14

I'm currently writing my MSc dissertation for an MSc in Artificial Intelligence. My university allocates a 3-month period for the whole project.

At this point in the project, I realised my reasoning behind the model is somewhat flawed, but that is not a valid reason to request an extension. I am stuck in a loop of writing and re-writing my observations, but I can't seem to make any progress as I keep re-writing the same portions of the dissertation.

How should I approach writing the dissertation, given my reasoning was somewhat flawed from the start and I have no way of changing/re-implementing the model?

EDIT: I have spoken to my advisor, but as he is one of the two people marking my work, he can't really give me guidance on the topic outside of "explain your reasoning". I realised the issue quite recently and he said he would be unavailable until a couple of days before the deadline, so I can't currently contact him regarding my issues.

3
  • 6
    What led you to conclude that the original model was flawed? The chain of logic leading to that conclusion could be a fruitful discussion in your thesis. If it's because your analysis turned up inconsistencies or errors or failed validation tests, then your work has value in closing down non-productive paths to future researchers. Always remember that negative results are still results.
    – pjs
    Commented Aug 19, 2023 at 19:19
  • 1
    Related: academia.stackexchange.com/questions/30995/… Commented Aug 20, 2023 at 1:59
  • 1
    "my advisor [...] can't really give me guidance" This is a problem in the system wherever you're studying. The "advisor" should "advice", which is synonymous to "give guidance". If your advisor is not teaching you how to do research and how to write it down, they're not your advisor. Commented Aug 21, 2023 at 16:48

2 Answers 2

31

Assuming that you don't have the time anymore to change and correct things, you could explain the model and your reasoning that led to it, flawed or not, and then you do a Section "Critical Discussion" where you state arguments against it, and how you think it should be changed, and then you declare this as "future work".

Note that generally formal models are never perfect and can always be criticised, and one can find surprisingly bad flaws even in some publications. Of course some issues are more serious than others, but still, in my view, if an MSc student realises at some point that their model approach is flawed and the thesis shows that they understand this now and what the problem is, even without redoing things I'd think that's a good thing. Obviously it may depend a bit on how bad the problem is and whether I'd have expected the student to realise this earlier. But then as a supervisor I could see it as my own job to warn you in time about this, and if I didn't, I may have missed this myself, and so I can't blame you that much (even though I may not have taken that much time to think about it).

I can't know about the MSc thesis marking culture in your place, but as a supervisor I know that the time for doing an MSc project is very limited. It may well happen that a student goes for some time in a wrong direction, and time may not be enough to fix it even if eventually understood. So expectations are limited, and for me as a thesis marker it wouldn't be a big problem. (Of course it may depend a bit on what your ambitions are as it may cost you a few percent, but I wouldn't expect it to break anyone's neck.)

It's the scientific process that we try to solve a problem, go wrong, and then understand things better by learning from our mistakes.

2
  • So, as I've just found the issue, would it be sensible to describe the (flawed) reasoning in the methodologies, then describe the experiments I've performed (as a result of that flawed reasoning) in the results section, followed by the discussion about the flaws? I'm worried that the experiments will seem like me just throwing things at the wall and seeing what sticks, given the inherent flaw in the reasoning and the fact that I was testing various scenarios to see how the model performed in those scenarios. Commented Aug 19, 2023 at 11:38
  • 2
    @AlessandroN. Ultimately you have to make the call yourself as what works "best" may depend on the specifics. My idea was to start off as normal, i.e., as if it were not flawed (although you can mention in the Introduction that there will be important criticism later), and then criticise it later. But how well that works in your exact situation and what your alternatives are I can't tell. Commented Aug 19, 2023 at 11:51
2

This is an MSc memoir, a compulsory step to your diploma (at least here in France, and generally in Europe).

Nobody cares about the MSc memoir and it is unlikely someone will read it if it is not really interesting.

Yours, according to you is not interesting because the idea is flawed. Your best approach is to say so and explain why it is flawed. And be done.

Your supervisor thinks the same so do not worry about writing a masterpiece.

There are thousands of MSc memoirs that have zero useful content, and they are still on the right side of the face, compared to those where there are lies added. Do not lie or hide anything - this is something that is not only immoral (YMMV) but also potentially a risk for you.

8
  • 4
    Has the OP stated this somewhere? If it’s the thesis project of a UK taught masters (3 month project for a 12 month masters is typical), then it could be a third of the total credit, and might even determine the final classification by itself.
    – origimbo
    Commented Aug 21, 2023 at 2:57
  • 4
    I feel strongly that you shouldn't be making that type of assumption, and you definitely shouldn't be trivialising something that OP has clearly put substantial effort into. However I agree with your "say so and explain why it is flawed": an exploration of why the incorrect assumptions were made and what the consequences were could lead to a very interesting thesis which would be of lasting use to others in the field. Commented Aug 21, 2023 at 5:15
  • 1
    @MarkMorganLloyd: this is not a trivialization but a constatation. I have read countless MSc theses from four universities (in France, Germany, Poland and Scotland) and in general they are worthless. This is not a critique of the authors but rather of the system - the students do what they have to do in order to get their diploma and we pretend that we care about how they write a memoir.
    – WoJ
    Commented Aug 21, 2023 at 8:13
  • 1
    @MarkMorganLloyd Arguably, if the student's thesis advisor can't give them advice on the project because they'll be grading it, the institution isn't demonstrating much effort/expectation towards students producing influential work.
    – Bryan Krause
    Commented Aug 21, 2023 at 13:01
  • 2
    @AlessandroN. That's certainly better, but communicating results is a huge part of the academic research process. I think the point WoJ is making in this post is that you need only to write something suitable to your examiners, and it's probably more important that you understand and communicate flaws than it is that you produce work that is broadly useful (meaning useful besides your own learning and assessing your learning). If your goal was to produce work that is broadly useful, your advisor would be helping you write it up in a way that is broadly useful.
    – Bryan Krause
    Commented Aug 21, 2023 at 21:58

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .