Finishing 2025

Another year comes to an end. I again failed to write too many articles on this blog, but I am not too angry about that. It was a busy year, and one that did not really help with feeling burned out at all. I want to reflect on that ... or just mumble into the room and pretend that anyone cares.

At work, this year was mostly overshadowed by a long migration project. Moving a customer from some old software to our current product can feel very good, but the circumstances here did not allow for that. Why is that?

If you know something about burn-outs, you might know that one big factor is the feeling that whatever you do is meaningless or damned to fail either way, which makes it very exhausting to actually care. With time, you kind of forget how to care about your work (and more) anymore. With this migration project, this was very much the case.

The crux is this: we want to migrate the customer, but the customer is not actually interested in this. The customer is okay with it, as long as the behaviour of the software does not change. The previous behaviour however is neither documented nor does anyone really know it. Even the customer could not tell us what we need to do, but they absolutely know if anything changes.

Well, at least we have the source code for the old software ... which is half a century old, has global state everywhere, lots and lots of files and no good software architecture or any modern nonsense like that.

So, we did a migration project with no clear requirements, but clear deadlines. And - surprise! - we did not meet the deadline, nor the extended deadline, nor the next one. That is because we discovered new requirements that took much longer than expected.

Another exhausting factor in my life this year was the ever-more-omnipresence of large language models everywhere. Reading news platforms like reddit or hacker news just is not fun anymore if LLMs are almost the only topic left. And if you find a cool programming project on these pages, more often than not you will notice that it was done using LLMs.

I am absolutely enjoying the positive influence that LLMs can have on our lives. I like to use it for my work too. But I do not generate code with it that I intend to actually use. I cannot stand that so much people differ in this. LLMs cannot reason, which is very important during programming. Sure, they have become better at simulating reasoning for some cases, but then they still fail catastrophically the moment after, and gaslight you totally.

I just find that I cannot trust the internet anymore. Nothing is real anymore and if I find a cool article, it might be based on lies and hallucinations. Or maybe it used another article as a source, which again is based on lies and hallucinations. This feels like "The Machine Stops" from E. M. Forster. In my mind, every developer should read this short story once. The aspect I am referring to is that in the story, the "intellectual" people of the world dislike original thought and prefer to learn about anything through as many other people as possible. That way, they only ingest knowledge that was filtered sufficiently.

Whenever I find the energy to write more on this blog, I want to concentrate again on original content based on my own experiments. I believe that this is the content the internet needs (even if it does not deserve it).

My promise to you is: Nothing in this blog is knowingly taken from LLMs, except if I make it obvious (like that ChatGPT Fizz-Buzz article of mine).

I learned two new languages this year: Toki Pona and Esperanto. Well, I am still learning the latter one, but Toki Pona was simple to learn, if you put your mind to it.

Toki Pona is a constructed language created by Sonja Lang. It uses around 140 words, but is still able to express most concepts (with some vagueness, of course). I started learning it in April and feel that I could read and write it proficiently by June, but I also was kind of obsessive here.

While learning Toki Pona, I changed my mind on Esperanto. I was sneering at the people who learned it, since I felt that it has no uses. Yes, it might have been a great thing as an international auxiliary language, but it seems to have failed. (Fun aside: I somewhat feel that Raku is the Esperanto of the programming world. It is such a perfect language, except for reality ...)

This feeling changed to curiosity, and as I dived into some of the very well-made free online courses on Esperanto, I noticed that I really enjoy the language. I can read most texts if I have a dictionary at hand, and I only really learned it for another two months until I had no energy for that left in August.

I feel that learning new languages and correspondingly learning to think in different languages gives me new perspectives and stimulates my intellectual curiosity, which again has a positive effect on my other projects. I guess this just goes with my recurring time of finding your own joy.

Well, for once, I hope that I might have more energy again in the second half of the year - the first will still be very stressful. I also hope that more islands of non-LLM internet might appear, but I am afraid that "vibecoding" is here to stay. At the same time, I believe strongly that our current "AI" is a bubble and must pop sometime, and I am very curious how the world will look then. I do not believe that "free AI for everyone" is sustainable.

I want to continue my language learning activities and I also want to do more programming projects that I write about. I would love to code a C64 game and write about it here, for example. Doing original stuff for fun and not plainly for my growth as a professional developer or something like that (as determined by my manager or likewise).

We will find out as time passed. That said: Have a good 2026 and find your own joy!