This is the first real actual wordvomit for my new goal. Let's see how it goes.

I tried writing yesterday's on Kitty, and while it worked fine I noticed many bugs with the markdown editor. Foremost among them is that the cursor tends to shift its vertical position when the post gets somewhat long. Not sure why. I use an underlying library that handles the actual implementation of the editor, so it could very well be that has a bug and I need to update it. But I won't do that now.

I'm at the beach till pretty much the end of the week. I only took my phone with me so that means no programming of any kind. I thought about bringing the PC as well but then decided against it. I know that even if tell myself I would bring it just to write there would always be the temptation to use it for something else. God forbid I get the urge to use it to check work stuff while I'm away on vacation! I left some experiments running for a whitepaper we're writing, so it's very likely I would've talked myself into just popping in for a second to see how they were doing. We all know that “just a second” can easily become an hour or more if we let it.

…

I was thinking that these vomits are not about what kind of thing “can I write today”, but about the kind of thing I see myself writing in fifty or a hundred days from now. Perhaps I'm thinking too much into the future, but supposing I actually manage to stick with this then what will my vomits (I may try to find a better word for them) look like. Will they be as rambly? Will the discussion seem as convoluted? Again, I'm really writing these just for myself. It's a sort of public journal I guess. The fact it's public helps give accountability AND pushes me to try and write better, but the goal is not to appeal to a wider public.

Another thing I thought is that this is very much like Balugust, where the goal is to get something out and published every day, no matter what it is. But why do we need to wait for blaugust to do that? Usually I abstain from publishing this kind of freer form writing because I don't think anyone is interested in reading it. In my mind it's more akin to an LLM generated text than anything else. Though perhaps that's a mistaken assumption, as LLMs really just tend to sound businesslike or sycophantic. Maybe this kind of writing, as convoluted as it is, is actually as human as it gets? Anyway, I'll make a different RSS feed on my blog for this, so at least people can decide if they care about these vomits or not. Maybe, with time, they'll start coming out in more coherent shapes, and might be of more popular interest. These first few ones are just for breaking the ice so to speak, and getting used to the medium. They'll stay around mostly for documentation purposes.

…

Before I sat down and began writing this entry I actually had some ideas going around in my mind for things I thought would be fun to talk about. However, it seems I digressed even before starting and never got to any of them in this post! Maybe my brain needs a little bit of a “writing runway” before being able to actually take off and feel comfortable with what it's doing.

I realized these vomits can be anything, from semi-stream of consciousness, like it's been so far, to simpler discursive posts. So yeah, I think initially I'll keep a list of ideas and maybe talk about them at some point.

It would actually be fun to keep, at the end of every one of these posts, a list of ideas that I still care about talking, brought over from previous days, as well as ideas for the current day. Not only is it a way for me to remember them, but it's also a sort of implicit documentation system to track what I was thinking about during a given period of my life.

While this sounds fun, it also sounds like a lot of work, and I don't want to add unnecessary friction just yet. First and foremost the goal is to get used to the process. To build the habit. After I'm settled in I can start adding embellishments.

Oh look at that! I'm already at 700 words. Well I guess I'll just post my ideas as a list then, so I can track them somewhere.

• I work with LLMs at my job, but I never really used them for anything but work. Lately (this week) I've been exploring with role-playing with LLMs. Creating fictitious characters that have certain goals for me. This came about after reading a short post from someone who saw their OCD scores go down after they started using LLMs for this sort of thing.
  • For me, the idea is that such a character is basically an interactive book that you can talk to and it can offer advice in return.
  • There's also the fact that (for some people like me) it's easier to talk to an LLM about personal stuff than to an actual person. It might not be ideal, and maybe it's even a bit sad, but not having to think about all the social norms is great. I know I don't have to worry about the LLM pitying me or getting frustrated that I'm acting childishly.
• I've been working on a silly new project that's a honeypot for dishonest web scrapers. It has all the required rules that tell respectful scrapers to ignore the site, so it will only trap “bad” ones. All the content on the site is generated on the fly, so it's basically infinite. And it's all garbage text so LLMs using it for training will very likely be broken and misbehave. I had fun working on this and I thought it would be interesting to talk about how it works.
• I also thought it would be nice to make a sort of TIL section! For instance, today I learned that jeans were actually invented in Genoa. I think… I need to check but I'll maybe talk about it tomorrow as today's entry is already long enough.