31

Having just spoke with my advisor, and speaking on occasion with other professors, I have come to the impression that one should not "subpublish." Let me say what I mean. I'm a 3rd year graduate student right now, and the professors I talk to seem to say that it is not a good idea to publish many small papers along the way to a result: that is, holding fixed the total mathematical content across N papers, it is best if N is minimized. I was trying to understand why. My advisor cited the reason of "reputation" and I have invented my own reason that perhaps there is an upfront cost to the process of writing and submitting a paper so that although effort per length may be constant, it is to my advantage to pay the upfront cost as few times as possible. (Finding a referee is the only upfront cost that comes to mind, but keep in mind I've never done this before.)

Can someone expand upon how reputation is relevant here i.e. for fixed mathematical content why it looks bad to come out with many small papers? Are there also benefits to the community or the knowledge pool that I'm trying to expand however slightly by writing such papers if it all comes at once in one big paper? One would think that if I "subpublish" that actually since my results become available more immediately that it would benefit the community if I did so. The only person it would seem to possibly hurt is me if I were to get scooped.

Are there other ways in which upfront costs to the whole process of publishing may happen?

I am interested in all my questions in both advantages and disadvantages to me, and also to those who would use my work.

2

6 Answers 6

15

To address the reputation aspect:

Think about a typical reader of your papers. Most likely they will only look at one paper you have written. Unless you impress them with that one paper the likelihood of them reading anything else you have written depends on whether that first paper convinces them it is worth their time to read another. So the more content that is in each of your papers the more impressed the reader will be with you. If the first paper of yours someone reads has a small amount of mathematical content then they probably won't expect too much from any of your other papers.

In your formulation the ideal N may not always be 1, but each paper should actually have something worthwhile in its own right to say. You shouldn't expect someone to read more than one paper to get to only one meaningful idea.

A side comment, the author does not find referees. The editor does. Sometimes the author may suggest candidates but the editor or the suggested referee can always say no.

18

In math it's generally better to aim for fewer, longer, better papers. There's nothing wrong with publishing short papers if that's what best fits your discoveries. However, given the choice you should not publish your results piecemeal as you derive them, but rather try to collect and polish them and craft coherent, substantial papers. There are several reasons for this:

  1. Publishing partial results piece by piece makes it harder to write understandable papers. A few papers later, you'll wish you had done things a little differently, and you'll start writing things like "by the same idea as in the proof of Lemma 3 in [15]" because the statement of Lemma 3 doesn't quite say what you need now. But it's too late to change it, so your next paper is either cryptic or repetitive, and neither one impresses readers. Or you find an ad hoc way around it, but your paper ends up being a little less natural. Basically, this is a form of technical debt. Once you have enough technical debt, the only good solution is to write a clean exposition from scratch. Every topic eventually reaches this point (which is one reason we have expository papers and books), but in the meantime you'll minimize the problems if you gather related ideas in one place and write them as a coherent unit.

  2. If you break your work into papers in unnatural ways, you risk looking desperate for publications. Each individual paper may look like you don't understand the big picture or don't think your work is going to amount to anything substantial. These are not messages you want to send to readers.

  3. If you want a serious research career, you need to publish in good journals. Taking a long paper in a prestigious journal and breaking it into three short papers in less prestigious journals is bad for your CV. Experts might know those three papers amount to something substantial when combined, but a non-expert looking at your CV won't be able to tell.

  4. Many people in your subfield will form an opinion of you without having studied your work carefully. For example, plenty of people who have never read any of your papers will see you give talks at conferences or will see references to your work. Because they don't have a global context for your work, to a first approximation they will judge you by how good they think your average paper is.

One would think that if I "subpublish" that actually since my results become available more immediately that it would benefit the community if I did so.

If you reach a nontrivial milestone that is genuinely exciting or useful for other people, then that could be a good reason to publish a paper now, rather than sitting on the result while you try to complete a larger project. On the other hand, there's much less value in presenting a steady stream of partial results taken out of context.

Of course career pressures sometimes interfere with ideal publishing. If you're going on the job market soon, then writing a suboptimal paper now is probably better than writing an optimal paper later.

11

(My answer somewhat overlaps with previous answer, but contains parts that in my opinion need to be said. Beware that I am a fundamental mathematician, and that my advices should be taken with extra caution in other fields.)

