71

Inspired by the recent question on list of figures: what is the point of ending your introduction with a paragraph saying "The paper is structured as follows: in Section 1 we do BLAH, then in Section 2 we do BLAH, we move on with BLAH in Section 3, and conclude with Section 4 in which we do BLAH"?

Many papers in my area do it, my coauthors add them to our papers, and I kinda took the habit, too, but I never really understood why. It looks like a poor man's table of contents. After all, we already write what we plan to do in the introduction, and remind the reader in the conclusions.

59

Unlike Dave and Henry, I am going to argue against the style The paper is structured as follows..., though not against inclusion of content serving the same purpose.

Quite some time ago, I took to heart advice by Simon Peyton Jones on this, which can be found here (slide 19), also echoed by Sandro Etale in his advice on writing introductions. In essence it boils down to this:

Don't write The rest of this paper is structured as follows. Section 2 introduces the problem. Section 3 ... Finally, Section 8 concludes. That is most of the time only a waste of paper.

Instead, use forward references from the narrative in the introduction. The introduction (including the contributions) should survey the whole paper, and therefore forward reference every important part.

In a consequence, the position I take on this part of writing introduction is that while it is important to inform the reader about the structure of the paper so that he/she can take the path of few surprises and easily follow the discourse, it should be however done with style. Doing it in such a poor manner as "The paper is structured as follows ..." is simply bad literature. I would argue, that it is better to firstly, clearly state the contribution of the paper and then write up the paper's plot summary leading from the introduction to the culmination of the paper in supporting the claimed contributions and discussion. I mention the references to the individual sections, and sometimes even deeper structural parts, only in passing, or include them in parentheses.

On a similar note, regarding Conclusion section, I took to heart the advice of D. J. Bernstein on writing conclusions. Citing other authors, he suggest to simply drop the conclusions part from papers and put all the important conclusions into the introductory section. Well, unless there really is something important to say there. Since then, I conclude my papers either directly by a loose paragraph at the end of the Discussion section (renamed usually Discussion and final remarks), or if I feel like there is something important to say, by a standalone short section Final remarks. However, never re-iterating what was done in the paper. The reader is anyway free to "rewind/relist" in the paper (see the very last point by DJB).

P.S. The links to homepages of the two guys include many more good tutorials on aspects of writing/research in computer science.

  • 3
    I agree with this. I'd rather have an "annotated table of contents" where you provide forward references to each part of what you discuss in the paper. – Suresh Oct 1 '12 at 20:27
  • 2
    Following Bernstein's advice on conclusions is potentially an easy way to get your paper rejected if the authors' guidelines for the journal to which you're submitting explicitly require a conclusions section. – aeismail Oct 1 '12 at 21:39
  • 4
    @aeismail: of course sticking to such principles as I describe rigidly is dumb if it does not fit a particular situation. But so far, for conference papers and even book chapters, drastically shortening conclusions in comparison to the more standard approach, worked well for me. – walkmanyi Oct 1 '12 at 21:41
  • 2
    However, sometimes you may bump with reviewers that want or desire a "Conclusions" section or a short summary at the beginning. Other advice I usually get is:always check the Introduction and the conclusions, and it is a bit hard if the conclusions are embedded along the paper. – Leon palafox Oct 16 '12 at 5:43
  • 4
    @Leonpalafox The advice was to include the conclusions in the introduction, instead of as a separate section at the end. That makes it easier to check the introduction and conclusions, since all you need to check is the introduction! – David Richerby Aug 31 '14 at 8:29
35

The answer is really the obvious one: it is to alert the reader to the structure of the paper. In addition to this part in the introduction, the paper should also, every now and again, tell the reader where they are in the overall story of the paper. This makes it easier for the reader to read the paper, by giving them the global picture and pointing out from time to time where they are in the global picture.

A table of contents would be overkill (and it would take up too much space).

23

I find these paragraphs very useful (and often include them myself), precisely because it's a short version of a table of contents: often readers don't want to read the whole paper, but want to find one particular part of it. A summary of what's in each section makes this much easier.

  • 3
    @KonradRudolph: Except for all those fields where that's never the TOC. The question seems pretty clearly to be talking about fields where papers don't have a standard structure, and therefore providing some form of TOC in the introduction is useful. – Henry Sep 30 '12 at 21:03
  • Ah hmm. It just occurred to me that I’m obviously approaching this exclusively from a experimental science perspective. I see how it’s not really applicable anywhere else – Konrad Rudolph Oct 1 '12 at 8:55
  • 2
    If you want quick overview, you can just flick through the paper and see the clearly visable headings of each section and sub-section, no need to also list them in the introduction. That might even motivate people to right better headings. – Franz Kafka Dec 18 '13 at 12:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.