Given what I do for a living, I tend to get pretty much daily practice at the fundamentals of code craft like TDD and refactoring. And, after 27 years of professional programming, I would probably still make time every day or once a week to mindfully practice this stuff.
Why? Because – from experience – I believe that these aren’t skills you learn once and then retain indefinitely. On occasions when I’ve been away from code for an extended period, coming back to it I found myself doing sloppy work. I’d forget to do things. I’d fallen out of certain habits.
And I also don’t believe these are binary skills. There’s unit testing, and then there’s unit testing. Not all unit tests are created equal. Some are easier to understand than others. Some are less coupled to the implementation than others. Some run faster than others. Some cover more examples with less code than others. Some teams say “We do unit testing” and their code is still riddled with bugs. Some say “We do unit testing” and their code hasn’t had a bug reported since it went live.
No matter how good we think we are at writing tests or making design decisions or refactoring our code and so on, there’s almost always room for improvement. And as we gain experience in these skills, improving takes more and more work.
As a guitar player of some 30 years, I know that the hour or two of practice I get maybe twice a week these days barely keeps me at the level of technical ability I reached 20 years ago. Most of the visible progress I made happened in the first 5 years of playing. But there’s no top to that hill. If I stop practicing for a while – which I have for the last year or so for various reasons – the ball rolls back down. It takes some ongoing effort just to keep it where it is.
It feels the same with code craft: so many managers who came from programming backgrounds tell me that they lost their skills after a few years away from the code. Indeed, many put themselves on my courses just to try and keep their hand in. Use it or lose it.
An hour or two of mindful practice keeps me where I am as a programmer. It takes an hour or two more to move the ball uphill a bit. This is why I’m such a strong advocate of 20% time for developers. At the very least, half a day a week needs to be set aside to learn, to practice, and to try new things. Or an hour a day. Or however you want to slice it.
Putting aside no time – and I’ve seen this so many times – doesn’t just lead to dev teams who don’t improve, it leads to dev teams who progressively get worse. The code craft muscles atrophy, habits get lost, teams get sloppy, code gets crappy, and delivery slows down to a crawl.
If you prefer a mechanical analogy, think of development as a steamboat. Your goal is to deliver goods from London to New York, sailing every two weeks back and forth across the Atlantic. How much time do you set aside to maintain the engine? None? That’s just not sustainable. How many trips could you expect to make before the engine failed?
The developers on your team are the delivery engine of your software and systems. Their skills – not just technical – need regular maintenance, or the engine breaks down. Set aside some time and put aside some budget to help keep the team in good delivery shape. Not just the occasional training day once or twice a year; a regular investment in building and maintaining their skills. Once a week, or an hour a day – whatever works in your organisation.
(You may be thinking “Hey, we pay them to work. They can learn in their own time.” And that might explain why your efforts to increase diversity on your teams haven’t worked.)
My closing message here is that it’s not just junior programmers who need to work on the fundamentals. We all do.
Here’s a tip from an old dog: a great way to reinforce knowledge and skill is to teach them. My advice for managers is to combine regular practice with mentoring. Your more experienced developers mentoring new hires will find – as I have – that the mentoring process forces us to really think about not just what we’re doing, but why we’re doing it. And it encourages us to be perhaps that little bit more exacting about how we do do it, because we’re trying to set a good example. It’s us as programmers, but in our Sunday best.