientific disciplines and make FOCS/STOC more of a venue for presenting results and ideas before an audience which was the original intent (rather than a publishing venue of ones papers). As an added advantage, since the "number" of conference talks would hardly be considered as an important data point by search/tenure committees etc., this would substantially reduce the unhealthy competitive streak that mars these conferences in recent times.

    ReplyDelete
  2. I think it has more to do with the 10 page limit and the added onus on the author to provide full proofs ... this means that there is lack of space for informal discussions...

    There is no page limit on the appendix. You can write full proofs, put them in the appendix, then write what you want in the 10 pages.

    ReplyDelete
  3. I have a tradition of submit and forget.

    I wish I could follow this. More often than not, I find myself fretting that I could have improved some aspect of the paper and that the program committee is going to reject my paper because of that :-)

    ReplyDelete
  4. Submit and forget is a fine strategy for conference papers. It works less well for journals: sometimes the editor also has a tendency to forget your submission...

    ReplyDelete
  5. The 2-pager makes bad writing self-fulfilling. Why bother with proper motivation at all in the first stage if you get to work on it later?

    Note that the two versions are supposed to be independent, so neglecting the introduction in the main submission would be a bad idea.

    ReplyDelete
  6. The perceived need for this change reflects a change in the reward system for conference submissions. Prior to the extensive sub-refereeing that is possible with electronic versions of papers and the sometimes detailed written reviews that are now expected, each submitted paper got many fewer minutes of reviewing time since the number of papers per PC member has roughly remained constant over time.

    Before these changes, a lot of salesmanship in the first couple of pages was essential. If you didn't grab the PC member's attention in those 2 pages, you could not expect the rest of your paper even to be read. Most of the rhetorical exclamation marks would be smoothed away in the final paper but much of the hype came through. (I recall part of a FOCS business meeting in the mid-90's decrying this hype.)

    We've gone to the other extreme in that many authors not only don't put in the hype, they don't even ease PC members into reading their papers, since they are reasonably confident that some expert in, or close to, their area will be asked to look at their paper.

    I would have liked this to be optional - namely, that you could either submit something separate or stand pat and have the first two post-abstract pages of your original submission be used automatically instead. Logistically this wasn't possible, apparently.

    ReplyDelete