2

I wrote a paper (P1) which went through peer review and got published. The paper introduced a new graphical user interface (GUI) primitive.

I then wrote P2, which extends P1, turning it into a new class of UI, which had no analogs at the time of writing (and maybe still doesn't, but the window of opportunity is shrinking).

P2 is not just a thin layer on top of P1, it comes with substantial enhancements that are the result of our "lessons learned from P1" analysis + new ideas.

P2 stands on its own, and it refers to P1 as if it were written by other people, per conference requirements. P2 also provides a short intro about P1, such that someone unfamiliar with P1 can still get the picture. It also provides comparisons to analogs of P1, explaining how our approach differs and arguing that our approach is better.

The set of P1 analogs consists of ~5 papers. One is a preprint, 3 are just providing their designs but not including any usability evaluation details at all, or deferring that to future work. This leaves us with one good alternative, which we were not aware of at the time work on P2 started. However, we did refer to this alternative in our related work section and compared it to our own product (there are some ideological differences in our approaches).

P2 explains that we chose to base our work on P1, because at the time of writing P1 was the only artifact of its kind that has gone through formal usability tests and provides detailed results of their evaluation process. Of course, the fundamental underlying reason is that P2 is the continuation of our previous work on P1.

P2 has been submitted for peer review, and a reviewer rejected it, one of the reasons being "it only considers one paper [referring to P1] containing a very similar design".

What is the right way to explain why P2 is built on top of P1 without compromising anonymity?

p.s. note that the reviewer in question seems to have forgotten about the section where we compare P1 to alternatives. It could be that that section is not convincing enough, not well-written - but in that case the reviewer would have said "it is not convincing enough" rather than "only considers one paper". The review is otherwise thorough, I cannot say that I have the impression the reviewer hasn't read the paper fully.

ralien
  • 145
  • 4
  • 5
    I think the reviewer is suggesting that you should either cite more P1-like papers or that the step from P1 to P2 is too small to warrant a separate publication. Which of the two (or maybe both) may be more clear from the context, hard to judge based on one sentence. Your suggestion to state something like: "It is only based on P1 because we are the authors of P1" probably does not address the reviewer's concerns. – Louic Jul 29 '20 at 10:01
  • I think you are going in the wrong direction, by trying to hide that you are the author of P1. I think you should have openly acknowledged that you are continuing your prior work and just remove your name from the reference (possibly providing a nameless copy as supplementary material). – mlk Jul 29 '20 at 10:08
  • 1
    @mlk That's a delicate suggestion, especially since the reviewers might be aware of P1 and it's author -- I think there's a chance that this strategy could lead to a desk rejection. – lighthouse keeper Jul 29 '20 at 10:33
  • 2
    @ralien The reviewer's concerns are (1) you only considered one prior paper, (2) the design from P1 is too similar to P2. Pointing out that you authored P1 does not remove any of these concerns. For (1), you either need to include more papers or better explain why one paper is sufficient. For (2), you need to better explain how P2 improves on P1. There's also a chance that P2 is in fact not a significant enough improvement of P1, and you cannot publish it as is. – lighthouse keeper Jul 29 '20 at 10:37
  • @mlk - the conference rules explicitly say authors should cite their prior work this way. – ralien Jul 29 '20 at 11:13
  • @Louic, the problem is that it is a relatively new area of research. The set of P1 analogs has ~4 items, of which only 1 (besides P1) has conducted a user validation study and published detailed results (the others were just "here's our artifact, user validation is a future step"). This analog didn't exist at the time we started work on P2. – ralien Jul 29 '20 at 11:18
  • @ralien then that is what you can tell the reviewer. – Louic Jul 29 '20 at 11:55
  • A lot of these comments should be answers. But write the paper the way you think it should be written and let the editor worry about blinding. That is their job, not yours. But @lighthousekeeper, seems to have something valuable to say here. – Buffy Jul 29 '20 at 12:42
  • @Buffy Thanks! I guess that instead of answering the question, I'm questioning the premise, hence the comment. – lighthouse keeper Jul 29 '20 at 13:09
  • 1
    @lighthousekeeper, I recognize you didn't answer, but I have the same concerns and wanted to emphasize them to the OP. – Buffy Jul 29 '20 at 14:00
  • @lighthousekeeper, you raise valid points, however I believe that is not the case. I updated the text of the question accordingly. The pool of other papers to consider boils down to one candidate, which was covered in our text. The delta between P1 and P2 is non-trivial, it wouldn't stand on its own otherwise. – ralien Jul 29 '20 at 22:25
  • @ralien if it's covered in the text maybe it's to easy to overlook. State at a prominent location, maybe in the introduction, that only two relevant papers exist. Regarding the non-trivial delta, the reviewer seems to have a different impression than you. In both cases I still don't see that it's an issue with blinding. – lighthouse keeper Jul 30 '20 at 05:39

0 Answers0