Monday, June 21, 2021

Collusion Rings

Grigory Yaroslavtsev now maintains the Theory Jobs Spreadsheet, a crowd-sourced listing of where your favorite CS theorists have taken new positions. Read his blog post for more info.

Michael Littman wrote a CACM Viewpoint on collusion rings designed to tilt the odds into getting papers into conferences. See also this post by T. N. Vijaykumar. I started to write a post on the topic but couldn't get it going so I asked the AI algorithm GPT-3 to write a guest post. Here is its story.


I have been part of a collusion ring and I know how it feels. I was a postdoc at the University of California, Berkeley, in the early 2000s when I first learned about the practice of “paper rings.” A paper ring is a group of researchers who agree to submit each other’s papers to conferences without disclosing this fact to the conference organizers. The idea is that if you have several submissions from the same conference, your chances of getting accepted are higher. The pressure to be part of such a ring can be immense. You are told that you need to publish or perish and that you need to get papers into top conferences. If you don’t play along, you will be left behind. I was lucky enough to have a mentor who warned me against being part of such rings. He told me that if I did not want my name on a paper, then I should not submit it to a conference. He also told me that if I did not want to be part of a paper ring, then I should not submit papers for others in the ring. He said that he would help me with my own research and that he would help me get my own papers into conferences. He was right: My own papers were accepted into conferences and they were published in journals.

I am now an associate professor at the University of Massachusetts Amherst and I have been on the other side of this issue as well. In 2012, I was asked by an editor at a top computer science conference to review a paper submitted by one of my students. The student had been part of a paper ring and had submitted his own paper along with those submitted by his collaborators in the ring. The editor told me that he had rejected the student’s paper because it was too similar to other papers submitted by members of the ring. However, he wanted me to review it so that he could reconsider accepting it for publication if I recommended doing so.

What should I do? Should I tell him about the paper ring? Should I recommend accepting the paper? Should I tell him that there is no way he should accept it? Should I tell him that he should reject it but then publish it anyway? What would you do? I decided not to tell him about the paper ring and not to recommend accepting the paper for publication. Instead, I told him that he should reject it but then publish it anyway because this would send a strong message about what we expect from our students and what we think about such practices. The editor agreed with my decision and published the student’s paper along with two other papers submitted by members of the same paper ring.

The student received his PhD degree but did not get any credit for his work on these three papers because they were considered as part of his dissertation work when he defended his thesis later that year. In retrospect, I think that my decision was wrong because it allowed these students to continue their unethical behavior without any consequences for them or their advisors (who were also part of this ring). But at least in this case, there were no negative consequences for anyone involved in this incident except for these students who did not get credit for their work on these three papers because they were considered as part of their dissertation work when they defended their thesis later that year.

7 comments:

  1. Did you use some particular site to generate this? Seems fun to play around with.

    ReplyDelete
    Replies
    1. I got a beta account at Open AI. It is indeed fun to play with.

      Delete
  2. How can you reject it but then publish it anyway???

    ReplyDelete
  3. Nice post, Lance! The quality is scarily good and threw me off for a surprisingly long time.

    ReplyDelete
  4. I enjoyed the algorithmic flavor to it!
    Not as crisp and succinct as your writing style,
    so there's still much to be improved, but nice
    demo.

    ReplyDelete
  5. What prompt did you use? Also, I find this somewhat in poor taste, given that a suicide was involved.

    ReplyDelete
    Replies
    1. I don't have the exact prompt, it was basically the sentence "Michael Littman wrote a CACM Viewpoint on collusion rings designed to tilt the odds into getting papers into conferences."

      And yes, in retrospect this wasn't the right topic for such an AI-generated post.

      Delete