8

Recently, I have reviewed an article for the IEEE (a very good article too). Within the article, the authors place great emphasis on a particular algorithm they developed and used, without disclosing any actual code - but explained very well that it could be replicated (which I won't do)..

Now, after I have provided an anonymous review, it occurred to me that I could further test their algorithm in my similar field of research (giving due credit of course).

How would one go about asking for details of an algorithm developed in an article in review?

2 Answers2

3

In a similar situation, I have sent an email to the editor asking to pass on a message and giving permission to reveal my name and contact details to the authors if they wish to respond.

Alternatively, but more problematic, you state that you may be able to replicate it. You could go ahead and do that and by the time your work is complete, the original article is likely to be available to be cited. If you were to go down this path, though, you should still be discussing this with the editor as part of your agreement as a reviewer is that you don't do this.

JenB
  • 6,357
  • 19
  • 31
  • Good advice - I am definitely not going to replicate the algorithm, for the reason you stated (disallowed), but also it would not be ethical. –  Oct 24 '15 at 13:08
  • "as part of your agreement as a reviewer is that you don't do this" - this is field-specific. In my CS subfield, the common understanding seems to be that the very reason why everyone should review every now and then (beside the idealistic, and thus unfortunately not very strong reason when lots of projects ask for your resources, reviewing being a community service) is precisely because one gets preliminary access to upcoming publications and thus can start building upon them particularly early. – O. R. Mapper Oct 24 '15 at 13:39
  • @O.R.Mapper In my CS subfield, the common understanding...can start building upon them particularly early — That may be the common understanding in your subfield of CS, but it strikes me as completely unethical. It certainly isn't the common understanding in my subfield of CS. – JeffE Oct 25 '15 at 16:31
  • @JeffE: I fail to see what is unethical about building upon published work and citing it properly. – O. R. Mapper Oct 25 '15 at 16:54
  • 3
    @O.R.Mapper Once the work is actually published, sure. But you're talking about exploiting your access to the work as a reviewer, which by definition happens before the work is published; exploiting that early confidential access is what I find unethical. A similar argument applies to grant proposals; the US National Science Foundation explicitly forbids reviewers from using or building on unpublished work described in proposals. – JeffE Oct 25 '15 at 20:39
  • @JeffE: I can see nothing unethical about this. The earlier the authors are cited after publication of their work, the better for them. Delays from publication to follow-up works are already long enough as it is. I see no need to artificially stall things even more, and I would rather argue it is unethical to disregard a somewhat relevant upcoming publication that would be suitable for citation in your current stage of research just because the only way you know of the upcoming publication is due to your involvement as a referee. Grant proposals are somewhat different as they can be ... – O. R. Mapper Oct 25 '15 at 21:16
  • ... considered to reflect the state of ongoing negotiations, where you could benefit from knowing about a competitor's position, which would benefit you, but not the competitor. Indeed, the document you linked to explicitly mentions personal benefits, whereas "general benefits" such as learning about the state of research in a discipline (i.e. expanding your knowledge about a discipline so you can build upon that, which would be generally beneficial, both to you, to the authors of the cited works, and to readers of the new works), is explicitly said to be acceptable. – O. R. Mapper Oct 25 '15 at 21:17
  • The reason it is potentially unethical is that the reviewer has gained advantage from access to confidential material - new research ideas etc. The paper authors may be pursuing those themselves. This paper is apparently good, but what if it gets rejected? All of these problems make it important to discuss potential use of the material. What if there's an embargo on the results for some reason and the reviewer using that information accidentally releases part of the results? It is likely that everyone will be happy, early citation etc, but that is no excuse to assume that it's okay. – JenB Oct 26 '15 at 10:50
  • See also http://academia.stackexchange.com/questions/24239/is-it-appropriate-to-ask-colleagues-opinions-when-reviewing-a-paper?rq=1 (and probably others) about confidentiality of review process. – JenB Oct 26 '15 at 10:52
  • @JenB: "The paper authors may be pursuing those themselves." - that's why I would agree it is unethical to implement anything that is already indicated to be pursued by the author in the future work section, or otherwise a likely follow-up. But most ideas that warrant a citation are not direct follow-ups. "what if it gets rejected?" - then the referee has a problem because they cannot publish their prepared paper as planned. That's the drawback one has to cope with in exchange for the benefit of early access. "What if there's an embargo on the results" - irrelevant, as I am not saying the ... – O. R. Mapper Oct 26 '15 at 17:10
  • ... referee should submit their related work before the reviewed work has been published. Of course, they have to wait for publication of the reviewed manuscript before they can start submitting theirs, in order to avoid a broken citation chain. The advantage lies merely in making the time between original submission and original publication overlap with the time that would otherwise lie between original publication and related submission. Which is why I think it is completely okay. – O. R. Mapper Oct 26 '15 at 17:13
0

It isn't wrong to request for the code of a paper you refer (after it is accepted for publication). I've provided the code of my algorithm for a few myself upon such requests coming from my readers.

This can be regardless of the fact that you've reviewed the article. But this information can be included in your request as a supporting factor thus indicating your genuineness as you could have asked so during the review.

You may request as soon as the paper is accepted to be published and is available online at least in the open archive as facilitated in some journal publishers.

Ébe Isaac
  • 12,715
  • 2
  • 30
  • 73
  • 3
    It isn't wrong for a reader of a published paper to request code from the author, but OP is a referee for a yet unpublished paper. The confidentiality of the refereeing process requires that OP either wait for the article to be published or pass an anonymous request through the editor. – JeffE Oct 25 '15 at 16:28
  • @JeffE: I have made the edit accordingly. – Ébe Isaac Oct 26 '15 at 01:33