elasticella

I am very bad at sticking to one topic per post. At the same time, I expect myself to have one clear topic per post. My drafts are many and my posts are few. Posting on Mastodon usually gets around this because my thoughts on a topic don’t have to be complete for me to feel like posting them, but (I rarely post there either and) my usual instance has been down. So, I’m back here.

Here’s what I’ve been working on.

Since I refuse to use LLM services personally, getting a job where these things are everywhere has been a whirlwind for me. There are so many little things that surprised me about how people here use and talk about LLMs. I learned how to download and host one locally so I could poke it around for a bit and figure out what the hell people are talking about. Unfortunately for me, setting up these things has some of the same draw as customizing a website or modding a video game. I ended up getting pulled into that for a bit. You can spend so much time comparing models, tweaking parameters, adjusting prompts… After the novelty wore off, I got tired of it pretty quickly and dropped it. Despite my temporary interest, my final opinion on these things is even more negative than before.

The good thing that came out of all that was the urge to double down on actually doing creative things. My Fandom Trumps Hate progress has been slow but I’m now at the last stage before my art is finished and ready to send to my bidder. The zine that I’ve horrifically over-scoped is also moving slowly, but it is moving.

I also made the self-gratifying decision of learning game development with a game engine specifically for Rust (Bevy). It is said to update with breaking changes about every three months. The concepts aren’t all new to me because I’ve worked with GLFW in C++ for game development, but I’ve never done any visuals in Rust and I’ve never even heard of ECS before. Maybe it’s a horrible choice. Maybe it’ll push me to do things with reasonable scopes and time frames, since they’re all learning projects for now. I just spent a couple hours the other night developing a system where you can click on a dialogue box to read the next bit of dialogue. Most game engines probably give you that out of the box. Still, I’m proud that I figured it out myself.

In the field of note-taking, I’ve been thinking about how to better incorporate fiction into my notes. I have no real place to save character notes or notes on my personal taste. My system was designed for studying. Stories appear intact in the input and output, but the process through mangles them and leaves them empty. They can’t be converted to a series of claims and principles and still emerge recognizable as a story. My previous struggle with this was my refusal to reserve a section of my notes specifically for dealing with fiction. The Zettelkasten system is not designed for sectioning. The idea is that not separating notes by topic makes it easier to draw conclusions across multiple topics. However, I’ve since accepted that (1) a single folder full of everything is horrible to look at and (2) some things simply have their own rhythm that can be better represented with a specialized structure.

So. I now have a new folder for fiction. I’m calling it “world-building” for now, and tagging its contents as various “story elements.” I expect either for the terminology to change or for me to forget it was ever a choice. So far, working with it has already been noticeably different from the usual ZK process. The ZK is about breaking down media and experience into evidence and theories, then collecting notes with recurring themes into a conclusion. This has been more like putting down a large outline, sectioning it into segments, and filling it in piece by piece. Top-down instead of bottom-up. The whole is the primary unit, instead of whole media only being the input and output with a sea of small notes in the middle. We’ll see how it goes for creating new things instead of just migrating old story notes (from memory, because I seem to have lost much of them).