Could you spare me 30 seconds of your time?

Or, in defence of the 4 page paper.

As anyone who knows me can probably attest, I can sometimes passionately defend one position and then change my mind completely and hold the opposite position just as fervently. This post is an example of one of these changes of mind…

Today I want to talk about the four-page paper.

The four-page (plus/minus epsilon) paper is almost a cliche in physics on par with that of the three-part novel of the 19th century: when it’s four pages long you can submit it to PRL, or, maybe with some extra work, Nature or Science. And these are gold-class journals: if you get enough PRLs then this is likely to gain the respect of you colleagues and, more importantly, impress hiring/tenure/promotion committees. So it’s not hard to see why the four-page paper is an attractive format to aim for. I don’t want to dwell on the rightness or wrongness of the academic system here. Like it or not this is the way things currently are, and I’m not going to try and change it right now. (That’s a long post for another year.) I just want to focus on what four-page papers are doing to scientific scholarship.

Much has been said about how the dominance of the four-page paper is a “bad thing” for scientific scholarship. Well, actually, at least, I’ve heard lots of people complaining about it (including myself). Now that I mention it I can’t find anything actually written down; maybe noone has the courage to do so? The brief summary of these criticisms is that four-page papers are, variously, incentivising authors to: (a) chop up perfectly good longer papers into many four-page “epsilon” papers; (b) “hype up” their results by talking about how revolutionary they are; and (c) forcing authors to crunch lots of valuable calculations into a ridiculously abbreviated form.

I hear comparatively fewer people talking about what’s good about the four-page paper. I wanted to set down some ideas here that I feel aren’t entirely fairly acknowledged in over-dinner rants. My actual position now is that the four-page paper format has it’s flaws, but isn’t *all* bad.

1. I agree, without question, that a paper without length restrictions can only be a better paper. Why? Well, introducing constraints to an optimisation problem can never improve the optimal value! But, sometimes, constraints force us to be more creative: e.g. vegetarian food can be much nicer. By being forced to communicate an idea in four pages you can be inspired to be much more creative…

2. Four-page papers are easier to read. I am a very lazy person and now I have less and less free time. I typically don’t even download papers which are 30 pages or more, I just look at the abstract. I don’t think I’m alone in this. The point here is that reader attention is a finite resource: four-page papers are not a major commitment, and could be skimmed rapidly for relevance. Longer papers require more energy. In this world where you have to compete simply for the attention of the reader you have to acknowledge that a little bit of marketing is in order.

3. Supplementary material! You can probably say the main idea in four pages. If you want to include the details use the supplementary material. This counts as four pages for me: the idea and argument are still expressed in four pages. The details are there for the interested reader.

My feeling is that the four-page paper functions extremely well as an extended abstract rather than a proper paper. But often an abstract is all that is required to inspire and motivate further research and communicate a new argument/technique. (I’m reminded of the tale that Landau would only read the abstract of a paper and then rederive its contents… I don’t think this is so unrealistic in many cases.)

Advertisements

7 Responses to Could you spare me 30 seconds of your time?

  1. I agree that 4 pages is perfect for an extended abstract. I also like this format. The problem is that PRL is definitely not like this at all. As it has become increasingly difficult to submit there (probably a good thing) people try to cram more and more results into the 4 allotted pages so that the paper isn’t rejected as not being important enough or not having broad enough appeal. But then they don’t bother to submit any supplementary material! This makes some PRLs nigh impenetrable. (I’m guilty of this myself, unfortunately…)

    This is much less of a problem with Science or Nature, but then again the fraction of papers in our field that are sent there (especially theory results) is pretty small, so they are negligible.

    • Joe Renes says:

      I’ve been coming around to this point of view myself, having dabbled at submitting stuff to nature physics. Took me a while to get the habit of discussing matters at some length without explicitly justifying the logical steps involved. Is supplementary info available at PRL in the same sense? That is, can one stuff all the details and proofs into the supplementary info, or do the four pages have to stand on their own? By the crammed nature of PRLs, I guess everyone assumes the latter.

  2. tobiasosborne says:

    Hi guys!

    Many thanks for your comments!

    I totally agree with the criticism that “people try to cram more and more results into the 4 allotted pages so that the paper isn’t rejected as not being important enough or not having broad enough appeal”.

    This is one of my pet dislikes, actually: when we do this (and I say “we” because I’m certainly no saint here!) it rather smacks of insecurity. Indeed, I am becoming more and more suspicious of a 4 page paper with 10 seminal ground-breaking results. Really, I’d love a 4 page paper with just *one* result that I could get clear intuition about, and fully understand.

