How long is a manuscript? All answers are wrong.

This week I got to do one of my favourite things: shorten a manuscript.* This one we’re targeting for a journal that has a 30 page limit on manuscript length. We started at 37 pages, and I was immensely pleased to hit about 29.7 (giving my coauthors just a little wiggle room to reject one or two of my cuts).

But it was a little weird.

The thing is, Goodhart’s Law exists. If you set up a metric and reward people for using it (in this case, for limboing just under the 30 page limit), they’ll do just exactly what you’re rewarding them for – and there will be unexpected, and maybe undesired, consequences. In this case, a decision to measure manuscript length as a number of pages rewarded me for behaviour that reduced page number, but was of no other value (or perhaps of negative value). I did things that wouldn’t actually reduce the length of the printed paper – only of the submission-formatted manuscript. I did things that might shorten the printed paper, but wouldn’t actually reduce the time it takes for someone to read it. And (I feel dirty admitting it) I may have done a thing or two that would shorten the paper, but actually lengthen the time it takes for someone to read it.

Here are some of the page-limit tricks I used, and couple that I bravely resisted.

  1. Hyphenation packs words in nicely. I turned on hyphenation and immediately saved ¼ page, with absolutely no other benefit to anyone.
  2. Kerning (spacing between characters). I have to admit that adjusting the kerning only saved a line or two, but I got to feel proud of myself in a very sophisticated way. Related but less sophisticated things like adjusting font size, margins, and line spacing were (unsurprisingly) foreseen and disallowed by the journal.
  3. Combining two paragraphs into one can save a line even without a cut, if each paragraph has a partial last line. Sometimes this is the right thing to do – but paragraphs are so powerful as a way of guiding readers that we should be skeptical of messing with them.
  4. Speaking of those partial last lines: when a paragraph’s last line has only a few words, a little trim can save a whole line. But if the last line is long, there’s no incentive to trim further. Result: every paragraph’s last line runs right to the margin.
  5. Because the Literature Cited consists of very short paragraphs for which tricks #3 & #4 don’t work, there’s a strong incentive to remove citations.
  6. Each Table and Figure is required to be on its own page. Therefore, combining two readable tables into one monster unreadable one could save an entire page; ditto for two figures that could be panels of a single figure. Whether or not they should be. (Be proud of me; I resisted the siren call of the 18-panel figure).
  7. Figure legends have to be their own page, so there’s no incentive to shorten those. Worse, there’s a temptation to move Methods details into them, which moves text from a section that “counts” into one that doesn’t. (Be proud of me again.) The Acknowledgements don’t have to be their own page, so those get cut down.

Now, I’m sure you can think of a dozen other tricks I could have used.** The point is that these tricks have nothing directly to do with improving the paper.  They’re focused entirely on saving pages; and if the length metric were words instead of pages, I’d deploy a whole different set of tricks. Hyphenation, for example, doesn’t cut words; but replacing two short words with a single long one does (despite the disservice it does to the reader). There are other ways to measure length – character counts, and more – but I can game them too. All ways of measuring length are wrong.

I completely understand the drive for brevity in papers. Publishers want short papers so they can publish more of them; readers want short papers so they can read more of them; and authors should want short papers (up to a point) because they’ll communicate more clearly with more readers. The devil, as always, is in the details!  And that’s one reason (of many) why the American Naturalist is one of my favourite journals.  In the Am Nat, it’s simple: “papers should be as long as they need to be to make their case concisely and effectively”.   As long as they need to be, but not longer. Now that’s hard to game.

© Stephen Heard  July 21, 2021

Image: A passage from the manuscript I was shortening (Isitt et al., in prep.). The changes illustrated cut three lines. How many words? I don’t know; I didn’t care.

*^Yes, I’m a little weird. But there’s something satisfying at watching a word count (or page count) slowly decrease, inching towards a clearly drawn goal. And if there’s one part of writing I’m good at, it’s cutting. Not in my first drafts, which are far too wordy – although that’s deliberate, because I’d rather just get something down that I can work with later. And not here on Scientist Sees Squirrel, where I admit to indulging myself a little in writing a bit more expansively. And definitely not in this footnote, which appears to be interminable. But you’re still here. Why on earth are you still here?

**^Moving text and figures into Supplemental Materials is an obvious one. I kept it off my list because it’s less tied to a page limit – Supplemental Materials can shorten to pretty much any metric. Partly as a result, excessive Supplemental Materials are arguably something of a plague.



2 thoughts on “How long is a manuscript? All answers are wrong.

  1. John Pastor

    Reminds me of a saying attributed to (who else) Albert Einstein: “A theory should be as simple as it needs to be, but no simpler.” Or something like that. I think he was arguing against excessive dependence on Occam’s Razor to justify one theory over another.

    Liked by 2 people

  2. Pingback: Does your scientific paper smell? – Brushing Up Science

Comment on this post:

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

You are commenting using your 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 )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.