Deliberate practice applied to coding offers some unique opportunities. Unlike other skills, like learning to play the cello (to pick one that I have some experience with), you can go very far without a personal mentor. The feedback from the computer is about as objective as it gets. It will let you know exactly how good your code is.
This also helps remove the emotional component – positive and negative – that can sometimes impede progress with an in-person mentor. This doesn’t remove all emotion, however. Just about everyone who’s worked in a professional coding shop has witnessed the rare occurrence of a coder cursing at or even physically attacking their computer because their code isn’t working as expected. Those are surreal moments when an avalanche of cognitive biases and unconscious behavior become visible to all but the coder. That’s a topic for for a different post. Suffice it to say, learning how to control your emotions, channel frustration, and ignite curiosity is part of what distinguishes good coders from great coders.
Which gets met to finding quality feedback. While I’ve made a good living writing mountains of proprietary code for various business and corporations, I earned my coding chops by working on or authoring open source projects. This was the best source I found for getting feedback on my code. It also taught me another important lesson: Do not attach your identity to the code you write. Like any noob, I had a lot of pride in my early code that was pretty much untested outside my little work environment. In the open source world, the feedback was often swift and harsh. Or, at least is was when my identity was attached to it. Learning to separate work product from identity revealed just how much emotional spin I was putting on what was in hindsight reasonable feedback. I have concerns that the current climate in the coding world is opting for soft feedback and good feelings over legitimate and reasonable feedback. This, too, is for another post.
It’s worth giving some thought about the the pros and cons of working with an actual person for mentorship. Along with good instruction, a single mentor will pass along their own limitations and biases. Not necessarily a bad thing, just something to be aware of. So multiple mentors are better than just one, which starts to move down the path of actively participating in open source projects. By “actively” I mean not just contributing code, but studying the code (and it’s history) of existing successful projects. There are usually many ways to solve a problem with software. Work to understand why one approach is better than another. Insights like this are best gained, in my experience, by studying good code.
Somewhat related, if you are working from a book or a training program, actually type in the examples – character by character. Don’t cheat yourself by copy-pasting code examples. This is the muscle memory component to coding that you will find when learning other more physical skills (like playing the cello.) If you really want to experience the gnarly edge, ditch the IDE and code with at text editor. I still do all my coding in vim and this keyboard.
Another approach to deliberate practice is the idea of coding “katas.” This never clicked with me. I attribute this to having studied martial arts for 25 years, most of that time at the black belt level. Mapping the human psycho/physical world and the purpose of katas in the dojo to the machine world is too much of a mis-match. Much is lost in the translation, in my experience. The katas in the dojo, regardless the art form, translated easily to other styles and practices. The coding “katas” are more tightly coupled to the coding language in which they are written. In my view, it’s yet another example of swiping a cool sounding word and concept and force-fitting it to another domain. A software version of cargo cults – expecting form to create function. “Black belt” or “Ninja” coder are other force-fits. Yet again, something for another post.
But those are my limitations. Your experience will no doubt be different. As learning exercises and proficiency tracks, many of the coding “katas” look to be very good.
(For related thoughts on how building your own tools can deepen your understanding of a skill, see “Tools for Practice.” The examples in the article combine software development and cello practice.)