Biocurious is a weblog about biology, quantified.

Scientists still can't write good

by PhilipJ on 12 November 2007

One of the first entries written for this blog over two years ago was on the sad state of writing in the sciences, and, not surprisingly, little has changed. The September issue of Nature Physics went so far as to include a guideline for writing papers (open access, PDF), largely because “so many papers deserve to be better written than they are”. I couldn’t agree more, but I have ended up frustrated with this editorial, and I’ll explain why.

According to Nature Physics, your paper’s title must be succinct and informative, but should also tempt a reader to at least look at the first paragraph. That first paragraph should set the stage for the “story” of your paper, and why that story should be told to begin with. The story of the paper should then underlie the rest of the article’s structure, and overly technical details should be left to the Methods or Supplemental Information sections.

You should explain, not hype! Avoid cliches “like the plague”, and refrain from unnecessary use of adjectives. Be quantitative in lieu of using superlative prefixes (femtosecond instead of ultrashort, and I doubt you really are studying a quantum nanobiological effect).

This is all sound advice, but vanity journals have really only themselves to blame for the overhyping and the near-ubiquitous quantum-, nano-, and bio-fication of (novel!, which they somehow forgot to mention) scientific results.

Furthermore, pushing aside the actual science into the all-too-brief Methods and Supplemental Information sections is rarely helpful. Methods are never complete enough to reproduce an experiment, and in the case of Supplemental Information (which is relegated to an extra file to download)*, often replete with errors. Real science is technical, and I thought that journals like Nature introduced the Editor’s Summaries and News and Views features so that the basic results of a paper could be read and understood by everyone. We either need to find better ways to incorporate those bothersome, technical details (which are the heart of science, in my opinion) into our science papers, or give Methods and Supplemental Information sections their due with proper editing and sufficient detail. Currently, science articles fail at both.

The best advice given in the editorial, however, is on the final paragraph of a paper:

It is commonly advised that a paper should begin by stating what will be said, continue by saying what is to be said, and then conclude by summarizing what has been said. This is bad advice that recommends lazy composition. Conclusions are not mandatory, and those that merely summarize the preceding results and discussion are unnecessary (and, for publication in Nature Physics, will be edited out). Rather, the concluding paragraphs should offer something new to the reader.

They go on to quote Jonathan Shewchuk:

“Here’s a simple test: if somebody reads your conclusions before reading the rest of your paper, will they fully understand them? If the answer is ‘yes’, there’s probably something wrong. A good conclusion says things that become significant after the paper has been read. A good conclusion gives perspective to sights that haven’t yet been seen at the introduction. A conclusion is about the implications of what the reader has learned.”

This in particular is something I will strive to follow in my own writing, and is by far the best advice of the day. Abstracts already play the role of brief accounts of the paper, conclusions shouldn’t be another repetition.

Unfortunately, the editorial didn’t link to what it considers to be some prime examples of well written science, a real shame given that being able to write a good science paper will only come to a new author after reading lots of well written science papers. If anyone has any examples of particularly well written papers, please share them in the comments!

* I’d like to know the relative number of downloads a supplemental information file gets compared to its parent article!

  1. Andre    3354 days ago    #

    One example that sticks out in my mind is this paper from Paul Wiggins on DNA mechanics. Phil Nelson is also an author and I think he writes beautifully so I imagine he had a hand in it. Regardless, whoever contributed to the text itself did a great job.

    Speaking of supplementary information, the supplementary information for the Wiggins article gives some background and more details of their theory and is also very well written.

  2. such.ire    3354 days ago    #

    If you’ll allow for me to go back to old papers, I think Francis Crick always writes exceedingly well. His papers aren’t just clear; they’re full of character. You can hear him reaching down and telling you how everything was done, why the results were cool, and what they meant.

    Try Crick et al, Nature, 192: 1227-32, where he presents the triplet nature of the genetic code.

  3. Doug    3354 days ago    #

    A title critique:

    From Google [define well]:
    “Definitions of well on the Web:
    (often used as a combining form) in a good or proper or satisfactory manner or to a high standard (`good’ is a nonstandard dialectal variant for …”

  4. sam    3354 days ago    #

    Doug, i hope you’re kidding. or do you not understand what we humans call humor?

    and, as far as Nature including an editorial about good writing, now that is funny. reading a paper in Nature or Science is basically worthless unless it points you to the real paper in a “lesser” journal. an article in j phys chem b might be boring and poorly written, but at least there will be enough info to understand the experiment.

    and supplemental information is for suckers. it should be illegal.

  5. max    3353 days ago    #

    supplemental information isn’t for suckers. It’s great. It should be required that all bioinformatics papers in open access journals add their complete source code as a zip archive as supplemental information. The whole field would gain from it.

  6. unbalanced reaction    3351 days ago    #

    Sam, I couldn’t agree with you more. The Boss insists that our papers be written such that other groups could reproduce our work. Due to the page limitations (specifically in the Methods section) that many of the high impact factor journals place on submissions, we often submit to lesser, more technical journals. The result is better papers (I think), but it sucks that often very important results get relegated to these “lower” journals.

  7. PhilipJ    3351 days ago    #

    Getting your work out there is important, so a Nature or Science paper is excellent for that, but I think it should also be a (group imposed) requirement that a second paper with the details necessary to actually understand the result be published too. Anything in Science, Nature, PRL, etc, is much too short to really get much out of if you want to reproduce anything, or often even understand what the measurement really was.

    With word and page restrictions that exist in said journals, it rarely matters how well you write, it just isn’t possible to convey enough methodology.

  8. ponderingfool    3346 days ago    #

    Supplemental materials for Science can include long, in-depth Materials & Methods. The question is whether the authors are willing to put the time into it.

    The time issue is probably why papers are not well written. Good writing takes time. It is rare to find someone who can lay out their ideas in an elegant way in the first draft. It takes multiple drafts/revisions/edits to accomplish this. The paper is written and then deconstructed and then rewritten.

    All those involved are busy people. PIs, usually the most experienced writers in a group, have to write grants, attend to the articles of other lab members, give lectures, go to meetings, manage a group of people, serve on various committees, review articles/grant applications, and oh yes find time to have a life.

    Needless to say compromises must be made. Good enough to get published is usually the rubric people shoot for. Elegance in writing is a luxury.

  9. PhilipJ    3346 days ago    #

    Compared to the time it takes to do experiments worthy of publication, I don’t think it is a big compromise to take the effort to write a coherent paper.

    What’s the point of a result if others can’t reproduce it?

  10. ponderingfool    3345 days ago    #

    Compared to the time it takes to do experiments worthy of publication, I don’t think it is a big compromise to take the effort to write a coherent paper.

    What’s the point of a result if others can’t reproduce it?


    The point is to get a paper and then send time on experiments for the next paper, and so on and so forth. As long as the results are actually reproducible, having your competition spend time working out little details to get something to work can actually be to your advantage. It is time taken away from them doing the new experiments you are trying. Do I think that is good? No but that is what an overly competitive system can create.

    From the discussions with my advisors, it also seems more and more of the work associated with an article is getting placed in the labs of the authors as journals cut costs. This can not help matters at all.

  Textile help