What is the best way to write a research paper? More precisely, how to communicate the research you do or have done in written form? Obviously there will never be a conclusive answer. Styles differ dramatically depending upon the person. And different approaches can be successful. Nevertheless, there are certainly a few different natural alternatives, all with their own traps and pitfalls. Here I shall try to just give a snapshot of various approaches in the hope that it helps others to reflect upon the practical process of paper preparation.
The main decision one has to make is regarding timing. When do I even have to start trying to write results up? Personally, I believe there are two philosophies:
a) Write-once-finished (let’s call it “wronfi”): The paper is started once the research is finished, i.e., you deem the result you have obtained to be useful and relevant enough for a paper. Then you collect everything you need such as notes, data, simulations, etc and try to compile it into a coherent manuscript.
b) Write-as-you-go (let’s call it: “wraygo”): Every part or result is written up as it is discovered. This may sound like a mad idea. Just immediately trying to pin down everything in a paper format once it is discovered.
In fact, I believe that wronfi and wraygo both have merit, both are called for at times. In principle one may always tweak an experiment or a mathematical generality for an arbitrary period of time. There is always something to improve. Hence, following the wronfi philosophy completely, one never even starts to write a research paper. Writing up any little intermediate result daily is also dangerous. It is possible to get lost in little details. Switching between writing and doing an experiment or calculation can be time-consuming as one has to switch gears almost hourly. Therefore, following the wraygo approach blindly is not going to lead to acceptable results.
Personally, I believe that one should find out, which mix of the two approaches suits your personal style and working hours best. This process should not be taken too lightly. Doing it well usually means increased productivity and this implies additional available hours each day, probably the most valuable resource a scientist can have. But this raises the question: how to actually find out the balance between wronfi and wraygo in your daily scientific ‘diet’? In fact, it essentially does work like trying to lose weight to a certain degree. One can try different strategies and methods but how your body responds is actually quite difficult to determine a-priori. You may be more productive for a while writing things immediately but then struggle to find your focus. This is the well-known jo-jo effect, initially you lose weight just to gain it back again after a few weeks or months.
One trick that worked reasonably well for me is to try to check out those diets, which work out over longer periods of time. For example, when I do have an idea that I would like to write down – e.g. like this blog – then I don’t really feel like waiting. It somehow dilutes my thought process. Unfortunately, this wraygo-type approach has the drawback that I need a suitable computing environment to type everything, otherwise one has to type up hand-written notes, convert things between different formats, or – even worse – keep different versions on different computers (laptop, home, work, etc). However, there is a practical solution by just using a folder in a cloud storage data service. In this folder you can save all your scientific wraygo projects and always be able to add or modify them at minimal opportunity cost. For actual mathematical theorem-proving and doing simulations, I prefer a fixed environment in my office with a computing setup and a corner for doing all the pen-and-paper calculations. This naturally adds a wronfi element to my work. Although this is just one possible ‘diet’, I believe that using a certain scheme that works towards your strengths and tendencies can drastically improve productivity. Of course, there are many other scientists, who have thought about this process. For example, I remember that Terence Tao wrote once on his excellent blog, that he likes to ‘batch-job’ low-brain administrative tasks in one period so he has other periods free for demanding calculations and writing papers. I found that suggestion logical and tried it. Somehow it did not work out directly for me. Initially I had no clue why. The main point was that I had put the administrative batch-job parts on an empty afternoon, did them all in a row, it was horribly boring and I was unmotivated afterwards. The solution was to put them onto a day, where there are already a lot of meetings, i.e., schedule them into the gaps of the day. This has worked out a lot better for me. Maybe for someone else it would work better to schedule them for an evening session, or directly before breakfast. The motivating point I am trying to make is that you should go and try to find out for yourself, it really pays off. Once the correct wronfi-wraygo balance has been determined, I would almost guarantee anyone to be more productive and happy in the paper-writing process.