My question: Are there ways to point out errors of somebody's scientific work without sounding harsh or as if I am discrediting that person? A lot can be found about criticizing well and proper feedback, but that usually concerns a face-to-face conversation between two people who know each other. My situation concerns a scientific work of somebody unfamiliar.

The situation: I have written a scientific article that corrects and improves a computer model of another scientist, say of Johnson. To justify the relevance of my article I have to point out the errors of the former model. Currently I say this like "Johnson errs twice by proposing equation ...", "Johnson's model lacks ...", "Johnson employs a method unfit for the purpose of ...".

The issue: I thought my wording was clear and business-like, I also start by giving credits to Johnson for being the first with his model. Yet three proof readers found my criticism harsh, one of them even said my words are discrediting Johnson. Clearly I am currently not attaining the right goal, people now read it as harsh criticism instead of an explanation what is wrong with Johnson's model.

My (somewhat) fix: Instead of "Johnson makes two mistakes by proposing equation ..." I could say "Johnson proposes equation x. Equation x contains two mistakes.". That is, I tag his equations, and then shift focus from the person Johnson to his equations, thereby making it sound less personal. Likewise, instead of "Johnson's model" call it "the former model". From a writing standpoint, however, this sometimes reads artificial.

To make matters more complex: Johnson's model has flaws yet he presents very good results, sometimes unusual good results. I explain why this is unusual. I do not want to sound harsh here, yet I also do not want sound as if his results are plausible.

Personal sidenote: Rephrasing my wording sometimes feel like twisting myself into difficult postures. That I have to omit the word "Johnson" to sound fine. And even when Johnson did a poor job I should obfuscate this.

  • 3
    Hi Bart, have you checked academia.se ? seems a very good first step, and if you need to replace some useful tips found over there within the "interpersonal way of dealing with the problem", maybe let us know? ;)
    – OldPadawan
    Feb 20, 2019 at 12:24
  • 1
    As the preferred methods of giving and receiving feedback may differ per country, could you clarify a country/culture for both you and Johnson (if you know where Johnson is from..)?
    – Tinkeringbell
    Feb 20, 2019 at 12:38
  • 1
    @OldPadawan: actually, I find interpersonal.se provides better and more creative answers, answers that I did not thought about.
    – Bart
    Feb 20, 2019 at 14:14
  • 1
    @Tinkeringbell: Johnson is from Germany. Now I understand I should not have given him an English name, haha. I am from the Netherlands (neighboring country of Germany).
    – Bart
    Feb 20, 2019 at 14:16
  • 1
    @Tinkeringbell: No bother. One proof reader from Germany (my co-author) and two proof readers from the Netherlands.
    – Bart
    Feb 20, 2019 at 19:48

5 Answers 5


You did not give much information on what errors johnson has committed. But my suggestion would be to assume that Johnson did not commit errors but instead oversights.

Maybe he merely set the groundwork for your more advanced formulas and you can give him credit for that instead of discrediting him for his shortcomings.


Johnson has already proposed a solution for problem X. The solution works fine in cases A-J. However during my research, I discovered factor Y also had a significant effect on the quality of the results. So I modified Johnson's solution to include factor Y (and Z and aleph).

This way you give credit, you still mention shortcomings and you justify your own approach.


And just as a further point of consideration. Imagine that you are the one who is missing something. Maybe at some later date another researcher (let's call him Bort) will find a mistake in your paper invalidating all your findings. Not only will you be in a similar position as Johnson is in now, but also will your criticism of Johnson stand in a much different light.

  • 1
    To be frank, what I think happened is that Johnson worked sloppily. That he derived his model in an ad hoc manner, what seemed intuitive and logical to him personally, instead of working on a rigorous base. His tests are idealized. If he tested thoroughly he would have known his model would not work in most situations. His model is just too flawed to be groundwork.
    – Bart
    Feb 20, 2019 at 19:57
  • 2
    @Bart: i understand your frustration. But given that it is so obvious to you, it might be just as obvious for your readers. There is nothing to gain from ridiculing Johnson. Just give him credit, where it is possible and stay neutral otherwise. Otherwise you might drift off into a rant.
    – BestGuess
    Feb 21, 2019 at 9:58
  • Well, I give him credit for being the first with the idea. Furthermore, I am not ridiculing Johnson, I am criticizing his work. But my current tone sounds too direct and personal. My question is therefore to hear ideas from you people in what ways I can sound more neutral.
    – Bart
    Feb 21, 2019 at 13:52
  • 1
    Thanks for the edit. Yes, I thought about that one. What I did is substitute Johnson with my own name or a friend's name, and then reread. Indeed, several sentences then sound painful, so those sentences have to be rewritten.
    – Bart
    Feb 21, 2019 at 19:48

As a scientific author, what I can tell you is that in a scientific paper you want to avoid negative language.

Mentioning mistakes, errors, and the such is negative language. It comes across as if you are looking down on the person you are basing your work upon.

So instead of saying

Johnson proposes equation x. Equation x contains two mistakes

it is much better to say

Johnson proposes equation x. Equation x can be improved by ....

and so on.

Speak of improving, of extending the validity of the equations / models.

Speak of extending the results, of making the model more general, of improving it.

Said otherwise, you want your paper to shine of its own light (i.e. your positive contributions), not due to someone else's darkness (i.e. someone else's mistakes).

  • Thanks for the suggestion, and I do that now if that is possible. But you also have to justify why Johnson's equation is wrong. Otherwise the reader may think Johnson's equation is another possibility.
    – Bart
    Feb 22, 2019 at 14:58
  • @Bart obviously I don't know the details of your specific situation, but as BestGuess suggests in their answer, you could say "Johnson's equation is valid only in case A. In this work we explore how this could be excessively restrictive and how to expand the equation validity"
    – Federico
    Feb 22, 2019 at 15:21

Disclaimer first: I've never actually reviewed a research paper, but I have a few years of experience doing code reviews (online, not face-to-face) with coworkers. I think reviewing a paper from someone you don't know in person can be compared to the first few code reviews you do for a new coworker. From what I know there isn't much cultural difference when it comes to giving feedback between Germany and The Netherlands, the country I live in, so hopefully this will work for you.

Yet three proof readers found my criticism harsh, one of them even said my words are discrediting Johnson. Clearly I am currently not attaining the right goal, people now read it as harsh criticism instead of an explanation what is wrong with Johnson's model.

The stereotypical feedback culture for many Western European countries assumes that feedback can be direct, and you don't need to hide behind things like positive statements. However, direct feedback should not come across as a personal attack. Remember you're giving feedback on the equation used, the code written, not judging the person behind that work.

You're indeed using a lot of 'Johsons' in your feedback, from what I can see. And you write that 'Johnson proposes equation X. Equation X contains two mistakes' reads too artificial for you.

My proposal would be to leave out the Johnson entirely. People will know he wrote the paper you're reviewing, his name is on it. You're giving feedback on what's in the paper, on the equation. So just limit to 'Equation X contains two mistakes' should be fine, and will come across much more impersonal.

To be even nicer, don't make absolute statements. Even when you're 100% sure something is wrong, phrasing it like 'It looks like equation X contains two mistakes, X and Y' might soften the blow even more. When I get code reviews written like that back from co-workers, it feels a little bit less like someone has taken a red pen and marked something 'WRONG', and a bit more like 'hey Tink, you might've missed something here, take a good look at X and Y again'.

Rephrasing my wording sometimes feel like twisting myself into difficult postures. That I have to omit the word "Johnson" to sound fine. And even when Johnson did a poor job I should obfuscate this.

That last line gives me the impression that you're upset with Johnson for doing a poor job. That's understandable, but it's also what's probably making your communications appear as too personal and harsh to the other reviewers. Try to distance your feelings about how well Johnson did their job from what is wrong with the paper Johnson wrote.

By giving honest criticism (X is wrong, Y doesn't work as intended, result Z seems implausible) you're already showing that Johnson did a poor job, you're not obfuscating it. Yes, it might feel a bit weird, it might feel like you're going out of your way to twist your language if you're used to a more personal, direct way of communication. But doing this regularly makes this come easier every time, and it's a valuable skill to be able to communicate in an impersonal way. So I do recommend you keep trying.

  • "My proposal would be to leave out the Johnson entirely. ..." I name other people, e.g. "Anderson's method [1] for Y shows that". Because (a) readers know the work from the name, not needing to consult the References. (b) the reading becomes more natural and personal than "The method in [1] for Y shows that". It would look inconsistent and awkward if I suddenly change that style for Johnson. I guess this has become both a criticism issue and a writing issue. I think I am looking for a way to introduce Johnson's model and then speak of his model in a natural way without mentioning his name.
    – Bart
    Feb 21, 2019 at 13:46
  • "To be even nicer, don't make absolute statements." I think here the code review differs from a published article. The code reviewer only has to warn you, and then you know what to do. I have to warn a broad audience, and I may not and should not leave my audience in doubt with statements "It looks like".
    – Bart
    Feb 21, 2019 at 13:47
  • "That last line gives me the impression that you're upset with Johnson for doing a poor job. ..." Yes, I added that line to make you people understand a bit my view. I spent a lot of time trying to get his model working. And if he did a proper job I would not have wasted so much time with his model. But this not only concerns me, others may fall into the same trap, therefore I prefer(red) not to obfuscate his poor work.
    – Bart
    Feb 21, 2019 at 13:47

I was in a similar situation writing a paper last year. I wouldn't say the Johnson of my paper worked sloppily, but that his model could be greatly improved by simple changes usually known by people in another field than his, but not usually known by people in his field. I assume that in your case you work in the same field, which makes your situation not exactly the same as mine, but I believe my situation is sufficiently similar.

In any case, my advice would be to change your language to be positive, i.e. focus on the positive sides of improving the model rather than pointing out the shortcomings in a really negative way. I promise you that the readers of your paper will see that Johnson were in fact sloppy (if it is as bad as you say), but will understand that you need to keep a professional tone. If you do so, Johnson might actually thank you for improving the work he started.

I would suggest to change negative language from

Johnsons model does not work in cases A and B shown by simple cases X and Y

To something like

Johnsons model can be improved to also work in cases A and B as it currently falls short in cases X and Y

If the X and Y cases are in fact simple, the readers will know and conclude that Johnson should have thought about them, all the while you can keep a professional tone.

  • Thanks, you gave me a bit of certainty that readers understand it if my wording becomes slightly less concrete during this delicate passage of the article.
    – Bart
    Feb 22, 2019 at 14:58

Thanks for all your answers and comments. Together with all the reading about how to criticize well, they taught me to look at my issue from other perspectives. And while interacting with you people I learned to look at criticism with a more emphatic view. I did not find one striking approach to resolve my issue but employed a variety of approaches and writing rhetorics. Two approaches worth mentioning here are
1) Cultivate a mindset on the work and not on the person. Simple, concrete example: I renamed directory 'CriticizeJohnson' to 'CriticizeWorkJohnson'. With such a mindset the writing becomes more natural, in contrast to patching your writing.
2) Substitute Johnson with your own name or a friend's name. Reading certain sentences then indeed feels painful. It allowed me to identify the sentences that are likely too harsh/direct. One the other hand, you may also experience a sentence as a fair admonition.

Two reads worth mentioning:
1) https://medium.com/@jamesheathers/how-not-to-be-a-crank-819103800502
2) https://academia.stackexchange.com/a/120705/62668 Best summary of the answers I found on academia.se

Lastly, formulating good criticism takes time and thought.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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