State formally, reason informally

Evan Chen’s Infinitely Large Napkin is my go-to resource when I want to learn about some new area of mathematics (or at least it used to be; I’m increasingly often running into the issue that it doesn’t have a chapter about what I want to learn – at barely over 900 pages it’s just way too short). I recently asked myself what it is about the Napkin that I like so much, and this post is part of my answer.

Note that I’m describing the way I like to learn math, not the objectively best way. If you happen to like learning that way too, maybe this post can give you a more explicit idea of what “that way” is. Otherwise, it will at least give you some insight into the brain of someone who learns differently than you do, and maybe that will help if you want to teach math to other people.

One way of teaching mathematical concepts is the approach typically employed by textbooks: give some definitions, probably a few examples, state a few theorems, prove them. In some cases, you might also get an intuitive explanation of what these definitions are all about or why a theorem is interesting or how the proof roughly works. But often, these explanations are sparse and confined to the beginning of a section, so as not to dilute the mathematical purity of the remaining text.

Another style of explanation is the one that consists almost entirely of hand-waving. Evan Chen describes it as follows:

Someone tells you about the hairy ball theorem in the form “you can’t comb the hair on a spherical cat”, then doesn’t tell you anything about why it should be true, what it means to actually “comb the hair”, or any of the underlying theory, leaving you with just some vague notion in your head.

We could think of these two approaches as opposite ends of a “rigor”-spectrum. Somewhere in the middle, you might have the way that theoretical physics is often taught: few things are formally defined, but concepts are at least made explicit enough that you are able to use them in calculations, and equations are usually derived (though less rigorously than in math).

Where does the Napkin fit in on this axis? Is it more or less rigorous than theoretical physics courses? I think that’s the wrong question to ask because the one-dimensional model of approaches to teaching math is too simplistic. The space of all the ways you could explain mathematical concepts is very large and treating it as one-dimensional isn’t such a great approximation (though I think that the “rigor”-axis might be the best one can do with only one dimension).

So what’s a better model? I want to suggest thinking about mathematical explanations using a two-dimensional approximation. On one axis, we have the rigor used when stating definitions or results, while the other axis is the rigor of the derivations of these results. Textbooks have both formal statements and formal derivations, casual hand-wavy explanations have neither. Physics is somewhere in between, though to me the derivations often seem more rigorous than the statements¹. Finally, the Napkin takes the opposite approach: it states things with essentially the same rigor as textbooks but then places emphasis on very informal derivations or explanations of why these statements are true.

I find that this approach – stating things formally but reasoning about them informally – works extremely well for me when I first learn about a subject. I’m happy to take someone’s word that a statement is true, at least initially and if the statement seems like it should be true. But if I don’t know precisely what the statement is, I find it much harder to get to grips with the subject.

As a caveat, note that I said “when I first learn about a subject”. This approach doesn’t teach how to write formal proofs, and it’s by design not as comprehensive as a textbook. But sometimes an intuitive understanding is enough for my purposes, and even if I know I’ll later want to learn a subject in more depth, I find it helpful to build this intuitive understanding before going through all the details.