I have the chance every year to supervise MSc theses on topics related to my research. While discussing with my colleagues, we realised there was a pattern in what the students did not get right during their thesis. Thanks to many iterations and feedback from these colleagues (I list them at the bottom), we have come up with the following 9 laws:
-
Meet meet meet: schedule meetings well in advance with your supervisor. A master thesis trains you, among other things, to manage your own project as you would do later in your job. So set an agenda, suggest several time slots and as soon as possible prepare presentations so you can also have feedback on this important aspect. And always take notes during the meetings; never assume you will remember it (because you don’t).
-
Keep it your own: as a rough approximation, the content should be composed of 25 to 35% previous work, 65 to 75% your own contribution, i.e. presentation and in-depth analysis of the results
-
Keep it short and to the point: organise your text so the reader get the main message quickly. Don’t hesitate to highlight the main results in lists or tables, so the reader can easily have a first overview. Don’t present background information that you are not using in your document, e.g. state-of-the-art information that are only remotely connected to the results you are showing afterward.
-
Keep it referenced: show that you have read the scientific literature and cite meaningful journal articles to support or illustrate your points. If you present information, it should either be the result of what you have elaborated before, what you have shown in a figure/table or what comes from a references. If it is merely your opinion, this weakens your text.
-
Keep it tidy: make nice and consistent graphs focused on messages, check spelling, only show significant digits, use SI units. Check that units on the left and right side of the equal sign are the same. Define every symbol or acronyms the first time you introduce them, and strongly limit the number of acronyms in your text as it makes it unreadable.
-
Keep it attractive and consistent: if the reader first skim through it, use nice illustration with message oriented captions (as opposed to « schematic of my system ») as an opportunity to get your message across. If blue is 1 and red is 2 on Figure x, then don’t change this in Figure x+1.
-
Keep it ecological: ask the reader, well in advance of your defence, if he/she wishes to have a printed version or a PDF. In case of a printed version, always double-sided.
-
Keep it shareable and organised: use the summary as a press release, setup a shared repository for your data, include uncertainties, organise your documents and code so that someone else can start from it (and send the bundle to your advisor).
-
Start early and fail quickly: have a head start and try to provide your supervisor with results and written documents as quickly as possible. It will force you to achieve something, perhaps failing in delivering what your supervisor expect but at least learning a huge amount in the process. Depending on your field, the first document might take more or less time, to be discussed with your supervisor. As a rule of thumb try to finish the first semester with a solid first iteration, one you would be able to do a presentation of.
Here is a list of contributors sometimes with just a few words in a discussion and not ranked by importance:
Julien Blondeau, Tim De Troyer, Hervé Jeanmart, Mark Runacres, Svend Bram, Kevin Van Geem.