Title: Some Thoughts on Academic Papers

As the W3C's MathML specification says, "The academic research community produces large volumes of dense scientific material". This material is in the form of papers, be they in special-interest journals published by university departments and other research establishments, or in the proceedings of conferences. The prestige of a researcher, or of an establishment, seems to be measured by the quality and quantity of their papers, as well as by the eminence of the journals and conferences that consent to publishing them.

Writing a paper is often a matter of describing a problem, citing other people's work on it, describing what you've done, and saying something interesting in conclusion, all within the word limit. You also have to write an abstract (a brief summary of the paper that people can use to decide whether or not they want to read the full paper), and possibly choose a small number of keywords for indexing. Since there are an enormous number of published papers out there, your choice of words can be very important, particularly in the title, abstract and keywords, if you want to catch the attention of the right people. The TEX family of programs is of enormous help because it can automate many of the formatting conventions.

The process of writing a paper, finding a suitable journal or conference, and getting it in, can take time, and I wonder how many papers are obsolete by the time they are published. Some researchers work by first looking for interesting conferences and then writing papers for them. The scenic location of a conference is sometimes a motivating factor!

Often, the only editorial decision is whether or not to publish; editors cannot change your words. Papers written by people who are not fluent in the language, for example, have all the grammatical and stylistic problems intact. This can, of course, happen to fluent authors too, particularly if they are forced to reduce the word count under a time constraint, which usually sacrifices readability. Often papers contain ambiguous statements that stumble the native reader because their true meaning is not the obvious one. Many editors would not dare try fixing these even if they were allowed to, because they might not be confident that their interpretation of the text is correct. In some cases, editors can draw your attention to things they think are problems, perhaps saying that they will not publish unless you fix them, but you have to be the one who fixes them; if a paper is published in your name then it implies that you had final control over every single word.

One exception is a journal that offered to publish a cut-down version of my undergraduate dissertation as an academic paper. I wanted to publish as soon as possible in order to stop anyone from taking out a software patent on my project, but I was tied up with final exams. Their editorial team kindly offered to do the cutting-down for me. Unfortunately they introduced a number of errors in the process, and since my final proofreading had to be rushed, I'm not convinced I caught them all. I hope the consequences will not be too bad.

I have not yet seen a paper that does not have quite a number of references in its bibliography. Citing other people's work is often considered necessary, and it also helps to keep the word count down because in many cases you can point to a previously published explanation of something rather than having to write another. The assumption is, of course, that the reader will have no trouble in following up the references. Most references are to other academic papers, and indeed many researchers find related work by performing a "citation search", that is, given a paper that interests them, they search for other papers that cite it. Increasingly, papers are citing Web pages, although these are often transient (besides anything else, the author might have to change ISP), so it is better to find a paper to cite, or at least find a URL that is unlikely to disappear. Citation is usually a sign of respect, so I was amazed to find that ACCU's Web Access Gateway has been cited at least twice by IBM researchers in the proceedings of the ACM (ASSETS 2000, November 13-15, Virginia, pages 171 and 179).

Finding material to cite can be difficult for the new researcher. However, if you can locate the proceedings of a conference, or even a single paper, that is relevant to your area, then you can find more material by following references and doing citation searches. You might then be taken aback by the sheer number of papers that have already been written in your area; it can be disappointing when you thought you had a totally new idea. However, I'm told that this is the sign of a promising area of research. Many researchers build personal databases of references that they may want to cite, and you can of course cite your own work if it has been published earlier. Given, though, that life is too short to read every paper, I do wonder how many researchers out there are inadvertently duplicating each other's efforts because they don't know that another area has relevance to theirs.

The copyright of an academic paper is usually retained by the publisher, but non-commercial distribution is usually acceptable. However, some organisations prohibit you from putting your paper on a Web page (or other server) without permission; presumably you have to say "contact me if you want a copy", although I haven't seen any researchers actually do this.