This whole "read code" thing ("write code that's easy to read", "learn to read code") is imprecise at best. Whats important is to learn code. You want to learn to write code that's easy to learn (not easy to read). And you want to learn to learn code.

If you write code that's easy to learn, and if you learn to learn code, you'll simply remember more code. The more code you know, the more productive you gonna be. That's because at one end of the spectrum, you know nothing about the code and you simply cant do nor solve anything. And at the other end of the spectrum, you know all the code, in which case you gonna be able to do whatever you want by definition: analyse it, expand it, rewrite it, refactor it, talk about it, fix it, etc.

This also explains why you want to write code with fewer lines of code (or fewer AST nodes): the more lines (or AST nodes), the more you have to learn. The more you have to learn, the less you know, the less you know, the closer you are from the bad end of the spectrum.