15

A section in a paper I'm authoring goes something like the below:

We want to do x. Upon first inspection, x could be easily done using approach A.
A, however, has the drawback that <elaboration of A's drawbacks>. For this reason, we propose B.

The <elaboration of A's drawbacks> is rather long though, making me question whether or not to keep it in since I'm dismissing this approach either way. Is it best to have this in or to just assume my reader will understand why B makes more sense than A in cases like this?

6
  • 1
    Do you have a character limit or why do think it is a problem to elaborate on the shortcomings of A? Nov 16 at 13:03
  • 3
    Can you cite a paper which describes A in details?
    – Nobody
    Nov 16 at 13:07
  • 3
    @Sursula I do not have a character limit. I ask because I have previously gotten the feedback (about a different paper) that I can be unneededly verbose at times, so I am trying to be more mindful about this and was wondering if this was a case where this might apply. Nov 16 at 13:09
  • 8
    @ElliotSolskjaer If you say that As shortcomings are not obvious, never assume that readers will reach this conclusion on their on. Being verbose and and explaining vital information are two completely different things. Nov 16 at 13:15
  • 3
    If the discussion of method A is in a well-delimited section titled "Drawbacks of method A", it is easy for a reader who doesn't care to skip it, so you can still include a long discussion for those readers who will care. Nov 17 at 8:41

7 Answers 7

38

My suggestion is that you write the paper as you think it should be written. If you get negative feedback from reviewers then deal with it then. But there is no reason to anticipate what the reviewers will suggest if you have reasons for your arguments. Expect that some changes will need to be made, but look at the reviews first.

Just. Do. It.

2
  • 5
    Ah, you got the ✔️ for saying "Just do it". You may want to contact a certain company for royalties for your advertisement of them! :D Nov 18 at 10:02
  • 2
    @AnderBiguri, funny you say that. I noticed that the checkmark looks a bit like a swoosh.
    – Buffy
    Nov 18 at 11:27
21

It sounds like you're potentially entering the pitfall of trying to write the paper following the exact thought process that you followed in your research. You may have first considered doing A, but then realized that that wouldn't work, and then thought about using B. Often the benefit of hindsight gives much more clarity.

In this case, you end up spending a lot of time and space describing what you didn't do, before getting to the important bit of what you did do. In practice this may make it harder to follow. It could be better to start with saying "we did B". This of course may raise the question "why didn't they just do A?". If this is an obvious question, then it should be addressed. It may however be better to do this after you have introduced B. This may give the benefit of being able to explain why/how doing B addresses the drawbacks of doing A, while explaining what those drawbacks are.

2
  • 11
    I more often see people enter the opposite pitfall. They write the paper as if they had magically guessed the final approach directly, and while the result is there, the paper fails to give much intuition and understanding. Of course there is a balance, and it will depend a lot on each case, but in a number of cases it makes sense to start with an explanation of why A fails and how that suggests that B might work better. Nov 17 at 8:48
  • 4
    Starting with "this is why A doesn't work" can be a fine option as long as you put it in a section of its own so people who don't care can skip it.
    – Javier
    Nov 17 at 14:40
9

This is typically what Appendices are for: to detail things which are not necessary to understand the paper, but could be useful or interesting to the reader.

Example:

We want to do x. x could be easily done using approach A, which however, has some drawbacks, among which 1 and 2 (see Appendix n for details). For this reason, we use B.

1
4

My advice is to identify what the challenges or requirements are and then explain how B meets them (and other methods do not). I encourage you to focus on the positive -- what you are able to achieve -- rather than the negative -- the ways that other people have failed. Also, in many cases I find it is helpful to first list the requirements or goals or challenges, then explain how your approach meets those requirements or goals or addresses those challenges.

Bad: "We use method B."

  • This is bad, because there is no explanation of why you have selected B.

Bad: "We use method B, because method A is flawed."

  • This is bad, because it does not explain concretely why B is better than A.

Better, but not ideal: "We use method B, because method A has problems P, Q, R."

  • This is better, because this provides some explanation of why you selected method B and specifically what the advantages are, but not ideal, because it is focused on tearing down someone else's work, rather than on what you're trying to achieve.

Best: "Ideally, we would like an evaluation method that has properties X, Y, Z. We ultimately chose to use method B, because it provides all three properties. Others have used method A, which provides property X but does not provide property Y, Z. In our setting Z is particularly important."

  • In my opinion, this is best, because it is focused on the positive - what you are trying to accomplish and what you have achieved; and does so in a way that explains why you chose the method you did, while treating prior work with respect.

  • If there are multiple options for evaluation methodology, you might even consider a table where you compare all the options: you can have one row per candidate method, one column per requirement or desirable property, and checkboxes in the cells indicating which methods meet which requirements. This can provide a visual summary that makes it easy to quickly understand why you have chosen your particular method. You should probably supplement it with a detailed definition of each property and discussion of why each property is relevant and important.

If some of the explanation gets too detailed or lengthy and is distracting, one option is to put that detailed explanation in an appendix, and summarize the main idea in the body of the paper with a reference to the appendix for further elaboration.

2
  • 1
    The asker refers to the problem that X, Y and Z are rather long, e.g. some property that might not become apparent without working through half of the explanation of A.
    – user253751
    Nov 17 at 15:04
  • 2
    @user253751, I understand - and I thought I addressed that situation in the last paragraph: "If some of the explanation gets too detailed or lengthy..." Is there some revision to my answer that you would suggest to improve it?
    – D.W.
    Nov 17 at 17:59
3

Assuming that A is a well-known approach and B is something which will be unknown to people using A, you should keep your paper exactly as you propose. Providing a description of A (and its drawbacks) gives your readers a reference point and a good feeling of whether they may be interested in B.

Let's assume that A refers to using a microwave oven, and X is an egg. You tried to use A on X and discovered that X tends to explode when exposed to A. Then you discovered an alternative approach B (a steam cooker), found that it works on X rather well, and wrote an article about it.

Now imagine that I'm a fellow researcher who tries to work with Y (tomatoes) and I'm confronted with a similar problem when using A on them. If I see an article about A not working for X, I will be really interested in trying out B, because the problem you have solved is quite similar to mine. On the other hand, I will likely skip an article about steam-cooking eggs as irrelevant, or even not see it at all because I will search for "tomatoes", "explosion" and "microwave", not for "eggs" and "steam cooker".

0

Following up to @Clef's answer, if you're writing a conference paper which has a short page requirement and can't add an appendix easily, people often cite a longer version of their paper on something open-publication like arxiv:

We want to do x. x could be easily done using approach A, which however, has some drawbacks, among which 1 and 2 (see an extended discussion in the Appendix of [citation of arxiv version] for details). For this reason, we use B.

Of course, this doesn't work if the paper is supposed to be double blind.

0

Depending on the context, it may be possible to recast the text so it focuses on the problem/task rather than the method. This could help clarify a non-obvious aspect of the problem, without getting too bogged down in details of a method you are not using in your paper.

For example:

"We need to solve problem X. At first glance, problem X looks like a standard example of general problem class Y, which has a known solution A. However, problem X actually has unusual feature Z, which presents an obstacle to using method A in its standard form. Therefore, we use method B to get around this obstacle."

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .