Routine for effectively proofreading accepted paper

Question

When I am fortunate enough to have a paper get accepted for publication, I often dread the final round of proofreading before I submit the LaTeX source files. Often, I read the paper start to finish multiple times and make incremental/superficial changes, but always have a difficult time calling it properly "proofread". In addition, I find this approach (multiple full proofreads) seems less useful for checking technical aspects of a paper (e.g., mathematical derivations or involved pseuedo-code) which is arguably more important.

I am wondering what types of routines people have developed for this situation. I would prefer to be much more systematic, as my current approach is both inefficient (time-wise) and doesn't seem very effective. Specifically, what do you do in order to call the paper "proofread" and feel comfortable sending it off to never be edited again?

Answer 1

It is very difficult to proofread your own work. You brain knows what you want to say and too often, in my experience, you "see" what isn't written because your mind produces the correct statement. And, my experience is that the harder I try the worse I do. Fatigue? Reinforcement?

My best advice is to make a deal with a colleague in your (sub) field and collaborate by proofing each other's work. Their mind doesn't have preconceptions about what is (should be) on the page, so they are much more likely to catch those things, especially those at the margins, that your mind refuses to see.

I no longer have the opportunity for that (retired) and I often find truly bone-headed errors in my technical writing. I've swapped symbols for "and" and "or" for example creating true nonsense.

A text editor at a book publisher can help with this, but they aren't often used for papers, and they often lack the required technical expertise.

I often find my own errors if I lay the work aside for a while and come back to it, but that doesn't work very well for papers. A week's pause is probably untenable.

Get fresh eyes if possible. Even a graduate student might be able to do it for a bit of cash.

Answer 2

Two things that our group follows for every submission:

(1) Atleast two people read it together (our submissions are always multi-author, so this isn't hard to arrange). This is typically the lead writer and a co-author. It helps because the chance of two persons simultaneously missing an error is low.

(2) Reading it out aloud. Vocalising somewhat takes your brain off the driver's seat, and you're less likely to gloss over things that are incorrect but obviously understood by you. I imagine this would be difficult if there are a large number of equations, but otherwise it is surprisingly effective, and fast.