Ask HN: How to teach myself not to overcomplicate? | Hacker News Reader

Ask HN: How to teach myself not to overcomplicate?

4 pointseuos11 comments

All my personal projects seem to fail because I jump off the deep end.

E.g. I want to implement a simple image hosting to learn GCP and have some fun with React/Tailwind (I had AWS experience). Instead of building the hosting itself I start with user authentication, reading about Passport.js, JWT, specs, etc. I spend a lot of time on this, until I get bored and set the project aside...

I have 20+ years of experience and I worked on stuff like Node.js and Chromium - so I know a lot of "what can go wrong" in a cloud application - security, scalability. How can I learn to ignore those concerns when I toy with code for fun?

View on HN

Start by determining what absolutely needs to be included in a minimum viable product, and prioritize ruthlessly. Amazon does this using "press-release-driven development."

Create a simple proof-of-concept prototype, maybe of just one small tiny aspect or feature of the system and then continuously iterate. Don't try to boil the ocean all at once. The positive feedback from having something that works, and then being able to try out changes and improve it is much more pleasant than the continuous frustration of having something large and complex that doesn't work which you keep needing to fix bugs in.

In your case, I'd start by building a "Hello World!"-style app using some simple framework like Python Flask (https://flask.palletsprojects.com/en/2.2.x/quickstart/) or the node server of your choice. If it's longer than 50 lines, then you're doing it wrong. No CSS, no client-side JavaScript, just a plain HTML form that lets you POST an image from a browser to the server filesystem. Then once that works, start adding more functionality: add another endpoint where you can see the POST-ed image, once you can see the images, give them content-addressable hashes, post the images to some cloud service, etc. But first get something working locally that you can rapidly iterate on. Good luck!

Well all complex systems start from simple systems. A complex system based on another complex system becomes chaotic, quickly. Consider this when building anything computer-related. Start with first principles and remember the K.I.S.S principle. Seems you're hungup on your environment. I like to have a base Debian install and only add things when needed, instead of architecting 'teh one true environment for every use-case'.

“Remember this adage developed by the U.S. Navy in 1960, KISS: Keep It Simple Stupid. Avoid unnecessary complexity.”

From:

KISS For Health, Wealth, and Happiness

Cats, Coca-Cola, And Documentary Movies That Show “Simple” Is A Force Of Good

https://moviewise.substack.com/p/kiss-for-health-wealth-and-...

You're doing this because you're working on toy projects with no real stakes. You overcomplicate because you can afford to.

Establish a forcing function in the form of users. Onboard them and include a way for them to report bugs (provide them with an issue template) and also maybe add a Slack workspace for the project. This will do two things:

- Bring issues you were not aware of to your attention.

- Prioritize your backlog with frequency/severity: by the time the fifth user complains about something you were aware of (frequency), you'll drop whatever non-essential thing distracting you and get to fixing that issue. By the time something really bad happens once (severity), you get to work on it as soon as possible.

I can't get to the stage where I have users

Could you elaborate, please?

I don't reach a stage where something is running. E.g. I have half-broken login page (with backend services) instead of actual image hosting.

Precisely. Get users to your product and you'll reach the stage of finishing and getting working things out of the door.

Try to set your self some limits. Run your projects as real projects: with defined scope and limited resources (time).

Oh, my resources are limited. Full-time job, 2 kids and such... That's what makes progress slower and that overhead so painful.

Can you pretend that your project is to build MVP that just needs to work? And consider security features/fixes to be something for next release?

Reply on news.ycombinator.com