Wednesday, April 27, 2011

Don't Blame the Tech

Short Announcements: STOC Poster Submission Deadline is Monday. Jaime Morgenstern set up a google groups page for students to find roommates for STOC, FCRC and other theory conferences, "particularly useful for women and students at smaller departments."

I usually agree with Moshe Vardi when he writes about conferences but not so much on his latest CACM editorial Technology Has Social Consequences. Moshe blames electronic program committee software for declining ethics of program committee members, increased PC service and more farming out papers to junior reviewers.

Moshe has a rosy memory of the past. I knew more than one PC member back in the day that would work on a paper submitted to a conference. The slowness of paper dissemination back then hid it better. Junior people and students always served as subreviewers, one professor even tried to run a course on reviewing FOCS papers.

What's changed is that computer science has grown and specialized. We have more conferences and each conference needs an ever larger PC to cover all the specialized subfields. Program committees have become unwieldy with ethics and quality reviews harder to enforce. These problems won't be avoided whether having an electronic or physical meeting. A strong PC chair makes more of a difference than the venue of a meeting.


  1. One problem I have with the review process (and this applies equally to conferences and journals) is that there is no 'punishment' for sub-standard reviews, or for PIs failing to proof and correct the reviews made by their students (which are often very poor). We desperately need a carrot-stick reward structure for reviewers. The pressure to publish and the dearth of scientists qualified to evaluate the glut of papers being produced will only lead to an exacerbation of the current problem.

  2. Isn't arxiv one way of protecting one's conference (and journal) submissions against this sort of unethical reviewer behavior?

  3. One problem I've had as a reviewer is the following: I get a paper submitted to a (good) journal that has already been published in a (good) conference. The paper is written by a student and his/her famous advisor. The paper is terribly written, bad notation, overly complicated proofs, etc. I want to write a thorough review. However, why should I spend more time on the paper than the famous advisor? In these cases, usually the famous person has a good enough reputation that I know they know better than to submit such a poorly written paper, but just don't care and are happy to stick their name on whatever their student writes (they could have been actually involved in the research--I'm not making a judgement about this). Writing a paper also takes time, and there is a trend in our community to publish as fast as possible and move onto the next project before spending adequate time writing up the previous results. Then the reviewer has to do the thankless job that really the authors should be doing, and if they don't, perhaps this is one case in which the reviewer is considered to have done a poor job, when the burden is really on the authors themselves.

  4. Anon- when I get a paper like that I write
    a QUICK negative report saying

    ``Rejected because it is badly written.
    The results may be good, or not, but this
    cannot be determined. If they fix it up and resubmit I would be happy to look at it again. I am not going to bother listing what is wrong with it since even a cursory reading by all of the authors
    will reveal it. Then have it proofread by a non-author. If the resulting resubmission is still badly written then I will recommend the paper be rejected without invitation to resubmit.''

    I honestly believe that the referee system
    is truly anonymous so there is no
    danger of getting someone mad at you.
    However, this (alas) is also a negative in view of Mihai Pop's comments.

  5. Wouldn't it make sense to ensure that when a paper is published, both the paper and the reviews are made public, including the reviewer?

  6. Short question: Does any of the four conference hotels at FCRC/STOC offer free internet (when reserving a room with the ACM group code)?

  7. BTW, is there any news from the job market. It is already the end of the season.

    Mark Braverman--> Princeton

  8. David Steurer -> Cornell

  9. Anonymous #5 :

    I could agree on publishing the review, but not the name of the reviewer.

  10. This comment has been removed by the author.

  11. Mohammad Hossein BATENI--> Google Research

  12. Here are some answers to questions raised earlier:
    "Short question: Does any of the four conference hotels at FCRC/STOC offer free internet"

    The Fairmont offers free internet *if* you join their "President's Club", which costs nothing to join.

    "is there any news from the job market. It is already the end of the season."

    Swastik Kopparty -> Rutgers

    Shubhangi Saraf -> Rutgers

    David Cash -> Rutgers