Two years. Two years between the first ideas and the submit of the paper, which has gone on its journey today. Sounds like a long time, but to be honest it is not. To show this I would like to explain in this post a little bit the generalised basic steps of my work towards a paper. I will not take the submitted paper from today as an example, because its creation was quite unusual. Therefore, I will stick with the general approach, which is divided in several phases:
At the beginning there is usually a task, which have to be fulfilled. This might be a project proposal, which have to be executed, or a missing piece of a thesis or report. It is usually relatively broad in its specifications and therefore the main role is the delivering of the motivation for the work. It requires also a lot of reading of papers, whether someone else has not already solved all the problems at hand. Starting on this basis, the fun part begins, by finding a solution, which might help to fulfil the given requirements of the task. This phase is fun in the sense of weird ideas, which are connected to each other, but also frustrating, since some things do simply not work. It might be for example an equation, which seems to be correct, but where you do not find the right steps in the derivation. It is a process, which might need months to finish and it is usually solved by hiking, standing under a shower or in your dreams. Such things might sound funny, but the truth is that good ideas not often come along when you sit behind a desk in a grey office.
At this point you can see that I am usually working on methods and new approaches to existing problems. Apart from the well structured paper work a decisive element within this development is try and error. When the idea works on paper it is time to play. Implementation, applying to different datasets, finding out the strength and weaknesses of the method etc. When the implementation is relatively short, it might be necessary to reprogram the code several times to use it in applications, which have not been on the plan at the beginning. Well planed code helps here a lot, but with every implementation a programer has to make decisions, which might proof at another point as the wrong direction. Highly flexible code also costs time, which might be not handy when there is a deadline on approach. In between reports, theses, conferences have to be written or prepared, which show the ideas to a broader audience for the first time. Additionally, a lot of discussions have to be performed during that time. It is the phase, where solutions for errors or uncertainties have to be found and where each iteration and each new application helps in the end.
After this phase the decision will be made, which application is the best to show the strength and much more important the weaknesses of the method. It will be also decided, which procedure of events is the best to deliver a story for the given task. Reasons behind this are here that a paper in the end is a story. It should have a highlight, it should have an interesting background, and yes, usually every scientists like a happy end and not so much a drama. Nevertheless, this is not decided by the own will, but by the data and application. What ever is tried before, there might be the situation that things simply do not add up. When this is the case a critical story has to be accepted. After having a story in your mind, it leads to a story on paper. Here as well, several iterations might be necessary, since it is possible that even at this stage things do not fit together. No, to be honest, at this stage everything should be working great, but it sometimes don’t. Then you go back, check and recheck everything, look for other datasets or applications and you try to verify that there is no error anywhere. All you do here is to get a manuscript, which gave you a clear conscience that you did everything to ensure that there is no methodological error and no surprises hit you after the publication.
After final discussions with the other authors the final submit take place. Now starts the wait for a reaction from the editor and the reviewers. Changes to the manuscript are usually necessary and now in an ideal case it pays of that you have played around before. Answering the reviews sometimes give you new insights and hopefully still allow you to publish the paper. And when it does you are happy that all the work before has payed of.
All in all, there are a lot of steps involved, and all need their time. Sometimes this all need two years or more. Nevertheless, you undoubtedly have several papers on which you work at the same time. Indeed, having one paper in each phase at a time is the simplest way to work. Frustrations are covered by the fun and creative work alternate with the structured one.
What I have described above is the methodological part of a paper, but there is usually also the real science part. Its not the method which makes a paper, but the story on the scientific question, which was delivered by the task. That is why I called this post a scientific cycle. The philosophy of science, represented by philosophers like Popper and Bacon, had found in the past centuries several ways how scientists should do their work. For my PhD I collected these steps and made a figure from this:
Methods play a huge role in this, since they are the main part of the theories and the analysis. But the final case has to be made on the ground of the specific application. It is a hypothesis, which has to be tested and an experiment, which has to be performed. Sometimes this can be seen directly, sometimes it is hidden in a lot of words. Nevertheless, real research has to be performed this way, since without this approach the results might be attackable.
Good papers are the combination of the right methods applied on the right questions. Both need a lot of work and therefore sometimes a lot of time.