How to finish a thesis where none of the experiments worked out?

Question

I am writing my Master Thesis in Computer Science and am quite desperate. I have been experimenting with a new set of mathematical tools my advisor has done research on.

In the last three month, the experiments did not work well and did not propose the superior results he proposed previously. Even though I use it in a different setting, I expected it to work at least reasonably well.

That it doesn't work should not be a problem to write my thesis. I only need to find the right explanations for it. However, I have not the slightest idea. These tools are a niche topic even in mathematics and I am a computer scientist.

Apart from this, I am afraid only doing an error analysis won't be sufficient for a master thesis. Since these experiments were actually the premise for further modifications and ideas, I am afraid those has become also useless.

Answer 1

The general rule is that research is about knowledge not confirming "expected" results. Knowing that something is false or doesn't work is knowledge and it can be very valuable.

Your professor made an hypothesis and you seem to have shown that it is false. That is worth knowing. It doesn't invalidate future work, but requires that you redirect it.

At a deeper level, knowing why something doesn't work is even more valuable. That is possible in fields like math and CS, but less so in things like humanities research where "proof" is elusive.

But an analysis of the why might be within reach and worth a writeup.