Is it better to retract a paper, or publish a letter calling the conclusions “unphysical?”

langd5_v030i025.inddSometimes publishers and authors decide it’s easier to retract a paper than leave it up for discussion by other scientists.

That seems to be the case here: The authors of a paper in Langmuir retracted it in September for a math mistake, but not before the journal refused to publish a comment criticizing the publication.

Here’s the notice for “Drainage of a thin liquid film between hydrophobic spheres: Boundary curvature effects:”

Our article published in Langmuir is a theoretical study of the think flim drainage process between two identical solid spheres. It highlighted the possible consequences of the boundary curvature effects, which have not been studied before. Recently, during the process of generalizing the work for two arbitrarily curved surfaces, one of the authors (A.F.) found errors in the mathematical derivation. These errors were also noticed by others. Consequently, the results presented in the article are not reliable.

We apologize for the confusion raised in the academic community.

Author Yongli Mi gave us more details:

There [is] not too much to tell besides the statement we made. We made a mistake in [a] derivation [in] the modeling work and published without noticing it. After the work was received, one of the editors in communication with others found our mistakes. I immediately contacted my post-doctoral research, Dr. Fang Angbo, and he told me the mistake he made was also found by him, so we retracted our paper.

However, physicist Olga Vinogradova emailed us to express her dissatisfaction with this explanation. She and a co-author submitted a criticism of Mi and Fang’s paper to Langmuir in June, but the journal chose not to publish it:

I’m very disappointed with the Langmuir Journal. The paper was approved by 3 referees and an Editor, and none of them (!) detected numerous fundamental mistakes there. This clearly indicates the low standard of reviewing in Langmuir. After Evgeny Asmolov and I saw this paper, we immediately decided to write a comment. Some mistakes were obvious, but one of them required a detailed clarification (since to detect it you should be a professional in math). You could find the details here http://arxiv.org/abs/1406.1614

We submitted a comment to Langmuir, and several referees recommended to publish it. However, the editors of Langmuir rejected our comment by telling us that after reading it Fang and Mi agreed with our critics and would like to withdraw their paper. I believe this was a wrong decision since comments are an important part of a scientific life (and can prevent other newcomers to the field from making the same mistakes).

It is indeed ridiculous to read now the editorial note about this retraction. Asmolov and I found mistakes (in June), but not Fang (in September)! What the Langmuir writes in this editorial note is simply not true. Better ask the editors why they declined our comment and why they are now trying to misinform the readers. One cannot hide this story since our comment is in arxiv. Many people have read it and know the true story.

We’ve reached out to the editor, and will update with any new information. In the meantime, take our poll:

24 thoughts on “Is it better to retract a paper, or publish a letter calling the conclusions “unphysical?””

  1. The poll options here were not sufficient. I think both publication of the comment and retraction should have been done. That would clarify the issues, give credit to the commenter, and indicate the paper as retracted.

    1. This is the best answer. Both. Interestingly, the retracted by Langmuir paper can still be downloaded with the Editorial Note telling that the paper contains errors, but readers now do not have any ideas where these errors are hidden since the Comment pointing out the calculation mistakes was rejected.

  2. Did Vinogradova directly contact the authors first? That would have spared them the work of writing a manuscript and the frustration of having it refused.

    1. Frequently, it seems a pointless attempt to contact directly the authors: they won’t bother replying, and wake up only if they feel pressured.
      (This is a general comment, which perhaps doesn’t apply to the present case).

    2. No. I think published Comments are much better than writing to authors since they make a discussion public and clarify mistakes for everybody who is interested. I’m in fact not so frustrated by a rejection of our comment. A much more disappointing thing is the “strange” Editorial Note since everybody knows that the paper by Fang and Mi was retracted as a result of our (rejected) Comment, but not because “A.F. found errors”.

  3. I have a letter under review for a very similar situation. It is a comment on an article in which a model is used incorrectly, leading to absurd and unphysical results. The fact that the paper was accepted in the first place was a clear failure of peer review. I might have ignored it if it wasn’t for the fact that, firstly, the journal is respected and a mark of quality research, and secondly the senior author is well publised and cited in the field, and really should have known better.

    I chose not to contact the authors or the editor first. Instead, I wrote the letter to lay out clearly where the mistake was, what the consequences were, and then use another data set to show that the model doesn’t do what it claims to. Thinking it through, I believe this was the correct course of action, although it wasn’t clear cut. I’m not familiar with Olga Vinogradova’s criticism, but it seems they came to a similar conclusion.

    I think journals should have guidelines regarding what to do when a fatal flaw is discovered in a published manuscript.

    1. I think it is important to let the readers know what the flaws were in the original paper. And it certainly helps to explain why it doesn’t work as is explained in the paper, to prevent other people from making the same mistakes. So in that sense retracing a paper might be worse than having a comment. Especially if the retraction notices are as short as this. Not explaining what exactly is wrong and/or why.

      1. I personally disagree. If the error is just simply an error that invalidates the paper (say a typo in an computer program) , the proper course of action is retraction (with a clear retraction notice).

        I don’t see the point of having two papers out in the literature for a gain of knowledge equal to zero.

        No author on earth likes to look like an idiot, therefore if you politely point him /her to the mistake, they will correct it promptly.

        I don’t see the point of writing a paper for that (except growing your own publication record).

        If the mistake is not that obvious and deserves a bit of discussion, then only it may make sense to write a paper.

        1. I disagree. I personally think that the retraction is absolutely necessary if scientific misconduct is detected (plagiarism, falsification of data, etc), but should NEVER happen in the case of scientific mistakes. Mistakes are made by the authors (editors, positive referees…), and this is their problems now. The scientifically wrong paper should be open for a discussion. In this particular case my arguments to the Langmuir Editors were:

          1) The paper by Fang and Mi has appeared online in December 2013 and has already confused the scientific community. By publishing their paper the Langmuir journal has given to readers the wrong message that their theory should now be used instead of Langmuir 1995 publication. Therefore, there is a danger that experimentalists (that might not have strong mathematical background to follow the derivation by Mi and Wang) are already using the erroneous theory to fit data, which could lead to wrong physical conclusions and stimulate a series of erroneous publications.

          2) The publication of our Comment would prevent other newcomers to the field from making similar mistakes. As we explained, some mistakes can only be detected by professionals in mathematics and hydrodynamics.

          3) The Comments are a natural part of scientific discussions, which increases the interest of readers to the field and also plays some educational role.

          1. I think you are exactly right Olga.

            I am not in favour of retractions of flawed papers where there is no fraud. The caveat being that some papers are so dangerous (e.g. people will die if paper X is allowed to remain as part of the literature) that they do need to be retracted. If every flawed paper were retracted it would be mayhem – don’t people realise that half the scientific literature would have to go?

            In this case, the authors have accepted their mistake. The journal should have published your comment and – if they wished – the authors could have appended their agreement. That way, the research community can learn from the error and all the key information is in one place.

            The scientific journals are – en masse – completely and utterly failing to understand how to take part in productive Web 2.0 discussions. They are still trying to suppress online discussion of their publications. In the era of RW and PubPeer how gormless is that!?

            Let s/he who is without error in any of their scientific publications be the first to disagree…

          2. I fully agree with you. Everybody could make a mistake (especially in high risk/high gain science), which will be one day corrected by others. That’s how science develops, and this is not a reason to retract the erroneous paper. As Evgeny Asmolov and I write in the last paragraph of our Comment, Fang and Mi correctly indicated the timely problem, but failed to solve it. So what? To retract their paper and to hide Comments on it? This is the worst possible solution. If we look at the old journals, one can find hottest debates which resulted in really great new knowledge, and numerous published Comments-Replies are often more valuable (and famous) than original papers that started a discussion.

          3. Still unconvinced… Maybe it’s because of a few specifics of your field and of this particular case that I don’t know.

            A few arguments don’t make sense to me:

            1) Only scientific misconduct should lead to retraction.
            No, absolutely no. Scientific misconduct, honest mistakes and even new knowledge should lead to retraction. When a flaw impairs a paper to the point that nothing good can be taken of it, it should be retracted. In my field, that has happened many time (people realizing that they were working with the wrong cell line, or with the wrong DNA construct from the beginning, such that 100% of the data is useless).
            I’m not talking about papers making the wrong conclusions from their data.Such papers, I agree, should not be retracted, because data may still be useable and reinterpretable.
            Check the “doing the right thing” section in retraction watch to see plenty of retractions that are not due to misconduct.

            2) The publication of a Comment would prevent other from making similar mistakes.
            Again, how could people get any inspiration from a retracted paper? If the paper is retracted it means it should never have existed. It will never be read, nor cited anymore (if you download it, you get a big RETRACTED watermark in the middle). That is the best way to prevent further mistakes.
            Writing a comment could, on the other hand, confuse the newcomers into thinking that this is a controversial matter (which it is not, since the authors agree with you).

