Lessons from writing my dissertation

2 April 2026

I have just wrapped up my third year dissertation, a harrowing experience that deeply affirmed my desire to pursue a PhD. In an effort to milk some value out of the numerous mistakes I made along the way, I have collated below what I wish I knew before I started.

1. Start writing within the first month of your project. Everyone says this, few do it. But I really really hope that I’ll pull it off next time. It is deeply unpleasant to have to write up the entire thing in one fell swoop.
2. It is incredibly important to maintain a lab notebook. Track every single change you make, record what worked and what didn’t work. Perfect logs will make it much easier to restore state, and more importantly, in the (highly likely) circumstance that you fail at (1), logs make it much easier to write up materials/methods and results in one go.
3. For any kind of data analysis, APPEND ONLY. Never delete or overwrite, no matter how useless a particular approach seems to be. Use a nested tree-style structure with Jupyter or Marimo notebooks, and for each change, duplicate the notebook entire. Name each notebook like 1.1.1, 1.1.2 and what exactly the notebook does, e.g. 1.3.2 segmenting at 0.7.ipynb. Then reference this indexing scheme in your lab notebook. Again, this is insurance if you fail at consistently doing (2), as good semantic hierarchy is a form of logs in itself.
4. As a general principle, do research as if everything you do should be interpretable to a third party. It is difficult to consistently act to the benefit of your future self. Instead, cultivate delusions of grandeur about your repository; imagine it as a sacred and hallowed folder that will form the basis of several neo-religions in a thousand years, and any ambiguity or lapse in documentation will trigger world-historical interpretive schisms portending intergalactic nuclear armageddon.

I am acutely aware that much of this may be common sense, but experience and suffering tends to add new texture and profundity to even the most clichéd wisdom. Besides, it is common sense that I lacked, so I hope it is of some use to fellow green-eared undergraduates.

The dissertation itself is of little absolute value. But I can say with no exaggeration that the opportunity to work with a fantastic supervisor and lab, learn some difficult lessons, and have a taste of scientific exploration and autonomy were transformative to my self-regard and career aspirations.