The past several weeks I have been almost continually working toward one deadline or another, always getting things done within 24 hours (though in some cases, 24 minutes) of the deadline.
I think there is general agreement that the TCS culture is deadline driven, and I generally hear people say (even while complaining) that this is a good thing because it forces people to get things done that would otherwise be delayed indefinitely.
I disagree with this on two points.
- First is the implicit assumption that “things will not get done” without some external deadline. I don’t think that’s true. Everyone has their own (soft) deadlines: students want to propose and graduate; professors want to get tenure and report results to the funding agencies at the end of the year. Moreover, it seems to me that when someone is interested in a result the paper gets written; it is mainly results that even the authors themselves are not excited about that drag on forever.
- Second is the implication that it is a good to have some external motivation “forcing” papers to get written. Here is where I strongly disagree. The net effect of deadlines is that, overall, papers submitted to conference are half-baked and poorly written; in many cases both the authors and the reviewers (not to mention the eventual readers of the paper) would be better served if the authors were “allowed” a few more days to polish their work, or even a few more weeks to work out extensions or generalizations of their results. (Of course, one can hope that the authors continue to revise their paper after submission, but usually it is just on the the next deadline.)
Is there a solution to this? Let me first note that the physical sciences do not work the way we do: journals come out weekly, and one can submit a paper at any time once it is ready. (This is not to say there is no pressure to publish results quickly, only that there is not an arbitrary cut-off time for submission.) For better or worse, though, that model is not going to work for TCS.
One thing we could do, though I’m not sure how exactly it would play out, is to be more lenient when it comes to deadlines. For example, why couldn’t we allow authors to continue to revise their papers after the deadline (though with no guarantee that the reviewer will read anything but the originally submitted version)? Given that most reviews aren’t done until at least a month after the submission deadline, this seems like it wouldn’t be too difficult to implement. (And even if a reviewer read the original version of a paper, he might want to read [limited sections of] a revised version if aspects of the proof were clarified.) Seems to me this would help authors as well as reviewers.
Another idea that would seem to help (though it is not precisely related to the deadline issue discussed above) is to allow a paper to be considered for more than one conference; for example, a paper that was submitted to (say) TCC could be considered for (say) PKC if it were rejected from TCC. (Reviews could be forwarded so as to minimize the extra work on the second PC considering the paper.) This would be a huge benefit to authors, who now sometimes have to wait several months to have a paper published just because it got rejected from the first conference it was submitted to.