I would say that in most cases a good way to go is to get each paper tell one story. This both means that the story should be complete, and that there should not be two stories in it.

Now, in many cases choosing the perimeter of a paper is a subtle issue and your interest and the interest of the community do not necessarily align completely.

Pros for publishing several smaller papers:

  • It can be difficult to get a long paper published in a good journal, because of a limited number of pages published each year (this sounds weird at a time when most papers are distributed electronically, but many math journals have this issue). For example I had a long paper, containing different independent results on a common object, rejected by a journal after the referee found the results nice, but the paper too long for its worth. Maybe one of the result would have been equally sexy on its own, and the paper would certainly have been much shorter.

  • Your CV will have more lines, and this can sometimes help. Every hiring committee I attended to put an emphasis (too strong in my opinion) on the number of papers of each applicant.

  • If there are different results that may be of interest to different communities, putting them in one paper mean that they will appear in only one journal, maybe not read by both communities, and that there will be only one title and one item and review in each database (MathReviews, Zentrallblatt). This can make one or both of the results more difficult to find, less visible, therefore less useful to the community.

Pros for publishing less, longer papers

  • It can be difficult to publish a short paper in a good journal, as it might appear as not tackling a challenging issue. Of course, a short paper solving a long-standing open problem will be easily published, but when you have to convince an editor and a referee that your question is good and that answering it deserves merit, short paper may (unfortunately) hurt you.

  • Long papers can impress people on CV. I have heard in hiring committees remarks like "she publishes 30 pages papers, this is serious work" or "Ok, he has a lot of papers, but he mostly seems efficient in maximizing the number of CV lines from little mathematical content". If your introductions are all the same, it can look like salami publishing; in our publish or perish era, one is expected to salami publish without looking like he or she is salami publishing.

  • Sometimes, two or more results complement each other and are together worth more than the sum of their individual worth. In this case, a unique large paper may be much better than several smaller papers that individually will look minor.

Concluding comment

You see that I gave each argument both ways. This is why I think the issue is subtle, and that each case has to be considered closely. The "one story" guideline can help but is not very precise. Finally I would advise to try to maximize the benefit of readers (remembering they have limited time); try to imagine what a referee that does not yet know your work will feel reading your paper, and make your choice so that he or she thinks "what a nice result!" (the result needs to be somewhat impressive by itself, but also to be clear and clearly delimited). In most cases and in the long run I believe it will also be close to maximize your own benefit.

3

This is highly field specific. In many STEM fields, multiple small papers are preferred because the field moves very quickly and people would get scooped if it took them 2-3 years to write a paper. This is much less of a concern in mathematics. Most mathematicians are very careful and take their time to write thorough and complete larger papers.

1

In some situations, people will judge you by counting the number of papers you have written. Administrators in funding organizations or universities might do this, for example, or HR people working for job search agencies or prospective employers. They won't read the papers, and they wouldn't understand them even if they did. So, in these situations, more papers is better.

6
  • 2
    These people will also look at which journals the papers were published in and it's important to them that you're publishing in good journals. The top journals aren't interested in publishing the mediocre papers that result from slicing what should be one paper into several parts. It's much better for your career to have one paper in a good journal than three papers in a poor one. Feb 15, 2014 at 10:45
  • 2
    @DavidRicherby In contrast, I've gotten several papers in good journals by separating out a project - the notion that a larger paper is inherently better is one that is not invariably true.
    – Fomite
    Feb 19, 2014 at 16:37
  • @Fomite Of course. If a project contains several units that are relatively self-contained and each is interesting in its own right, it can make good sense to publish each unit as a single paper. Feb 19, 2014 at 17:12
  • 1
    @David -- in my experience, HR people working for job search agencies or prospective employers will not care which journals you published in. They might be influenced by impressive-sounding names, but not much more.
    – bubba
    Mar 5, 2014 at 2:45
  • 1
    @David -- yes, of course. But, I assume "Academia.SE" is for people who are currently in academia, not just those who intend to remain in academia for ever. I thought it would be obvious from references to "HR people" and "job search agencies" that I wasn't talking about jobs in academia.
    – bubba
    Mar 23, 2014 at 7:45
0

@Jeff-Publishing your work in fewer larger papers or more smaller papers depends on your level, which means that when you are in your formative years, I recommend you to publish your work in more smaller papers and when you are in your summative years, I recommend you to publish your work in fewer larger papers

0

You must log in to answer this question.

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