Monday, June 16, 2008

Posting a SUBMITTED paper

(Can still register for COMPLEXITY 2008 at here.)

In a comment on my post SODA 2009 CFP that Shiva Kintali pointed out that SODA is encouraging authors who submit to post their submissions on their own websites. He is correct that they are encouraging it, see here, but is he correct that it is an excellent idea? This raises a few questions.
  1. You are on the SODA committee. You read a paper that is very good and that you can build on. You make sure it gets rejected from SODA and then write your own paper that is similar. Obviously unethical. But since her paper is on line early she can easily prove that she obtained the results first. Does just having a paper on your website count? I would say yes, and this would be a good reason to post. However, I doubt this scenario is common.
  2. You are on the SODA committee. You read a paper that you can build on. You really want to get a legit copy of it so you can start working, fully intending to credit previous work. You go to the authors website. Its not there! Perhaps you are motivated to get it INTO SODA so that you can use it and reference it. This might be a good reason for the author to NOT post to give them an edge. However, I doubt this scenario is common.
  3. If you are on a program committee or a subreferee or refereeing a grant then you are morally obligated to not steal any work that you see in that capacity. You are also not even allowed to build on the work. But are you allowed to look at the authors website to find the work so you can then build on it (and properly credit it)? I would think so. What if the author does not want you to to this. Then she would not put the paper on her website. But now SODA is urging her to do so. Is this appropriate on the part of SODA? Shouldn't the author have the choice of perhaps not going public yet because she wants to work on the problem some more? Of course, at this point its is just urging to post on your website not mandatory. But will this urging turn into mandatory at some point?

9 comments:

  1. Re "since her paper is on line early she can easily prove that she obtained the results first": I would encourage authors who are concerned about having a provable timestamp to post the papers to a repository that provides timestamps, such as arxiv.org — I would think that posting to a personal web site is not good enough.

    And I have had a situation in the past where I and someone else submitted papers on very similar topics, mine was rejected from that conference (accepted to another later), the other was accepted to the first conference, and the other one used the dreaded phrase "we initiate the study of...". I don't remember whether I had made a preprint available in that case (sometimes I do, sometimes I don't) but it would have been of aid in that little priority dispute.

    ReplyDelete
  2. I personally don't believe in posting papers before they have officially been accepted for publication, unless the author is not planning on seeking a publication. I don't know why a conference would "encourage" publishing papers on the author's own web site, especially since most conferences require signing over the copyright to another organization (such as IEEE). Researchers, including myself, do post their publications on web sites, but if there is no organization maintaining the copyright, it is open to anyone to use as he/she pleases without limit.

    ReplyDelete
  3. Researchers, including myself, do post their publications on web sites, but if there is no organization maintaining the copyright, it is open to anyone to use as he/she pleases without limit.

    This is absolutely false under current copyright law (under the Berne Convention, which covers practically every country in the world). In particular, when you write a paper you automatically hold the copyright, with no copyright registration or the like required.

    Maybe you are making the point that it is difficult (and potentially expensive) to enforce your copyright in the courts. That's true, but it's not a big deal considering that researchers make no money off papers anyway. The real fear is losing scholarly credit, and that's best addressed by publicizing your work as much as possible under your name, not by minimizing distribution. If someone steals your work, and you can prove it is really your work, then they will be shunned by the community and you will suffer little or no harm.

    ReplyDelete
  4. I think the issue Bill was trying to point out was the timing; i.e, asking authors to post their paper upon submission, rather than upon acceptance. Once a paper is accepted, it probably will not hurt you to have it seen between acceptance and publication.

    If I submit a paper, and it is rejected, then I can, without competition, put additional effort into strengthening the result for the next conference. But if I posted the initial result publicly at the same time that I submitted it, then others may be working to strengthen it also, and if they strengthen and publish the result before I do, then they will get all the credit, even though my initial paper (which was rejected) did most of the work.

    But I don't think Bill was saying there is anything wrong with encouraging authors to post their accepted papers online; at that point, they have already "locked in a profit" from the work, and at that point, if someone else extends the result, this will help your career, rather than hurt it. You will get credit for the initial work, and your work will even be seen as more valuable because it is cited.

    ReplyDelete
  5. I have wondered about a similar issue that perhaps someone might care to comment on.

    In some areas, the time period from idea conception to actual paper is long enough (perhaps due to empirical work such as experiments) that I have often seen people provide sneak previews of results in their talks, somewhat overlapping with the review timeline of the conference or journal. That seems to provide even less protection in a priority conflict, especially if the paper is rejected for technical reasons and the eventual publication is delayed.

    Does anyone have any thoughts on what the preferred protocols are/should be?

    ReplyDelete
  6. "But if I posted the initial result publicly at the same time that I submitted it, then others may be working to strengthen it also, and if they strengthen and publish the result before I do, then they will get all the credit, even though my initial paper (which was rejected) did most of the work."

    Say what?

    Making your submission an official technical report or posting it on a preprint server like the ArXiv establishes priority for your results. (Even just posting it on your web site in advance gives a weak time stamp.)

    On the other hand, if your work can be so easily improved, why are you submitting it? You're not done! If you're not willing to publicly take credit for your work, you just shouldn't submit it.

    I would love to see conferences move to a one-line submission process: Submit the ArXiv ID number of your paper. Deadlines and page limits can still be enforced easily; full versions can be posted as revisions. Authors can submit results without worrying about losing priority. PCs can post electronic proceedings as an ArXiv overlay immediately after the decisions are made. Without the overhead of paper publication, submissions can be considered much closer to the conference date. Some mechanism would have to be made for authors who want to preserve patent or other IP rights, but those would be rare and easily-handled exceptions.

    And now we begin the endless discussion of page limits. Where's my beer?

    ReplyDelete
  7. Re "On the other hand, if your work can be so easily improved, why are you submitting it? You're not done! If you're not willing to publicly take credit for your work, you just shouldn't submit it."

    I think that's a bit too glib. Many nice results are of the sort where the essential insight may look relatively simple once presented (as opposed to the longer process of arriving at it) - so, if the paper is rejected and someone else benefits from the insights, perhaps even adapting it slightly to avoid an exact duplication, then the original author loses out. Of course, I do admit that the arXiv solution seems sensible.

    In the end, can there ever be a good solution to the priority issue? Even when people don't post their papers ahead of acceptance, there tend to be battles. A famous recent case was a complaint by E.C.G. Sudarshan who argues that he has been cheated out of the Nobel Prize twice - the second time for a reason not unlike the priority battles being discussed here.

    ReplyDelete
  8. Once a researcher has proved a result, publicizing it fast enables research to move forward. For the field as a whole, it is better that results get early exposure.

    Posting a paper as soon as it is submitted may create issues of ownership of ideas and results, but that's secondary. It does help give other people a more timely access to new ideas.

    ReplyDelete
  9. But since her paper is on line early she can easily prove that she obtained the results first.
    ====================================
    james

    Link Building

    ReplyDelete