63

I received a journal paper for a review. The paper suggests a solution to an important problem, but the solution is very complicated and has mediocre performance. While reading the paper, I thought of a different solution, which is much simpler and has better performance (to give some scale: their solution is 1 page to write and 5 pages to prove; my solution is 5 lines to write and half a page to prove).

What should I do now?

A. Write my solution in the review, hoping the authors will remember to "thank the anonymous reviewer"?

B. Recommend that the paper published as is, then write a paper of my own with my improved solution?

C. Something else?

EDIT: Before I found the superior solution, I thought of recommending "resubmit with major revision", with the reasons that I mentioned above: your solution is too complicated and its performance is too weak.

However, now I am not sure it is the good course of action. A major revision may take a long time. The authors might even decide to delay or stop working on the paper without telling the editor (or me). Meanwhile, I will have to postpone the publishing of the superior solution for an indefinite amount of time. What do you think?

  • 6
    Is the problem you have found a better solution to essentially the whole paper, or is it just part of larger whole? – Jim Conant Jun 30 '15 at 5:26
  • 31
    Just because you discovered something better doesn't mean the paper under review doesn't deserve to get published. Would you have come up with your idea if you hadn't read the paper under review? You said it solves an important problem, that's worth publishing! – Marc Claesen Jun 30 '15 at 6:35
  • 2
    In my opinion, this makes it reasonable for you to ask the editor's permission to contact the authors. – metsburg Jun 30 '15 at 9:15
  • 1
    I will assume that there are subtle fatal errors in your superior solution that will take independent review to discover. Then it makes sense to review the paper on its own merits. Does it advance the current state of the art? – emory Jul 1 '15 at 16:30
50

I didn't get a clear impression of your opinion of the authors' work. "Solving an important problem" ought to make for an excellent paper, but the rest of your description seems to partially negate that. (In particular, I don't quite know what a "nonstandard input parameter" is, but it sounds a bit like they are cheating or that the legitimacy of their solution could be a matter of debate.)

Either way, it seems to follow from what you say that you feel that you have solved an important problem, indeed in a better way than the authors. Unless you are such a known superstar in your field that this is a commonplace occurrence for you, I would think that you deserve to get more out of this than an anonymous thanks. To me the options seem to be:

1) If the authors' paper is strong enough to be published, say so. You should not hide your opinions about their shortcomings, but you don't have to include your new ideas in your referee report. Then:

1a) If just by responding to your criticism (if they care to) the authors replicate your better solution, then it seems that you did not add critical value, and I would just recommend accepting their modified version and letting the matter end there.

But

1b) If your conscientious refereeing produces a version that is worth publishing but does not replicate your superior solution, then you should prepare your solution for publication. The timing of this is something that you should ask for other opinions about. It would be ideal if the authors had circulated their results in preprint form. If not, you may need to delay submission until their paper actually appears.

2) If you think the paper is not publishable in its current form in the journal to which it was submitted but that it inspired you to solve the problem in a better way, this may be a case for you to ask permission to contact the authors and add your work to theirs.

3) If the work is really not valuable and your solution has little or nothing to do with theirs, then you could just recommend the paper for rejection and then submit your own solution elsewhere. Examine your conscience carefully before doing this though: imagine that the authors found out exactly what you did. Would they see it your way?

  • 25
    "Examine your conscience carefully before doing this though: imagine that the authors found out exactly what you did. Would they see it your way?" This is a crucial point. And, honestly, there is almost no way that the authors will see it the OP's way. In my community, rejecting a paper and then writing a related (even a better) one yourself is the stuff that produces lifelong enemies. – xLeitix Jun 30 '15 at 14:47
  • 9
    @xLeitix Well, it would also be very hard to defend ethically. After all, the causuality and facts goes this way, "there is a problem" -> "original paper solves a problem" -> "someone gets inspired by reading the paper" -> "that someone devises a better solution". Those are simply facts, I don't see a way to skip a step here. Fact that the original paper was not published at the moment does not matter at all IMO, it should be cited (even as unpublished or private communication) just to reflect facts and give proper credit. After all without original the better solution would not be as it is. – luk32 Jun 30 '15 at 18:53
  • 1
    I believe that you should always act as if you have not seen a paper that you are reviewing for the purposes of your own research. If you want to write a better follow up, you have to wait until a copy of the original paper is publicly available first, in my opinion. – Lembik Jul 1 '15 at 7:06
  • 2
    @Lembik: That may be the topic for a different question, and maybe it is indeed field-dependent, but the people (professors, postdocs) who taught me about how to conduct research over the years uniformly expressed that having early access to related work and thus being able to proceed with one's own research earlier is one of the best ways how reviewing is not only a service to the community, but also a chance for the reviewer. (Waiting would still occur, but only related to publishing one's own research, which would then have to cite the reviewed work.) – O. R. Mapper Jul 1 '15 at 11:34
  • @O.R.Mapper You are right that this may be a topic for a different question. I just note that acting as you suggest gives you an unfortunate conflict of interest when reviewing the paper, apart from anything else. – Lembik Jul 1 '15 at 11:36
15

Option C is to contact the authors, aiming to become coauthor of a better paper. This should be done via the editor of the journal/proceeding you work for, to avoid the impression that you try to exploit your position to get free publications. The editor writes a letter to the authors stating that the anonymous referee has found a better solution, and ask whether they want to write a joint paper. If they refuse, you can still choose option B.

This option is only viable, if the result you improve is a significant part of the paper, and the authors have every right to differ in their judgement from you, so a refusal does not mean that they are mean/not cooperative/envious/whatever.

At a later stage in your carreer, i.e. once you have a good position, you should opt for A. This is not only common interest, but also helping yourself. It does not make a difference whether you have 76 or 82 publications, but editors read your reports and notice, and usually editors are influential people.

As a sideremark I would recommend you to take a little more care about the "anonymous referee" stuff. If the authors read this post, they will know whom to blame.

8

Ethically speaking, both, A and B, are perfectly valid options in my opinion. If the paper itself provides a significant contribution to the state of the art, I would recommend to publish the paper. Although your solution might be superior (which, please correct me if I am wrong, you have not yet proven) the proposed method might cover different aspects or might be of relevance in a different way. Depending on your motivation to publish a paper with your novel method yourself, this would be the logical next step. Why give your ideas to the current authors "for free" (meaning without real credit)?

  • It is not obvious that such incremental improvements are worthy of publishing on its own. Giving ideas to authors for free is exactly what reviewing is all about (almost). – Pål GD Jun 30 '15 at 7:33
  • 11
    @PålGD Reviewing is primarily about assessing the given work. – Kimball Jun 30 '15 at 8:19
  • 2
    @PålGD Giving small ideas to somebody's paper for free is indeed a common part of reviewing. However, the situation here appears to be that the reviewer would be giving the authors a whole new paper for free. Having that much input to the paper deserves nothing short of co-authorship. Indeed, if the entire original content of the paper has been superseded, one would question even the "co-" part. – David Richerby Jun 30 '15 at 12:05
4

Option B strikes me as the best solution here, as it seems the original paper gave you the basis from which to create a better solution. Their paper could also help other researchers finding other solutions.

So unless you feel that their contribution is not enough of a step forward, or their proof is missing pieces or incorrect, then it would seem to merit publication.

They get their publication credit for the paper, and you have the beginnings of a new paper.

  • Did it give the basis, or just focus the OP's attention on the problem? To me, it looks like the latter, but who knows... – Deduplicator Jun 30 '15 at 15:04

protected by ff524 Jun 30 '15 at 13:37

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.