Saturday, August 22, 2009

Paper Length

Spacecataz says:
To be fair, said former student has been in a political slap fight with a scientific adversary, and this has been the main reason for the ridiculous size of the paper. The other guy keeps demanding more and more in the paper because he doesn't believe said student's results.

This raises a couple of excellent points:

1. How long is too long of a paper.

2. What do you do when reviewers or exterior people keep asking for more and more from the paper.

The first is relatively easy to figure out. Journals have guidelines for this, and the typical one is that papers should be no longer than 10 pages. While this is often difficult to do, it is a very good goal. In fact, some journals penalize you for going over 10 pages to the tune of $100+/page over 10 pages. For a 20 page paper, that is $1000. In publishing land, that is not a HUGE sum of money, but it is costly. For the specific paper that we are discussing, once it is journal format, it will be something on the order of 40-50 pages, which will cost $3000-$4000 extra, just in penalties. Not to mention the color figures and normal page fees, etc. This is looking to be a $8,000 paper. That, my friends, is expensive for a publication.

Next you have to consider how much people can swallow in one bite. A ten page paper is something that can be read in one long pull. Once you good at reading papers (i.e., you can skim it and get the salient points very easily), it is easy to digest a 10 page paper in under an hour. When you go to a paper that is over 20 pages in length, it is excessively difficult to actually accomplish this. A 20+ page paper no longer has a central focus - it has many areas of focus. This makes it harder to skim and gleam the major point(s). The details start to take over, since it is probably chocked full of them. In fact, this is probably the main difference between a 10 and a 20 page paper - the details. In a 10 page paper, there is a central (simple) theme that is conveyed with the minimum number of details to get the point across. In a 20 page paper, it may either have multiple points or one point with an overkill on the details. These details become the primary object of the paper, and the central point loses its importance. Envision forests and trees and all that crap.

On the second point, this is a little more tricky to deal with. Let's say you have a paper and the referee disagrees with you fundamentally. Instead of just saying (or in addition to saying) that the paper is complete crap because it disagrees with their line of research, they say that they don't believe this aspect of the method or that aspect of the method or this detail or that detail. This gets you into a battle about (possibly) trivial things that take away from the central theme of the paper. What they are in essence doing is having you focus on the trees instead of the forest. If they get you to focus on the details instead of the central theme, then you lose readers and your effectiveness to communicate.

I imagine that this may not be intentional, since they may be interested in the details, but often times what I think happens is that the reviewer disagrees with the outcome of the study, so they are gunning for anything that they think is "incorrect". If you same something like, "this term is negligible", they will automatically say that it is not negligible and therefore the entire study is flawed. You then get into a debate about how that term is or is not negligible. Very few people probably care about this argument except the reviewer, but in order to satisfy the referee, you have to address it somehow. This can totally derail the review process.

Now, what to do about it? That is a fantastic question. There are a number of strategies, some of which are effective and some of which are not:

1. Keep the argument within the reply to the referee. This allows you to address their specific issues without detracting from the main point of the paper. You can always add a couple of sentences in the paper on these issues, but, unless it adds to the central theme, don't go into gory detail! Go into gory detail in the response only.

2. Deflect their arguments by trivializing them back in the response. Don't even acknowledge the importance of their argument. This is very risky and could easily backfire. Some referees don't ever want to see a paper again after they have reviewed it once, so this strategy works well on them, but often the referee will only get very angry after a casual dismissal. So, use this method rarely.

3. When I was a Post Doc, the guy that I worked for addressed all of the reviewers comments within the paper, then, when it was accepted and going through the copy editing process, he removed a significant amount of the additions. Don't do this.

4. Add some strong sentences in the conclusions about the assumptions that you have made. State them in the methodology section, and then again in the conclusions that that you have acknowledged that there are assumptions and that they could impact the findings. This way, you still have an extremely tight central theme, but you concur that there could be other findings with different assumptions. This pleases the reviewer because it says right at the end of the paper "I could be wrong". They read this part as the most important, but very few other people will. You please everyone.

The main problem with this is thinking rationally once you get the reviews back. Typically, you are just so pissed off that you are seeing red. You want blood. But, you have to set your emotions aside and think about how to effectively respond to the comments. I am not the best example of this. I get overly emotional about my papers and blow a fuse when I get a bad review (on every paper!) I try to put my feelings aside, but I have a very hard time doing it. I will try harder in the future.

Ok, I have to go and help Crab Mama put rocks around the fish pond....


spacecataz said...

Very good advice, Crabmaster. Good timing as well- I am currently dealing with several papers in review, and reading this is helpful.

Note: the captcha puzzle I need to solve in order to post this comment is "manglo". Man. Glo. Subliminal message?

Space Prof said...

Your option #4 is the best, but there is a 5th option: actually do the thing the reviewer is asking you to do. But, of course, this takes time. And if you really think it's negligible, then you don't want to do this.