Learning a programming language

The best way to learn a new programming langage is to get hand-on, and get in the zone. It's also the most fun.

Posted on

In the paper Teaching Programming Languages by Experimental and Adversarial Thinking, an approach to teaching programming languages is outlined that reflects a more natural approach to learning:

…a programming language is less a purely mathematical object and more like an object found in nature. In addition to any formal interfaces it may present, we should – and do – also view it as a target for experimentation. We subject it to a variety of tests. Most of all, we follow a loose version of the scientific method: we form a hypothesis about the language’s behavior; we construct tests (example programs and their expected outputs) consistent with the hypothesis; if the tests pass, we reinforce the hypothesis, otherwise we find it falsified; we use the falsification to generate a new hypothesis, which results in new tests. When we have exhausted our ability (or energy) to falsify, we accept our hypothesis as a tentative truth about the language, and use it to construct actual programs (whose behavior may – after painful debugging sessions – again falsify our understanding).

Basically, we learn a new programming language best by just jumping in and trying stuff.

This is super validating because it often feels like The Right Way to learn is a task list of the most boring words you have ever heard, including but not limited to:1 data types, variables, expressions, operators, classes, inheritance, structs, interfaces, enums, events, enumeration, standard libs, special language features, concurrency, I/O, networking, serialization, and on and on. I just fell asleep writing that.

The better way, in this case, is also the most fun: copy/paste something that looks interesting to you; change some stuff around; when something doesn’t work out, head over to Stack Overflow, check out the documenation, or ask questions on the language’s public chat; then change some more stuff and do it all again.


To learn a skill efficiently you need to get in the zone with some deep, deliberate “practice.”

The Importance of Deep Work & The 30-Hour Method for Learning a New Skill takes an interesting approach—divide your learning into seven or eight, four hour blocks of learning over the course of a few weeks and dive deep.

I like this informal structure because it gives you some bare-bones guidelines for what you need: time to get into the flow, some consistency to build up some working knowledge, and a timeframe for which you can expect to have a basic understanding of the subject matter.

The article gives a sample of how you might break it down. Something specific to learning a programming language might look something more like this:[^rusty]

  • Session 1: Read some posts about the language, watch some videos, etc.
  • Session 2: Install the language, go through the Getting Started
  • Session 3: Do a small project, something you are interested in
  • Session 4: Start a larger, more advanced project that will stretch you
  • Session 5: Wrap-up the larger project from Session 4
  • Session 6: Pick up a book, skim it, refactor based on your learning
  • Session 7: Implement more advanced functionality, or start a new project

But I think it’s important not to systematize this too much. If it’s not fun, it won’t get done.2


Anecdotally, I have been using a loose version of this to learn Rust, and it has been a breeze. It feels like all the pressure and drudgery of learning a new language has been replaced with fun and excitement, which I think is exactly what one would want.


  1. If you remember the Micro Machines commercials, read this in that voice [return]
  2. I just made that up right now and I think it’s pretty clever3 [return]
  3. But I’m sure it’s already been said by a million moms and teachers around the world [return]