You're Building the Wrong Stuff...

Timestamps

  • - Intro
  • - How AI affects software development
  • - How I got started and learned new skills
  • - How exploring deepens your skill level
  • - Develop at a depth you are uncomfortable with
  • - How do you know what to build?

I usually tell newer developers to build software that solves a real world problem, something you could get users for or eventually make money from, instead of another calculator or to-do app that's been done a thousand times. I still think that's good advice. But there's also a time to build something nobody asked for, because a lot of the biggest jumps in my skills started with software I made for myself with no specific goal in mind.

Both are valuable, this is just about the second one.

AI and learning at work

This isn't an AI video, but it's worth mentioning because it affects a lot of developers. AI doesn't just let you build the same features faster, the expectation becomes building 10 times as many features. When that's the requirement you don't get the time to actually understand each feature, and the quality of the software goes down.

You can push back and say let's keep the feature set clean and understand how everything works, but a lot of upper management is going to push features out the door regardless. So many developers feel like they're not learning at work like they used to, and that learning has to happen on their own time now.

My suggestion is something you can do in as little as an hour a day: work on something that has no outside expectations or requirements.

Stuff I built for myself

I'm super unorganized, so most of this code is nowhere to be found (it wasn't until a few years in to my career that I started saving most of my work on GitHub). You'll kinda just have to trust me.

  • First HTML and CSS page in high school. Before that I tried C++ with a giant book and gave up. A "for dummies" style HTML and CSS book got me a page that was atrocious, but it got me hooked on building software, which was a bigger win than finishing the C++ book would have been.
  • A Minecraft server on my own computer, with port forwarding and the networking set up so my friends could join. No paid hosting. That got me to pick up a book on web servers and web development, and it was probably my big jump from just playing games to building stuff other people could use.
  • Front end talking to a back end. Getting backend data to show up on a web page for the first time was a revelation, up to that point I was hard coding everything in HTML and CSS.
  • PTZ camera controls at my first job. Pan tilt zoom cameras. Nobody assigned this, I was just dorking around in the camera's API docs and built a page with arrow buttons that moved the camera from my own code, no vendor or third party security software. A coworker saw it and said we could actually use it in some of our custom software.
  • A cryptocurrency explorer to see the transactions and blocks for a blockchain I was using. It got a good amount of users and was my first real end to end app: front end, back end, deploying, validating data, keeping it secure.
  • Currycoin, my own cryptocurrency and blockchain from scratch in Rust. I wanted to learn Rust and understand the protocol itself, not just call some blockchain's API. It never got to the point of being fully usable, but I learned more in that time than months of messing around with crypto.

None of these came from a ticket, and every one taught me more than the projects where I was assigned tasks. Other than the explorer and the Minecraft server, I didn't have clear outcomes in mind, I just started building and saw where it went.

💡 Tip[!tip]
If you want to see what "understanding the protocol" means for something like a blockchain, the original Bitcoin whitepaper is short and readable, and The Rust Programming Language book is free online if Rust is your thing.

Why exploring teaches you more

At work you create a feature, ship it, and move on. You don't get to spend as long as you want learning it at the detail you want. When I was building the crypto stuff I spent a day or two just researching different approaches and structures so I did things right. When you have to give a daily report on your progress you can't really do that, so you pick something and go with it.

The self-initiated stuff is also way more efficient learning because you're actually motivated. I was up late working on Currycoin because I was excited about it, not avoiding it.

I don't think it has to be one or the other though. If something you built just to explore starts turning in to a real thing, figure out how to go deeper and make it software other people can use. Going from software that works for one person to software that works for many people is a skill of its own:

  • will other users break things?
  • will someone do something malicious?
  • do you have the security set up that you need?

The list goes on. When it's just you, you're pretty confident you'll use it the way you expect.

Go deeper than is comfortable

So beyond just exploring, I'd encourage developing at a depth you're uncomfortable with. Pick a project and keep going deeper, instead of building a shallow version of something interesting. With the blockchain I wanted to understand wallets and transferring tokens from one key to another, the complete end to end, not some cute little thing nobody would care about.

For portfolio projects, don't try to get as many on there as possible. Get a couple that are really good and go a little deeper than most people would. That's what shows you know how to build production quality software.

This doesn't mean build big. A whole blockchain and coin is ridiculously over the top. It could be:

  • a small but really good plugin for some software
  • a mod for a video game
  • a website with one tool that's actually valuable

Build it high quality, and have some pride in it instead of rushing it on to your portfolio.

For me right now it's one piece of software I've been working on for months. I'm at the point where I want to make really good software and not a bunch of random stuff, so a lot of my focus is on product design, which features it should have and which ones shouldn't, and even the marketing, sales and branding around it.

What should you build?

Two suggestions:

  1. Something you're interested in. Pokemon cards, Lego sets, piano, whatever.
  2. A pain you have, or an issue in some software you hate. Say you can't keep track of your cooking ingredients and you don't like how the other apps do it. Build your own just for yourself, and once it's good enough you can look at sharing it.

If you have more ideas than coding experience, you should probably spend more time on the fundamentals first. But if you have the coding ability and just don't know what to do with it because nobody is telling you what to do, go through the exercise of building something unprompted. Piecing features together in to software, not just writing the code, is a skill a lot of developers don't have.

If you're working on anything interesting, let me know in the comments on the video. Also let me know if you like this talking style of video or if you'd rather see more tutorial stuff.

Powered by Course Catalyst