Louise's thoughts

Writing papers

As you probably know I am writing papers right now. I have to admit that the 's' is what makes this more crazy than usual. Somehow they decided to place the paper deadline for my core-conference and my newly adopted 3rd favorite conference within a week of each other. And since I am this ambitious, hardworking, mad woman, I plan to submit to both of them. Where the two first describing characteristics are my own, the latter is actually a term that I have been called twice this week by the same person. I secretly wish, at least one of the times, he would have used one of the two first characteristics, but considering who he is, I take both as a compliment. So I am writing two papers at the same time. Actually three, but that is a different story.

After a week of franticly writing, erasing, re-capturing data and writing some more and then restructure the whole thing, I am now not only the happy owner of an almost finished paper submission, I also learned how paper styles differ vastly among researchers, even within the same field. When I was at Berkeley I learned one style of paper writing, which definitely paid off, because I got two good publications out of it, but this week taught me that there is never one way of doing things. To sum up the types of papers one can write for a conference, there are roughly two styles: The 'look what we did, see what we found, so we can conclude this' and the 'this is what we should consider so we did that, this is what we found and these are the consequences for the field that these findings have'. Instead of mentioning names I will call the first approach the Berkeley paper and the second one the Irvine paper.

The Berkeley paper follows a strict line of presenting the system/case study/architecture/algorithm that was developed or conducted. First we say what we will be saying, then we say the things we need to say and then we will say what we just said. This paper also have a strict order of sections such as introduction, related work, method, results, discussion, summary and/or conclusion, future work and acknowledgments. In the beginning of each section the purpose of the section is stated ('we say this in this section because of this and this') and in the end of each section, a summery of how far we are in the argument is stated ('this is what we have seen so far, this is what this section was about'). The Berkeley paper is therefore really easy to read and it is good for presenting systems that you have built or case studies with a limited array of findings.

The Irvine paper on the other hand tries to tell a story. We don't want to give away the punch line before the end and it builds up an argument before concluding on eminent points. The paper does not need any predetermined sections other than introduction and conclusion. The introduction is an appetizer to what we will be saying and what we have done. If it fits the purpose a review of some relevant literature is a section of its own, if not, the related literature is intertwined with the real material. No matter what, the material is continuously related to other literature. If a study has been conducted or a system developed this is presented, not unlike the Berkeley model, but where raw findings are essential for the Berkeley paper, the material can be more spread out here and colored with analytical insight. Again, we want to tell a story. Finally, it is essential to have a good conclusion which is not just a summery of what we found, but actual extra insight and concluding remarks in relation to the issues dealt with in the paper and the overall consequences for the field.

Needless to say I like the Irvine paper much better than the Berkeley model. This is what I always hoped that research could be about. Don't get me wrong, the Berkeley paper has its advantages, especially when it comes to reading speed and being merely a presentation of some new technology. But as I usually say when people ask me why I am in this soft and boring (their words) area of computer science, HCI: I think humans are much more interesting than computers. And likewise, I find stories so much more interesting than presentations.

Posted by Louise at March 13, 2004 08:55 PM