            I am totally pro scientific debate, and I agree that this may results in great knowledge, but only as long as there is something to debate. This doesn’t seem to be the case here, since everybody agree that there was a derivation mistake.

            Finally I don’t understand the sentence:”Mistakes are made by the authors (editors, positive referees…), and this is their problems now.”
            If they retract a paper, it is because they are aware of the problem and want to fix it. What else should they do?

          4. Readers can still download retracted Fang and Mi paper (at least in this Langmuir case the erroneous paper remains online), but what do they learn without a Comment pointing out the calculation mistakes?

          5. “Readers can still download retracted Fang and Mi paper (at least in this Langmuir case the erroneous paper remains online”

            If you download the pdf, there is a huge “retracted” watermark across the page. The html starts with a “withdrawn” notice. There is no way to ignore these.

            “but what do they learn without a Comment pointing out the calculation mistakes?”

            But what is there to learn from that paper, since everybody agrees that it’s wrong? Wouldn’t it be better if that paper never existed? Isn’t it better to send it to oblivion ASAP? Is there really an interest in knowing how they did a calculation mistake?

          6. Yes, it would be much better if this paper never existed. But! It was initially approved by 3 positive referees and the Langmuir editors, and was during 10 months online (without “retracted” water mark). Everybody agreed that it is wrong only after our Comment, but not before. And yes, many (especially theoretical) readers are interested where mathematical errors are to do not repeat them in the future. I realise, that the journals do not publish comments on retracted papers, but at least the Langmuir could give a correct explanation, instead of inventing these stories (about editors and authors that found unknown mistakes themselves). In this case one could simply write “E.S. Asmolov and O.I.Vinogradova found basic faults, which led to unphysical conclusions of Fang, A.; Mi, Y. Drainage of a Thin Liquid Film between Hydrophobic Spheres: Boundary Curvature Effects. Langmuir 2014, 30, 83-89, and clarified them in their Comment http://arxiv.org/abs/1406.1614 submitted to Langmuir in June 2014″

          7. “at least the Langmuir could give a correct explanation, instead of inventing these stories”

            Now we do agree!

            The problem is not so much that one needs a comment paper on every flawed paper. The problem is that one needs a clear and exhaustive retraction notice and a proper account of the facts. You seem to disagree with the fact that Fang may have independently found the errors and you are unhappy that you are referred to as “…and others”, rather than nominally. I can’t judge on that at all.

            But I think that, in principle, a retraction with a good retraction notice should be the way to go in such case.

          8. Right. I certainly disagree that Fang and the Editor “independently” found mistakes. The Editor was informed about mistakes at the beginning of June, which is documented by the submission date. According to the oficial ACS (and other publishers) procedure he should immediately informe the authors about the Comment on their paper by sending it to them for a Reply. So the word “independently” is no longer valid.

          9. “If we look at the old journals, one can find hottest debates which resulted in really great new knowledge, and numerous published Comments-Replies are often more valuable (and famous) than original papers that started a discussion.”

            How fascinating! How “old” do you mean? When did this practice die out?

            I begin to wonder if there isn’t an internet poll in this? For those that think flawed papers should be retracted, how flawed should they be? 10%? 70%? 30%? 50%? Or maybe most scientists think that published papers are only ever all right or all wrong?

          10. Earlier journals allowed one many rounds of Comments-Replies, but apparently this practice stopped 30 years ago or so. A good example of the scientific impact of these discussions is the development of the adhesion theory. The Derjaguin DMT model initially rejected the JKR theory. There have been many hot discussions about this and some specific details of the models. The one between Taybor and Derjaguin published in JCIS as many Comments-Replies is famous. It helped many of us to learn the adhesion field much better and stimulated many new research. Finally it was understood that both the JKR and DMT models are correct and quantify adhesive contact in two different limiting cases (depending on the parameters of elastic body). Clearly, Comments-Replies are very important scientific contributions.

          11. That is a nice example. Since the journals nowadays don’t want to host critiques of their published papers, in case it makes them look bad, they are not going to be part of the new scientific discussions enabled by the internet. Instead, it will take place at Arxiv, PubPeer and their ilk, without the dead weight of defensively minded editors.

    2. Well, my letter was accepted for publication today, so it seems that the Bioinformatics editors didn’t have reservations about publishing a comment that is critical of another one of their articles.

Leave a Reply to Ivan Oransky Cancel reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.