Thursday, January 15, 2009

Amitabh Varshney (Graphics Faculty at UMCP) emailing me the following:
Someone told me a while ago that there are some theory conferences in which members of the program committee are not allowed to submit papers. Is this really true? Does this still happen?
My first impulse was that theory conferences largely do not allow PC mems to submit (I knew that STOC, FOCS, COMPLEXITY did not, and that COLT did). By asking people (I could not find this info on websites) I have compiled the following list. Corrections and additions are welcome. PC stands for program committee.
  1. Do not allow PC mems to submit: STOC, FOCS, ICALP, MFCS, COMPLEXITY, SODA, RANDOM, APPROX, SoCG
  2. Up to the PC: LICS
  3. Allow it but there are various restrictions: CRYPTO, EUROCRYPT, RECOMB (e.g., only two such submissions per person, or some complicated thing having to do with if students are co-authors. The exact rules change from year to year.)
  4. Allow it with no restrictions: COLT, ALT, ISMB (biocomp), WABI (biocomp) EC (electronic Commerce)
  5. My friends outside of theory tell me that allowing submissions from PC mems is the norm. Note above quote from Amitabh Varshney.
Now some RANDOM comments on this. I can submit the comments to RANDOM unless I'm on the PC.
  1. Reason to allow it: if it is not allowed then people might decline to be on the PC.
  2. Reason to not allow it: avoid conflict of interest and the appearance of conflict of interest.
  3. For those that do allow it, the person who submitted is not allowed to be involved in the discussion. This works pretty well especially with meeting over-the-web.
  4. From what I've seen and heard there is not a problem with the PC mems having an advantage. In fact, RECOMB explicitly says that a PC mems paper has to meet a higher standard. For other conferences this has been the de facto rule.
  5. When COLT began the area of Learning Theory was small so they allowed PC mems to submit, else there would be too small a pool of people to submit. This is no longer true, but the rules live on.
  6. The first few years of COMPLEXITY (then called STRUCTURES) the PC mems were allowed to give invited talks. This struck me as being a good way to reward the PC mems. But if they all want to do it, that is too many invited talks.
  7. In COMPLEXITY this issue has not been revisited- it is not brought up at business meetings. According to Jeff Erickson's SODA BUSINESS MEETING BINGO it does come up at SODA business meetings. I've heard it also come up at SoCG business meetings.
  8. If you let PC mems submit they you can have a large PC without depleting the pool of people who can submit.
  9. I have no strong opinion on this issue. I'll leave that to the comments.


  1. Correction: RANDOM does allow PC submissions (at least, RANDOM 08 did).

  2. Do these conferences that disallow PC submissions make a difference between a PC member that submits a paper as the only author, or a PC member that is n-th co-author on a paper his/her grad student wrote?

    If not, I would oppose such a constraint, since it effectively disadvantages others - namely, grad students whose supervisor unfortunately is on the PC - to submit to a conference of their liking. I think this is not fair.

  3. I may be wrong, but I believe that LICS has disallowed submissions by PC members for a while.

    Do these conferences that disallow PC submissions make a difference between a PC member that submits a paper as the only author, or a PC member that is n-th co-author on a paper his/her grad student wrote?

    Those I am familiar with don't make such a distinction.

  4. Personally, I see no big problem with allowing PC members to submit if it is handled the right way. But it is interesting that different areas of CS seem to have come up with different answers to this question, I guess mainly due to different circumstances. For example, compare theory and databases:

    - DB conferences have huge committee (80+ members), theory conferences much smaller ones.

    - large DB conferences do get more submissions than STOC or FOCS, but this does not alone explain the committee sizes! Each DB committee member reads much fewer papers but serves on many more committees (esp. before tenure). One can discuss if larger committees result in better or worse decisions - not clear to me.

    - purely student-authored papers are more common in theory, while in DB it is sort of customary to have the advisor on every paper.

    So if you run things as in theory, disallowing submissions by PC members is not a big limitation on either committee members or their students. But you can't really look at the question without also looking at committee sizes and coauthoring customs.

  5. LATIN and WADS allow PC submissions.

  6. The last time I remember this issue being seriously discussed at SOCG was in relation to the (now defunct) applied track, back in the mid-90s. Since we were inviting people from communities where PC submissions were the norm, into a community where PC submissions were strictly forbidden, there was some initial confusion/disagreement. So in 1996, PC members on the theory track assumed they could not submit (and didn't), but PC members on the applied track assumed they could (and did). We eventually decided to stick with "the standard theory process".

    The idea has been floated in passing, usually in discussions about heavy committee loads. (This year, for example, each PC member is responsible for 35-40 papers.) But, as far as I can recall, the idea never gets any traction.

    And despite its appearance on my Bingo card, I've only heard the suggestion to allow PC submissions at one SODA business meeting. It was quickly shouted down.

    To Anon #2: No. No submissions from PC members means no submissions where even one coauthor is a PC member. There's no such thing as an "nth coauthor"; with very few exceptions, authors of theory papers are listed alphabetically.

  7. I have served in a number of PCs where we used EasyChair for the PC meeting, and there it is impossible for a PC member to even see the discussion about his/her paper.
    In that sense, a PC member is really like any other author and only receives the outcome, after the final decisions are made.

    Thus, I see no real reason for disallowing submission by PC members, given the current technology.

  8. We eventually decided to stick with "the standard theory process".

    This reminds me of the quip that there is nothing more conservative than the academic practices of otherwise generally progressive professors.

    This is the only group of people I know who can readily posit extending human rights to whales and chimps, but consider parallel tracks and blind submissions too risky an experiment to be tried.

  9. As a vocal person, I'll simply state my opinion I don't really see a problem with PC members submitting. It just means people leave the room (or don't get online access to the PC discussion) when their paper is discussed.

    I've heard the argument that it "gives the appearance" of "possible impropriety", but I just don't buy it. Odds are, if your community can't handle having PC members submit papers to the conference without unfairness, there's bigger problems in your community.