4

Hypothetical:

  • I found an error which fundamentally kills a published paper and re-opens a number of problems the paper claimed to solve.
  • Not only is there an error, but the results are actually wrong. Part of another publication (by a different author) proves this. Yes, I am 100% sure. Other experts have confirmed this in private communications with me.
  • I have nothing new to add, other than I can show the second paper proves the first paper is wrong.
  • I get no response from the first author when I let them know about the error.

What is my next step?

(A) contact editors of the journals with the incorrect results? and then what happens?

(B) post a paper to ARXIV?

(C) submit a paper to a journal?

(D) wait, do more research, and try to craft something more "substantial" for (B) or (C)

(E) do some or all of the above simultaneously? (please specify which ones)

I am not sure if my contribution thus far warrants (B) or (C), but I would like to receive some acknowledgement in print.

Something must be done because this error if unacknowledged will have a negative effect on future research.

Improve this question
2
  • 3
    What field is this?
    – Buffy
    Commented Apr 19, 2019 at 0:50
  • I upvoted @Buffy's comment and am commenting to reask her question. This is a question of academic culture, so it is highly field dependent.
    – Pete L. Clark
    Commented Apr 20, 2019 at 20:06

3 Answers 3

Reset to default
2

I have nothing new to add, other than I can show the second paper proves the first paper is wrong.

If the second paper already refutes the first paper, then there is nothing to do. If you could write a brief, non-trivial paper explaining how the second paper refutes the first (and the consequences of this), that's a different matter.

(A) contact editors of the journals with the incorrect results? and then what happens?

This may be field-dependent, but in general, I think this only be appropriate if the authors acted in bad faith (and thus the paper needs to be forcibly retracted). In this case, it sounds like the authors were simply wrong, so I wouldn't get between the authors and the journal.

(B) post a paper to ARXIV?

Particularly in math and physics, all papers are posted to arXiv. It's not a bad policy for other fields as well (I post all my machine learning papers to arXiv). But it sounds like your results are significant enough that you should publish them in an actual journal or conference, not "just" the arXiv.

(C) submit a paper to a journal?

(D) wait, do more research, and try to craft something more "substantial"

If you have enough "substance" to publish now, then I would go for it (and post the paper to the arXiv while you're at it). If not, doing more research is necessary -- you shouldn't publish a bad paper, particularly one that "calls out" another paper (all the more so if the other paper is well known).

Improve this answer
1
  • 2
    Thanks for answering. Regarding your first sentence, I should mention that the correct paper did not point out the error in the other one. FYI, in no way do I think the author acted in bad faith and/or refuses to cooperate. All I know is that I got no response from them. Also regarding your last sentence... it will be hard to avoid "calling out" the other paper, but your point is well-taken.
    – Anonymous Researcher
    Commented Apr 19, 2019 at 1:03
2

I think it is appropriate to write a short (even 1-page) Erratum paper and submit it to the journal in which the original paper was published. However, before doing so I would make sure you try hard to contact the authors and get their feedback, and possibly even their co-authorship on the Erratum. If they do not respond in a reasonable time, it's OK (in my opinion) to submit the Erratum to the journal, and you can include a note to the editor that explains the sequence of events that led to this point.

Your Erratum will be stronger if you can show how to fix the error, not just point out the error. But I understand this might not be possible.

Keep in mind that your goal in all of this is to help the research community by pointing out this error, rather than adding a pub to your CV. It sounds like you are approaching it with the right attitude.

The community deserves to know that the original paper contains this error, if it is an important one.

Improve this answer
10
  • I am not convinced erratum is the technical word for this. The fact that a paper pointing to the error does already exist mitigate the need for an action. I would certainly send a comment if able to fix it, but otherwise I would rather wait for a fixing to come to my mind, or better carefully pointing to two contradicting papers when needed (if needed) for discussing my own work. Of course the latter is possible only if those papers are related to your current research. I add a question: how the " two papers behave to each others"? Ignoring the point?
    – Alchimista
    Commented Apr 20, 2019 at 12:10
  • My understanding from the OP is that the “second paper” doesn’t explicitly argue that the first paper is wrong, but rather that some part of the second paper can be used to argue that the first paper is wrong, and that’s what Anonymous Researcher did. I could have misunderstood the OP though. If the second paper already explicitly argues that the first paper is wrong, then I agree, there is nothing to do here.
    – LarrySnyder610
    Commented Apr 20, 2019 at 12:14
  • No Larry I also think as you. Just not sure it is a good move to point out the error without a fix, but I am not sure honestly. Say, if one can disprove general relativity without fixing something, it will still make much sense. Opposite pointing to the fact that alchimista made a stupid model for an experiment important to specialists only, well better wait or try to discussing it without a frontal "attack", at least without proposing the correction.
    – Alchimista
    Commented Apr 20, 2019 at 12:20
  • 1
    I think it depends on the importance of the paper and of the error to the community the paper is aimed at. Even if the paper is aimed at specialists only, it might still be very important to correct the record and make sure future researchers don’t rely on the incorrect results, leading them down the wrong path. OTOH if the error is just a minor mistake, no action may be warranted. It would certainly be better if the OP could offer a fix, but if the paper was trying to solve a hard problem (as most research papers are) then developing a fix might be as hard as the original research topic.
    – LarrySnyder610
    Commented Apr 20, 2019 at 12:56
  • -1: That answer would work in an ideal world. But seriously, denouncing an important error to the people who made it and to those who published it...
    – Sylvain Ribault
    Commented Apr 20, 2019 at 19:00
-3

Definitely do not contact the journals who published the incorrect results. Their interest is to reject, bury or delay the wrong results' denunciation. To get a fair and professional answer from them is possible, but unlikely. And it would probably require an unreasonable amount of time and effort from you. Quoting James Heathers:

How does “originally rejected without review, appealed, then accepted after 18 months, 5 rounds of reviews, and 6 reviewers” sound? Does that sound like something that would ever happen to a paper?

Well, that’s what happens to criticism. Would it be more efficient to just upload it and get on with your life?

This leaves you with arXiv, if it is widely used in your community. (Failing that, a well-chosen website or social network could do.) Short rebuttals of papers on arXiv happen quite often.

Improve this answer
2
  • 4
    This seems awfully cynical. Reputable journals want to publish quality results.
    – Buffy
    Commented Apr 20, 2019 at 20:13
  • Maybe. But if you wait for the journal to correct such an important mistake, the correction is likely to appear too late for being useful, if at all.
    – Sylvain Ribault
    Commented Apr 21, 2019 at 18:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .