How to deal with a coauthor who introduces errors

Question

This situation has arisen a couple of times for me and I don't know how to deal with it:

I'm writing a paper that contains formal mathematical statements and proofs. I am very careful about ensuring that my statements are precise and that the proofs are correct. Usually my coauthors are equally careful, but sometimes I have a coauthor who isn't.

The coauthor will edit my careful statements and proofs, presumably with the goal of improving the exposition. But they end up inserting errors, such as dropping necessary assumptions from theorems or incorrectly simplifying something in the proof.

If this happens once or twice, I don't care - it's an honest mistake and they are improving the paper overall - but some people make a habit of it, sometimes to the point where they are more of a hindrance than a help. It really annoys me, as I need to fix their errors and I end up not "trusting" my coauthor - i.e. whenever they make edits I'll go through and check what they've done.

Here's an example: Essentially, the proof of our main theorem involved defining a function, proving several properties of that function, and then using those properties to bound the integral of the function. My coauthor "simplified" the proof by redefining the function to be the integral of the function I had defined. This meant that the function was now a constant and the rest of the proof made no sense. He obviously had good intentions, but didn't understand the proof he was editing. This error made it into the arxiv version of the paper. It was not a big deal and nobody said anything, but it was annoying, as I had correctly written that proof.

I keep my displeasure in this situation to myself, as I don't want to offend my coauthor or appear unreasonable, but I want to tell my coauthor to be more careful. It's also good for them to improve their work habits. Usually I'll leave a comment for my coauthors in the manuscript explaining why things have to be done this way, but that doesn't seem to get the message across.

Is there some way I can encourage my coauthors to drop this bad habit without offending them?

EDIT: To be clear, I'm talking about coauthors at approximately the same seniority level as myself.

EDIT: In the example, it wasn't just the two of us. There were other more senior authors. So I didn't feel like it was my responsibility to address this issue. I didn't talk to the other authors about it, but one of them did also express irritation.

Answer 1

I recommend two different approaches, depending on the other types of value contributed by your co-author:

• If your co-author is not contributing technical value to the project at all, then you don't need to make a big issue of things: just simply do not invite them to collaborate with you on your next paper.
• If your co-author is contributing in other areas, but inconsistently for the proofs, I would recommend using a good version control system (e.g., Subversion, mercurial, git). Then you have an easy mechanism by which you can do explicit review of the changes that they introduce and adopt precisely the ones that you are positively comfortable with.

Answer 2

This error made it into the arxiv version of the paper. It was not a big deal and nobody said anything [...]

Hmm. I see three telling issues just in this short text fragment, which I think provide some insight into the situation.

1. A serious error made its way into the arXiv version of the paper. Sorry, but to me that would be a very big deal, and I would be quite pissed.

2. You are saying that this was not a big deal to you. Maybe I'm reading too much into this statement -- it could be your understated personality; or maybe you were upset at the time but have gotten over it and are less upset now that the error has been corrected; or maybe you think that you shouldn't be upset, and that being upset is a sign of immaturity. Well, as I said above, in my opinion (the opinion of a random mathematician on the internet, which obviously counts for a lot :-) ), it is a big deal and deserves to be treated as such.

3. Nobody said anything. As I said: hmm; these are in my opinion the most important three words in your question. It sounds like your coauthors are "nice guys," as you say, and one of them actually made a good technical contribution so you may want to collaborate with him again. However, good communication is essential to a good collaboration, and what seems to be happening is that this nice guy and good technical contributor is simply unaware of the damage his sloppiness is causing (and possibly even unaware that he is being sloppy). I understand your reluctance to confront him -- maybe he's a sensitive person and will be offended, and we all hate to have awkward conversations in which we criticize someone else. Well, I don't see what other choice you have other than to terminate the collaboration, which would be a pity, or to carry the burden of working with a collaborator you can't trust, which would be very unfair and dispiriting for you and would not bode well for the future of the collaboration.

My advice: have a friendly chat with your coauthor. Bring up the example with the proof he messed up, and another couple of examples, and explain in as sensitive a way as you can that you would be really happy to continue collaborating with him, but that this issue of sloppiness and lack of trust is bothering you a lot. Focus as much as you can on positive things, e.g. by saying "you're a great researcher and I think you made a very good contribution to our paper, but ...". Here, after the "but", there should come a clear, direct and unambiguous description of what your coauthor did wrong; i.e., try to soften the criticism as much as you can by wrapping it with positive things, but don't be tempted to make the main criticism itself vague in order to soften it. The feedback has to be expressed directly, rather than in a passive way by leaving easy-to-ignore comments in the latex source file of the paper.

To summarize, I think you will find that although people generally don't enjoy too much being criticized, a lot of people actually appreciate being given constructive feedback that helps them improve, especially if it is delivered in a sensitive and positive way. In any case, even when they don't like it, sometimes it needs to be said. Good communication is important to keep any relationship, whether it's with an academic collaborator or anyone else, healthy.

Answer 3

My approach would be as follows:

1. Taking the 'blame' for focusing on this issue: Tell him you're kind finicky about phrasing and style in article's you're co-authoring, so you're asking for his patience when it comes to discussing changes and that s/he be receptive to your nitpicking more than s/he might expect.
2. Ask that you both use change tracking when making changes. If it's LaTeX, there's the changebar package, and a bunch of packages for making notes, or coloring text etc. If it's LibreOffice or MS Word, insist that you both use Track Changes.
3. Ask that, when s/he makes a significant change, that s/he also add some comment explaining why it was made.

This already puts you in a better position, I believe, since either s/he accepts your requests, in which case you can easily check whether they screwed something up (and s/he will think twice about each change as s/he makes it) - or s/he argues with you about it. The thing is, it's much better to argue about your wanting to supposedly be pedantic than to argue about your accusations of his/her being an "error introducer"...

Answer 4

I think the time to get your ground rules straight is as near the beginning of a collaboration as possible.

Here's a possible way to do it:

Hi all,

I'm excited about the project. Since it seems to be taking off, I wanted to take this opportunity to let you know a preference that I have. I'm pretty easy going in a lot of ways, but when it comes to proofs that I contribute, I'm meticulous to a fault.

If I contribute a proof to this project, and one of you wants to rewrite it, please flag that edit, and please don't wait until we're very close to publication to propose such a change.

Thanks. This will do a lot for my peace of mind.

Answer 5

There are two things you can, and should, do:

1. Point out errors when you see them. As long as your communications stay 100% about the work, this needn't be an emotionally charged action. People should have a thick skin about receiving constructive feedback, and if they don't, that's their problem rather than yours. You can let errors slide in early drafts and come back to them later, but it is your responsibility as a researcher not to submit or publicly post work you know to have significant flaws, or allow such work to be posted in your name. In fairness to you, it's also the responsibility of the person posting a preprint to get a go-ahead from all the authors, and it sounds like that didn't happen in the situation you mentioned.

2. If someone habitually does this sort of thing, avoid working with him/her if you have a choice.

Answer 6

This is a great question. Following on from what Dan Romik said about the best approach being to communicate your concerns - this is what I would do, and this is basically how I would do it (I am assuming certain things here, such as the strength of the relationship, or communication culture in the group, so you might want to pick and chose the aspects that work for you).

When picking a time for a meeting, I would make sure that I am as relaxed and cheerful as possible (well exercised and well rested), and that the co-author in question is also in good mental health. To arrange the meeting I would mention that I had something really minor to discuss with them. I would start that meeting by telling them how much I enjoyed working with them and some of the positives related to that experience. Eventually, I would get a point where I would say, however, while our work together has been very rewarding, I have one little concern that I would like to discuss with you. Having come around to the point I would ask if I could be honest with them - they will say yes (people nearly always do). At this point I would explain my concerns with their actions during the writing of the paper. When doing this I would use lots of qualifiers and try not to injure their ego. For example, having explained my concerns I would say things like, it is not big deal obviously...we all make mistakes and I have done similar things in the past etc.. I would also follow up by explaining that I wouldn't raise this concern with them if I didn't value the relationship and want to keep it improving. Finally, after doing this I would ask for their opinion, and ask if their was any feedback that they wanted to give me from our previous collaborations.

If this goes well and they respond to the feedback in a positive way then you will both have a stronger relationship and they may be come a collaborator for life - someone who has shown that they can change and improve and that they care about your opinion. If it doesn't go well then you have probably dodged a bullet as this issue was going to come to a head at some point so it was better to resolve it sooner rather than later.