    (Note to self: if I feel the urge to put in extra results then maybe I’d better face up to the fact that the paper is a bit weak.)

    Joe: I understand that PRL has recently changed their supplementary material policy, i.e., there is the possibility of depositing EPAPS (see e.g., http://forms.aps.org/author/epapsinstr.html).

    This is a *good thing* as far as I’m concerned.

  3. Joe Renes says:

    Thanks for the link. It’s a bit vague about how the paper is supposed to be structured, but I think we can all agree that gory proof details are “… of too limited reader interest to warrant publishing in full in the journal.” Do you know any examples where the main text probably wouldn’t warrant publication?

    I think it’s vital they move to this sort of format, or else their publications will just get more compact and less readable as you and Steve mentioned. Sometimes people use PRL as an extended abstract for a different paper (i.e. the longer, complete version), but there’s really no excuse for that now. Not that there really was before.

    My hope is that people will write fewer papers: more big ideas and less incremental research. And feel free to organize their papers on logical grounds (one idea at a time, as you mentioned), not collections of results which cross the appeal/importance threshold. While I’m in a wishful mood, I’d also like to win the lottery. 8 figures.

  4. Dave Bacon says:

    There are two conflicting places where PRL is good. One is for cute simple theory results, or precise well defined experimental results that really do fit in 4 pages. The other is where the 4 pages are an executive summary of a deeper longer result or of a new experimental method, etc. PRL does fine by the first of these, but for the second it seems a disaster. EPAPS is a great idea, but in practice doesn’t function very well. For example I had a paper where a referee didn’t even know there was stuff to read in the EPAPS! Plus, unlike Science and Nature this extra material is just dumped (raw LaTeX!) into an accessible location. No editing is done on this. I think if they could allow long papers with executive summaries for these types of papers they would be a lot better off. For example it would be great if I could write a PRA and if the result is interesting enough the editors could pick it out for PRL summary.

  5. aram says:

    Physicists like to hate on the CS conference model, but you are talking about a breezy extended introduction where you describe your technical contribution at a high level, the relevant previous work, and the important consequences of your work, before the “notation” section, and then the proofs.

    I am trolling a bit, but after Ashley and I rewrote 1001.0017 in FOCS mode, we decided it actually read better with 17 pages of appendices.

  6. tobiasosborne says:

    Dear Joe, Dave, and Aram,

    Many thanks for your comments!

    Concerning:
    “EPAPS is a great idea, but in practice doesn’t function very well. For example I had a paper where a referee didn’t even know there was stuff to read in the EPAPS!”

    My suspicion is that referees will take a while to get used to this. Probably once they have all tried to submit some EPAPS themselves this will ease up a bit (I hope!)…

    “Do you know any examples where the main text probably wouldn’t warrant publication?”

    I now know of a couple of examples where this has begun happening. The only one which springs to mind at the moment is:

    http://prl.aps.org/abstract/PRL/v105/i15/e150401

    “For example it would be great if I could write a PRA and if the result is interesting enough the editors could pick it out for PRL summary.”

    That’s a great idea! I really like this: the chance of an “upgrade” would surely make PRA a much more attractive (and possibly better) journal!

    Aram: your suggestion doesn’t sound entirely like trolling to me. I guess I think of the CS model as what would happen to a math discipline with a PRL-type option 😉 (ok, so I’m trolling now…)


    anyways

    I’m not entirely sure if I communicated my vision in the post so well…

    My vision for what I think would be a *great* journal is the following (based loosely on the 3-4 page format).

    Here are the rules: you’re meeting a physicist friend from another discipline (e.g., astrophysics or something if you are quantum info.) in a cafe for 1 hour. You have access to exactly 1 napkin measuring 15cm by 15cm. You are allowed to write on this.

    Your task is now to communicate your latest and greatest result to your friend using only the napkin and the one hour. Your friend is skeptical but not combative, so they’re willing to accept physical arguments in lieu of rigourous proofs.

    My ideal is a journal which publishes the transcript of such a conversation, in, say latex form. You’re allowed to attach an appendix of whatever you want. But the bit which is assessed is the transcript plus napkin.

    I find that this is the format where I learn the most from others. I think most results can only be made more accessible if such things were published.

    My feeling is that one of the goals of PRL etc. was/is/should be precisely to provide a venue for slightly more formal version of the above description. I.e., accessible presentations of ideas which aren’t bogged down by too many equations…

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: