This is a selfish, personal therapy session article, where I’ll be talking about my response to the rise of LLMs.¹

I’ll start off with a fact: my main hobby is working on personal programming projects.

I’d bet I’ve spent the majority of the evenings of my teenage and adult life tinkering away with something-or-other in a code editor. High school was the peak of excitement for me. Every new technology, every honed skill, filled a young Aaron with an incredible number of new ideas. Each was achievable with just a little bit more learning.

I learned PHP and built a product review website. I learned Visual Basic .NET and made a bespoke text editor with text-expansion macros. I learned Python and used Excel libraries to create a spreadsheet automation tool. I learned Java and wrote a clean web browser (read: WebView wrapper) for Android. And I made 10,000 more things that are gathering dust on the hard drives of old laptops, forgotten to time.

Even as life got busier, I kept this up through university, and through my career so far. Some of my proudest projects have been my most recent, like the Delta Radix calculator and Loconundrum.

---

I work on these projects because I enjoy it. I’m informed this is the typical reason for people to do hobbies.

There are two sources of fun, the amount each matters varying between each programming session.

The first is the process. How will I choose to structure my code? Will I sketch out some interface designs first? Am I going to learn a new programming language or library, or do I get some satisfaction by blasting through the problem with confidence using technology I already know?

The second is the end result. I’d like to be proud of what I’ve made! Maybe I can show it off to friends, family, or fellow engineers, or just silently put in on GitHub.

Not every project has an end result. Sometimes that’s disappointing, but even then, at least there was the process. Maybe it just wasn’t a great idea, and I can refine my thinking for next time. Maybe I structured the code completely wrong and need a refactor, but it’s 11:30pm, so I call it for the night and then never touch the project again.

---

It is October 2026. For the last few years, LLMs have been the hot technology.

Agentic harnesses which give LLMs access to programming tools can just do stuff. They can quickly complete a wide variety of programming tasks, ranging from “fix this bug” to “build an entire application from scratch”.

It’s trivial to use an LLM to build a CRUD web app, a command-line utility, or some library whose README advertises that “it’s not just another simple API wrapper, it’s a genuinely complete bidirectional interface—built from the ground-up. ⚡”

As I write this, a new trend seems to be using LLMs to decompile old video games. Just a year ago, this was a task requiring gruelling determination and a tight-knit community, all working towards a common goal.

---

Being aware of the capabilities of LLMs has effectively poisoned my ability to work on any hobby programming projects. Damaged, self-deprecating thinking now affects every single decision.

For the process, the mere existence of this technology is a constant source of temptation.² If you’re hitting a bug or trying to neatly solve an algorithmic problem, an LLM could solve that problem for you in an instant. Why spend time working through on it? Why get the reward of applying your knowledge to overcome it? Why do the thing you enjoy anymore?

For the end result, an LLM can just… get there. You don’t need the process at all. Fire up an agentic programming tool and just ask it to build whatever you want. “Jarvis, work on this project idea and have fun on my behalf. Make no mistakes.”

Given all this, what’s the use in applying my skills? LLM tools could just achieve the same idea, in a fraction of the time - while skipping any of the feeling of reward.³

Likewise, perseverance feels less rewarding than it used to. No longer have I attained some valuable knowledge. Whatever I’ve just learned could probably be churned out of an LLM in a fraction of the time.

I’ll acknowledge that this attitude is probably unhealthy. I think that’s a reflection on me, more than anything. I’ve always been told you’re not supposed to compare yourself to others - that’s probably especially true for an algorithm. Yet it’s really difficult to shake.

---

As my main hobby, I have ascribed a lot of my personal worth to my ability to program a computer. I could solve my own problems, or build websites for family and friends. There was a positive feedback loop from spending time to enhance this skill.

In today’s world, it’s hard not to see this as significantly less valuable than it used to be.

I don’t think I was fundamentally prepared for this, as a person. If that skill was a significant part of my feeling of worth and identity, then what the fuck does that mean for me?

---

Where do I go from here?

In retrospect, maybe putting all of my eggs in one hobby-basket was a bit dangerous in the long-term. Especially a hobby which is also an increasingly-common career, and thus something that is always trying to be made more efficient.

Aside from that, I need to try some new stuff. There’s a world full of hobbies and things to contribute to out there!

At the same time, I need to find a way to reframe how I approach hobby programming, shut out the noise, and keep building cool stuff by hand.

This is all a more philosophical, personal endeavour than I was ever expecting technology to confront me with. Technology might stop surprising me one day, but not for a while.