So I’ve been saying for a while that reading academic papers is a skill. However, what I didn’t realize is that having a pretty good idea of how to read a paper from one discipline doesn’t quite map to knowing how to approach a paper from another field. Academic papers are all confusing nightmares, but papers from different disciplines are nightmarish in different ways, which means that while some parts of the basic reading approach can stay the same, others will have to be discarded or relearned.

I’ve been thinking a lot about the value of writing recently, especially in regard to how it’s been influenced by the advent of generative AI. In the past few months, I’ve had a lot of conversations in which people claimed that generative AI is just as good at writing as humans now, or better, in many cases. While I agree that AI generation tools are, at this point, better at the mechanics of writing than the average person, I have always found their outputs to be shallow and devoid of interesting surprises.

Last summer, I was tasked with writing a report about the research I completed and submitting it to my supervisor by the end of the summer. I, of course, had never done this kind of writing before, so I learned several lessons. The hard way. The learning process was super painful, but it did come in handy when I had to write two other research reports in my classes last fall. Here are a few of the lessons I learned.

Over the last month and a half or so, I’ve come to the conclusion that I actually don’t know how to read, which is definitely a jarring realization to be having after over 18 years of formal education.

Okay, maybe I’m being a little bit cheeky here. I am not literally claiming to be illiterate or even functionally illiterate, and it would be stupid of me to do so, since clearly I am writing this blog post and have written many other blog posts where I reviewed books. However, one of the things my high school education didn’t prepare me for, and that my four years of training and education in engineering and computer science have completely failed to teach me, is how to both get through and learn from – “learn from” is the key term here – a large volume of readings on a weekly basis.

A while ago, someone in a Discord server I’m in asked how much of the math behind a system you need to know to implement it. I thought it was an interesting question, and I felt qualified to answer it, so I ended up writing quite a lengthy response. It just occurred to me that it might also be useful to other people, so I thought I would clean it up a little bit and archive it here.