I used to have a mental block against casually reading things that weren’t designed to be casually read, like old papers, textbooks, documentation, etc. If I was going to read for enjoyment, I’d read blogs or popular science/fiction books. Not sure why.

Over time, I’ve met people who read things I’d never think to read. I’ve gotten a lot of value from copying them, so I’m sharing my thoughts here.

Contents: Documentation | Primary Sources | Books and Textbooks

Documentation

For a programming assignment using a new language (erlang, I think), my original plan was to look up syntax and concepts as needed while I coded. Then, a friend told me he was reading the first few sections of the manual before writing any code. I’d have never thought of that; I’d only ever read a manual when I needed a specific piece of information.

I’ve started to do this with some of the internals of programs I use, and it’s really interesting; you learn not just how things work under the hood, but why they were designed that way, and the neat tricks that were used to make them work. Every time I open the bash manual, I learn five super useful features, four of which I promptly forget to use. Another great reference is The Architecture of Open Source Applications, which isn’t really documentation, but shows how different open source projects are designed. Reading these kinds of documents changes how I think about programming for the better.

Primary Sources

Another underrated source of interesting information is primary sources. Like, the original source. Wikipedia and blog posts are great for quickly learning something in a digestible way, but they don’t capture the context of the time. It’s one thing to look up the theory of evolution, but it’s another to read Darwin’s original work:

Although much remains obscure, and will long remain obscure, I can entertain no doubt, after the most deliberate study and dispassionate judgment of which I am capable, that the view which most naturalists entertain, and which I formerly entertained–namely, that each species has been independently created–is erroneous. I am fully convinced that species are not immutable; but that those belonging to what are called the same genera are lineal descendants of some other and generally extinct species, in the same manner as the acknowledged varieties of any one species are the descendants of that species.

This quote is so powerful to me; it’s like watching a pivotal moment in history from Darwin’s perspective. It’s also pretty fun to read the original perceptron paper from Rosenblatt:

In the meantime, the theory reported here clearly demonstrates the feasibility and fruitfulness of a quantitative statistical approach to the organization of cognitive systems. By the study of systems such as the perceptron, it is hoped that those fundamental laws of organization which are common to all information handling systems, machines and men included, may eventually be understood.

It’s crazy to imagine that someone had to prove statistical approaches to AI were feasible; we take it for granted now.

Books and Textbooks

Textbooks are fun, too. As my family can attest, I hated biology in school, but Karpathy inspired me to give it a second chance when he randomly started reading biology textbooks in his free time. Turns out I don’t hate biology either–I just hate people telling me what to do, as you can probably tell from this website.

You don’t need to read textbooks in order. You can just pick a random chapter and read it. There might be a chapter that builds on a previous chapter, but if you don’t understand it, you can just read its dependency–chances are it only requires reading a section or two, not every preceeding page. The figures are also really helpful for getting the gist of ideas quickly.

After years of reading books front-to-back until I abandoned or finished them, jumping around a book feels so freeing. Especially fiction–sure, you don’t get the full story, but you get the vibe, and can usually infer missing details.