Other How to publish if a mistake is found in a paper

[exponent137]

I found a mistake in one paper. It is not only a lapse, but it changes and anihilates meaning of the paper. How to publish this? In the same journal as a addenda, (because it can remark in short.)? Or it should be published as pa article in this or in another journal?

 

[Vanadium 50]

Is it possible that you are the one making the mistake?

 

[Drakkith]

If nothing else I would say contact the journal and see what they have to say.

 

[exponent137]

Is it possible that you are the one making the mistake?

No, I am not the only one who makes mistakes. :)

But, is it possible that addenda is given by a person, who is not an author? The problem is also, because such "errata" is very short, it cannot built the whole structure of a paper.

 

[exponent137]

"If nothing else I would say contact the journal and see what they have to say."
Yes, I will ask at the journal, but before I ask still for some your opinions.

But, once ago I asked here about one mistake (lapse) about one paper in arXiv. The answer was, that this is not a mistake. But an author of the paper confirmed that this is a mistake.

 

[ZapperZ]

This can't be answered without knowing the nature of the mistake.

If it is a typographical error, or if this was simply a honest error, then contact the editor or the author. Journals often publish corrections.

But if this is a mistake in a derivation, analysis, conclusion, then you write a rebuttal to the paper and send it to the journal. Read and understand the rebuttal paper policy of that journal, because it often has a stringent restriction, especially on its length.

If I were you, I would double check with someone else, preferably another expert in the same area, to verify that this is truly a mistake.

Zz.

 

[Choppy]

Normally if you believe that you have found a substantial error - "substantial" meaning that the conclusions would somehow need to be changed or that the readership of the journal would benefit from being aware of - what you can do is write a letter to the editor of the journal. You reference the paper, and bring forth your concern and demonstrate how the results change with this new information.

Usually the journal will contact the authors and give them a change to respond. In some cases they may simply decide to publish an errata. In other cases they may simply respond to your concern in a rebuttal to your letter, both of which are published, and are often themselves subject to peer review, if only by the editor.

Another option is simply to write another paper of your own that results in a different conclusion. You might choose this option if the particular paper is heavily cited or of other authors have made similar mistakes or if the mistake is one that has otherwise worked its way into the field.

EDIT: ZapperZ posted while I was writing.

 

[Timo]

I am a bit surprised by the replies so far. Contacting the authors of the paper in question about their possible mistake seemed such an obvious step to take to me ...

 

[ZapperZ]

I am a bit surprised by the replies so far. Contacting the authors of the paper in question about their possible mistake seemed such an obvious step to take to me ...

That was one of my suggestion. So why are you "surprised" by the replies so far? Again, it depends on the nature of the error. If it is a significant error in analysis etc., then it is perfectly valid to publish a rebuttal.

Besides, someone may want to contact the journal editors first if that person wants to remain anonymous, for whatever reason.

Zz.

 

[Timo]

That was one of my suggestion. So why are you "surprised" by the replies so far? Again, it depends on the nature of the error. If it is a significant error in analysis etc., then it is perfectly valid to publish a rebuttal.

Besides, someone may want to contact the journal editors first if that person wants to remain anonymous, for whatever reason.

As I said: Contacting the authors seems so obvious to me that I wouldn't come up with the idea to contact the publisher or flat-out write a paper. Irrespective of the type of error (except for typos, in which case "don't bother with anything" may be appropriate). The authors are the one who presumably made a mistake. They have thought about the topic for some time. They are experts in the field. And rumors are that some scientists publishing papers are actually interested in the topic and discussion about it. Of course one may imagine scenarios where one would not want to do this. But in my opinion the desire to publish something (which was indicated in the first post) and the desire to remain anonymous tend not to go well with another.

 

[ZapperZ]

As I said: Contacting the authors seems so obvious to me that I wouldn't come up with the idea to contact the publisher or flat-out write a paper. Irrespective of the type of error (except for typos, in which case "don't bother with anything" may be appropriate). The authors are the one who presumably made a mistake. They have thought about the topic for some time. They are experts in the field. And rumors are that some scientists publishing papers are actually interested in the topic and discussion about it. Of course one may imagine scenarios where one would not want to do this. But in my opinion the desire to publish something (which was indicated in the first post) and the desire to remain anonymous tend not to go well with another.

Read, for example, the Schon debacle. A couple of experts in the field contacted the editors of Nature and Science first to inquire about discrepancies they found in several of his papers. The nature of the inquiries remained private, but only later, upon their permission, did we found out who they were.

Read PRL. There have been several rebuttal papers on possible errors in the analysis of many papers. One can easily look at the most recent example from BICEP2 paper, where the error in analysis/conclusion is addressed in a separate paper. There are many more examples of that!

Zz.

 

[George Jones]

When I was a student, I read a paper in the Journal of Mathematical Physics that claimed the standard results (e.g., in Jackson) for Thomas precession were wrong by an additional angle $\phi$. My supervisor and I worked through the paper, found some mistakes, and sent a heads-up correspondence to the author. The author then published an Erratum in which he wrote

"The additional angle $\phi$ identified in Ref. 1 was unfortunately calculated incorrectly in Ref.1. There is an algebraic error following ..."

In the Erratum, the author then produced another, more complicated, expression for the additional angle $\phi$. We then published a paper showing that the new complicated expression was indeed correct, and that the new complicated expression was congruently zero in all cases, i.e., there was no correction factor!

 

[exponent137]

Let us assume that a mistake is clear and conclusion of the paper is wrong. Is it expected that the rebuttal will be published, or they can answer that the mistake is so clear that everyone can notice it? If mistake is 15 years old, is it too old? The paper was referenced once, but the mistake was not warned.

 

[Choppy]

Let us assume that a mistake is clear and conclusion of the paper is wrong. Is it expected that the rebuttal will be published, or they can answer that the mistake is so clear that everyone can notice it? If mistake is 15 years old, is it too old? The paper was referenced once, but the mistake was not warned.

The journal editor will decide whether to publish a letter that draws into question the conclusion of an earlier published paper. If in 15 years, the paper has only been cited once, it likely won't be a high priority either way.

 

[Student100]

Let us assume that a mistake is clear and conclusion of the paper is wrong. Is it expected that the rebuttal will be published, or they can answer that the mistake is so clear that everyone can notice it? If mistake is 15 years old, is it too old? The paper was referenced once, but the mistake was not warned.

Did you check the author who cited it? You may be 15 years too late to call out the error.

Otherwise, if the paper is 15 years old, and has been cited a grand total of once, it's doubtful anyone cares.

 

[exponent137]

The author who cited it was cited a lot of times. The paper with mistake is evident in arXiv. So it is not important, how old is.

 

[ZapperZ]

The author who cited it was cited a lot of times. The paper with mistake is evident in arXiv. So it is not important, how old is.

I don't know why this is dragging on, still.

If you think the error is substantial, then WRITE A REBUTTAL! Check the journal policy on rebuttals, such as how long it can be.

However, I strongly suggest you do a thorough citation search first to make sure no one else has written about it.

Zz.

 

[exponent137]

Rebuttal is written. No one else have written it. According to referee's answer, I estimate that it is 60/40 probability that it will be published. But he has the above doubt ... So I think about my chances in second resending.

 

[ZapperZ]

Rebuttal is written. No one else have written it. According to referee's answer, I estimate that it is 60/40 probability that it will be published. But he has the above doubt ... So I think about my chances in second resending.

What "above doubt"? You have stated nothing about the nature of the referee's feedback, other than mentioning how many times the original paper has been published and how long ago it was.

Zz.

 

[exponent137]

The final decision was that the mistake is such that every physicist with enough knowledge can found it. And that number of references is only moderate. They do not have some shorter format for rebuttals.

What I can do now?
Can I send this to another journal?
Can I sent it to arXiv? How to obtain an endorser for it, (general physics or quantum physics)? Is arXiv also uninterested in revealing of simple mistakes? (but this is not a lapse, because the conclusion is wrong.)

 

[Choppy]

It sounds like your answer lies in what you posted. The editors or referees felt that the error was not significant enough to warrant addressing directly. In some cases an error becomes blatantly obvious. Some journals will print a correction in these cases. Sometimes they just let it go because they judge the error inconsequential, which is likely the case here.

If you want to publish it in another journal, you would essentially have to write a paper and include a justification of why the error is important and warrants correction. And as a scientist, you would have to balance that against the other projects you're working on. Of this issue is important enough to you that you need to take time away from your work, then got for it. If not, then in the words of Princess Elsa: "Let it go."

 

[exponent137]

"Sometimes they just let it go because they judge the error inconsequential, which is likely the case here."
What do you think by "inconsenquential" (unimportant)? The article is completely wrong because of this error.

How to find an endorser in "general physics"? No one answers to the letters.

 

[micromass]

"Sometimes they just let it go because they judge the error inconsequential, which is likely the case here."
What do you think by "inconsenquential" (unimportant)? The article is completely wrong because of this error.

The editors clearly disagree with you.

Have you contacted the authors of the paper at all?

 

[exponent137]

The editors clearly disagree with you.

Have you contacted the authors of the paper at all?

Referee disagrees in what? He AGREES that the mistake in the paper is clear.

I have not contacted with the authors, because I wish to have a published rebuttal.
Sometimes I send information about lapses to some authors, but I have nothing because of this. But here is not a lapse, the whole idea of the paper is wrong.

 

[Hepth]

Remember, the ArXiV is a pre-print server. There is no obligation for the data and content in the papers to be the most recent to publishing. Many times there are last minute corrections to a paper, usually something the referee mentions, that is changed, then gets published in the proper journal correctly, but the authors never bother to update their arxiv submission. So always look at the published work, not just the Arxiv.

Maybe you are reticent to mention the exact article, but if you feel like a second set of eyes might help please PM me the link. If you don't have access to the fully published one for some reason and are just looking at the arxiv, then I can also check out the full result.

Also check to make sure there arent errata already.

But if its for sure a mistake, sometimes you let the author know. They will probably give you a mention in the errata. But I don't think its too common to publish the rebuttal if its merely correcting an error, even if it ruins the entire paper. Maybe I'm wrong.

 

[micromass]

Referee disagrees in what? He AGREES that the mistake in the paper is clear.

Sure, he might agree there is a mistake in the paper. He disagrees that it is important.

I have not contacted with the authors, because I wish to have a published rebuttal.

I'm sorry, but I feel this is a very selfish attitude. You might get a paper out of it, sure. But science is built on cooperation and free discussions. Science isn't supposed to be a race like this.

 

[Student100]

Why don't you just link the paper here?

 

[Hepth]

The moment he does I'm going to write my own paper real fast and submit it, stealing his idea and getting all that delicious credit... ;)

 

[micromass]

The moment he does I'm going to write my own paper real fast and submit it, stealing his idea and getting all that delicious credit... ;)

The editor probably stole his idea already and will be publishing a rebuttal in the next edition.

 

[Hepth]

The editor probably stole his idea already and will be publishing a rebuttal in the next edition.

Its really the only way to make a name for yourself in physics now.

 

[exponent137]

Sure, he might agree there is a mistake in the paper. He disagrees that it is important.
I'm sorry, but I feel this is a very selfish attitude. You might get a paper out of it, sure. But science is built on cooperation and free discussions. Science isn't supposed to be a race like this.

He agrees that the mistake is so important, that the paper lost all meaning. But he claims that this paper is not important. I claim, if it is published, it also needs rebuttal, if it is wrong.

I need publication that I will have a reference for future publications. I agree, this publication as a single is not important.

About Science, I have different experiences, very negative. also the endorsment sistem in arXiv is very unfair. OK, I found also very fine scientist, also all of you who gave me answers in Physics forums, and the main part of my professors. But I have also some very arrogant, and bad experiences. Scientist are people as all others.

 

[micromass]

also the endorsment sistem in arXiv is very unfair.

I don't get this. People always act like "publishing" on arXiv means something. It doesn't mean a thing. It's a pre-print server meant as a way of scientists to communicate further with the field. The endorsement system on arXiv works perfectly since every scientist who is only remotely connected to academia, gets to publish on arXiv. If you can't get an endorsement on arXiv, then nobody would read your paper anyway, so it doesn't mean a thing! What matters in science is actual published papers. ArXiv means nothing.

But I have also some very arrogant, and bad experiences. Scientist are people as all others.

Yes, there are arrogant and bad scientists. But I feel that, as a scientist, we should try to be above those bad elements.

 

[Hepth]

Does the correction have any other scientific implications that could be written about? More than just "you were wrong"; maybe "because of this error, it can be shown that EQ 4. implies that an additional source of strain on the detector comes from gravitational waves as the proton-antiproton system collapses in on itself during collision.."

 

[micromass]

Does the correction have any other scientific implications that could be written about? More than just "you were wrong"; maybe "because of this error, it can be shown that EQ 4. implies that an additional source of strain on the detector comes from gravitational waves as the proton-antiproton system collapses in on itself during collision.."

Right, try to do some further research and write a paper that IS correct.

 

[Vanadium 50]

also the endorsment sistem in arXiv is very unfair

Here is what the policy is.

A typical endorser would be asked to endorse about one person a year. The endorsement process is not peer review. You should know the person that you endorse or you should see the paper that the person intends to submit. We don't expect you to read the paper in detail, or verify that the work is correct, but you should check that the paper is appropriate for the subject area. You should not endorse the author if the author is unfamiliar with the basic facts of the field, or if the work is entirely disconnected with current work in the area.

What is unfair about that?

The bar is set very low. If your contribution doesn't meet even this level of understanding, I can see where it might be unfortunate, but not why it is necessarily unfair.

 

[exponent137]

Here is what the policy is.
What is unfair about that?

The bar is set very low. If your contribution doesn't meet even this level of understanding, I can see where it might be unfortunate, but not why it is necessarily unfair.

Some endorsers even do not answer, some are afraid that they will lost endorsment. They have only negative motivation, not positive. I expect also arguments and excange of arguments. This does not exist.

 

[micromass]

Some endorsers even do not answer,

Right. It's not their job to do this. You know what endorsement is for? It's when a professor has a very new student that wishes to publish but can't since he's new. In that case, he will endorse the student so his paper can appear on arXiv. The endorsement system is not built so random strangers get to email professors begging for endorsements.

some are afraid that they will lost endorsment.

Maybe that is a sign about the quality of the paper??

 

[Timo]

Some endorsers even do not answer, some are afraid that they will lost endorsment. They have only negative motivation, not positive. I expect also arguments and exchange of arguments. This does not exist.

I kind of understand your frustration. Especially since some voices on this forum argued that there was a sensible alternative to contacting the authors of the paper with the alleged error. But you have to admit that being annoyed about people's "negative motivations" because they don't want to engage in a discussion with you has a lot of irony to it.

 

[Choppy]

What do you think by "inconsenquential" (unimportant)? The article is completely wrong because of this error.

Obviously it's hard to say without seeing the actual paper, but it's entirely possible that the paper was seen as important enough to publish fifteen years ago (or whenever it originally came out), but now the field may have moved on. An open question back then could very well have a known solution now that's been validated by multiple independent groups. In such a case, an editor may not see any value in publishing an erratum.

 

[exponent137]

Another example:

I found a lapse in a paper:
http://arxiv.org/pdf/gr-qc/0103044v5.pdf
"with $u$ being a vector of length $\epsilon$ in ..."
I asked about it also in:
Baez's vizualisation of Ricci tensor
Then I wrote to the authors, and they corrected it, if we look the page 17 of:
http://arxiv.org/pdf/gr-qc/0103044v6.pdf
We can see, that the lapse is corrected:
"with $u$ being a unit vector"

Is it normal, that those, who warn for a lapse, are not mentioned?

 

[Timo]

Presumably. At least I have never seen an acknowledgment section saying "we want to thank Prof. Nogoodatall for long and insightful discussions about the topic, the NSF for funding and exponent137 for pointing out a typo".

 

[exponent137]

This means, if I will warn the authors about the mistake (what I was talking about in this topic), and they will write errata, I will not be mentioned.
Will they even write errata?

 

[Timo]

I don't know. My primary point was that for something like pointing out a typo or wrong choice of words I wouldn't know where to put a "thanks for XYZ to point that out". So to the exact question you asked my answer has been "probably yes". On the level where you bother to publish an erratum because the result of the paper has changed significantly this may be different. I would probably mention why I published the erratum since that sounds like making sense. But it is unlikely that we talk about a paper of mine - and even more unlikely that I would bother with an erratum. So I am not sure if knowing what I would do really helps you. I am not even sure how knowing what the authors would do will help you.

 

[micromass]

Is it normal, that those, who warn for a lapse, are not mentioned?

I really really fail to see what you would get out of such a mention.

 

[mathwonk]

Two possibilities:

1) write the authors of the error and see what they say.

If they do not react satisfactorily, i.e. if they do not retract the results in print, or if you do not wish to do this, then:

2) write a paper with a correct result, and at the end say this contradicts the results of the earlier paper.

I have seen such papers in math. E.g. In Lang's Algebra, 1st edition, chapter XI.3, he stated a theorem (thm. 7? p.281) generalizing a result of Artin, but later in a research article I think by James Cannon, in perhaps the Am. J. of Math, there appeared a paper with a theorem which contradicted Lang's "theorem". In the 2nd edition of Lang's book, section XI.3, p.402, thm 7 no longer appears.

I wrote about this mistake long ago here:

https://www.physicsforums.com/threads/criminal-offences-in-maths-textbooks.176436/

 

[russ_watters]

This means, if I will warn the authors about the mistake (what I was talking about in this topic), and they will write errata, I will not be mentioned.
Will they even write errata?

Since the paper is 15 years old and only cited once, I doubt they will write a correction. But is this all you are after here - recognition that you found an error? Congratulations! Is that good enough for you to stop wasting your time trying to get famous for finding an error in an obsolete and unimportant paper?

 

[Hepth]

To be fair, the second one he posted was cited by FOUR other papers...

But still, I would NEVER expect a mention for a simple correction. If you are asked to review it before publishing and then find it and it makes a huge difference, sure, because that's "useful" discussion. But 15 years later? No, that's just finding typos.

On that note, I do have ONE errata for a paper of mine where a student from Greece pointed out that I missed a factor of "2" when making a table. The change was not so important, but I mentioned his name explicitly in the errata. I really only did this because I knew it would make him happy, and I'm a nice person. Most people when confronted with errors in their works 6 months after publishing are not usually in such a nice mood.

 

[atyy]

Presumably. At least I have never seen an acknowledgment section saying "we want to thank Prof. Nogoodatall for long and insightful discussions about the topic, the NSF for funding and exponent137 for pointing out a typo".

vanhees71 posted http://arxiv.org/abs/hep-ph/0205014 in another thread. Footnote 6 "I would like to thank M. Beuthe for finding misprints in Eqs. (2.68) and (2.69) in the first version of the paper appeared in the electronic archive hep-ph."

 

[exponent137]

After all this story, how to publish the rebuttal?
The succesor of the original journal refutted it, because the mistake is too simple.

 

[micromass]

After all this story, how to publish the rebuttal?
The succesor of the original journal refutted it, because the mistake is too simple.

What other answer do you want than the ones you already got? It's not worth a publication.

And really, you haven't moved on from this at all?