CODE IN TOKYO

Why Practice Alone Isn't Enough for Engineers

For programmers and engineers, hands-on practice is fundamental.

You build things, break them, run them in production, hit real failures. That’s where the actual engineering muscle gets built. No argument there.

But separately from that, I’ve come to think the kind of studying that has nothing to do with your current job — textbook learning, certification prep, things like that — matters more than people give it credit for.

The reason is simple:

Practice alone narrows you.

When you do the same site, same stack, same architecture, same ops every day, you slowly turn into a “this-company-only engineer” without noticing. The Japanese term for it is takotsubo-ka - live in octopus pot — falling into a niche pot.

It’s especially dangerous if you’ve been at one place a long time. You’re writing code every day, so it feels like you’re growing. But in many cases, you’re just doing the same thing faster.

The Counter-Argument, and Why I Disagree

I get the usual pushback:

And honestly, none of those are entirely wrong. Plenty of people hold AWS cert and still can’t architect anything sensible.

But that doesn’t mean “you don’t need to know.”

What matters is having knowledge of work you’re not currently doing — kept somewhere in your head, ready to retrieve when needed.

The Specialist Who Couldn’t Move

I once knew an engineer who was extremely strong in Windows C++ desktop development. Win32 API, GUI internals, deep platform knowledge — all of it.

But outside that domain, basically zero interest.

AWS? “I’ve heard of it.” Docker? “Not really sure what that is.” Web? “Different world.”

If you’re at the very top of the industry in a narrow niche, that’s actually fine — the work comes to you regardless. But this person was deeply skilled in C++, just not at the level of being known industry-wide as a specialist.

And he was on an external contract, not a permanent role.

So I quietly wondered:

“What happens when this contract ends?”

He didn’t seem worried — he had family wealth backing him up — but most people don’t have that safety net.

AI Makes Breadth More Valuable, Not Less

In the AI era, this matters even more, because the rate of change is brutal.

You can now develop on a cloud you’ve never touched, with AI’s help, to a usable degree. Which means “do you know the basic concepts?” has become surprisingly load-bearing.

What’s ECS? What’s a load balancer? What’s authentication?

Even rough, fuzzy knowledge of these things changes how precisely you can instruct an AI — by a lot.

So even certification study is valuable. Not because the cert itself matters, but because it forces you to look at,

“the world outside the one you currently work in.”

If you only do practice, your field of view narrows surprisingly fast.