Gemini* and Gopher

• Personal/Opinions

  • Happy Easter! 🐰🥚🌸

    It's that magical time of year when chocolate is totally acceptable for breakfast and a mysterious rabbit somehow has a better delivery system than most couriers.

    I hope your Easter is filled with joy, laughter, and at least one egg you forgot about and will find in July.

    May your day be bright, your coffee strong, and your chocolate stash hidden from relatives.

  • Another Post By Bone

    Hey, it's me again. I don't typically write but I really should. Writing is fun. I think I'm gonna write more often starting with this one.

  • i am a bee landlord now

    It is a truth universally acknowledged that a small landowner with a half-dozen projects already underway is in want of more projects.

    It's garden start season, which means my basement is full of seedlings under grow lights, my chickens are released daily to scratch and turn the garden beds, I have three apple trees to pick up next week, and I'm perpetually behind on the mowing/raking/mulching/composting front. So of course I took some time out from all this to become a bee landlord.

    Beehives, for honey purposes, aren't something I can do. The township will allow me to have them, and my neighbors (probably) wouldn't even complain. But I am one of the approximately 2 million people in the US who are allergic to bee stings, and beekeepers *do* get stung.

  • I'm a flexivegan

    I have become vegetarian years ago. It had been a long time before that, that eating meat bothered me. Granted, it tastes good. But is it worth it?

    I have always been sensitive to everything related to torture, existential dread, helplessness, all this stuff. I can't stand living a pitiful life, where I could not escape, where I would have to suffer endlessly from my situation until I'd die. This is part of why I struggle with working full-time and especially in this economy. I feel the same for animals. They are born in prison, raised until adolescence if lucky enough, in the worst possible conditions, tortured on a daily basis and finally slaughtered. All without the single chance of escaping or even to be able to commit suicide. They have to endure it until a human decides it's time to die.

  • dream girl

     what's your ideal woman like      
                your dream girl             
              who do you dream of           
               waking up next to            
                 I want to ask              
               to torture myself            
                   but won't
    

• Algorithms

  • a voxel engine, part 1: the core algorithm

    I have a soft spot for algorithms that do one thing only, but do it well.

    This engine starts from a classic voxel-space terrain renderer: a heightmap, a colormap, and a screen filled column by column. Each map position stores one terrain height and one color. That is restrictive compared to full 3D, but also the source of its efficiency.

  • a voxel engine, part 2: making the math cheaper

    The reference version of this renderer is mathematically clean. It uses floating point, recomputes values per sample, and does not worry too much about where the time goes.

    That is a good place to start, but not a good place to stay.

    On constrained hardware, the question is not whether something works, but whether it works cheaply enough. And in this algorithm, the cost is dominated by one thing: stepping through the map and projecting heights for every column and every distance layer.

  • a voxel engine, part 3: reducing resolution on purpose

    A small device with a 1-bit display and limited resolution. No grayscale, no soft transitions. Everything has to work within that.

    Instead of trying to approximate a higher resolution image, I reduce the problem to something that fits the display.

    The renderer already works column by column. From there it is a small step to operate on larger blocks instead of individual pixels. Effectively, the resolution is reduced by a factor of two in each direction.

  • a voxel engine, part 4: making depth visible

    The basic renderer does not store depth. It draws the terrain front to back and keeps track of the highest pixel per column. Everything behind that is skipped. For a heightmap, this is usually enough.

    The problem appears once the renderer is no longer strictly uniform. To stay fast, it advances in depth spans. Parts of the map are skipped, others sampled more densely. As a result, there is no strict relation between a screen position and a single world position anymore.

• Technology and Free Software

  • Re: Parlor Trick

    I did manually feed Sean's post to ChatGPT so that it could have the full context, and it did pick up parlor trick from that. I wanted to see what my AI prompt injections would do had the AI been given the context of both Sean's post and the LLMs reply to it. The instructions I did didn't always fool the LLM, because apparently some LLMs have been trained on recognizing instructions in data, or they use certain tricks to try to distinguish data from instructions, etc. But there are different tricks to try to get around these, according to what I've read online. I hate that we have to do prompt injections now just to protect ourselves from AI, but hopefully we can find a way to do them so that it's not so disruptive to the reading experience.

  • Some comments about my being called out by an LLM and other random links about LLMs

    Lionel Dricot [1] sent me an email about yesterdays' post [2], saying it was more plausible to him that the author was a human playing an AI (Artificial Intelligence) (technically an LLM (Large Language Model)) rather than an AI. Then, I came across more comments [3] about it where the discussion turned to it being an LLM with significant human control behind it, maybe as part of someone's sociology thesis. Both are plausible, and it's hard for me to figure out which is more plausible, an LLM-based chatbot autonomously running with internet access [4], or one with more human agency behind it. I don't know which scenario is worse.

  • Internet/Gemini

    • Dear reader, I hate you

      Truth is, I don't hate you personally, I just hate the idea of you.

      And that is perhaps the reason I haven't been writing, despite having a burning desire to write. I just don't want you in on what I've been doing and thinking. At the core of this all is the fact that I keep writing in english. Whenever somebody asks me what languages I speak I always say: Spanish, some Chinese, and some Russian. I always get the same reply: "what about english?", to which I answer: "I'm trying to forget all about english."

    • “Twitter (I refuse to call it X)”

      It’s a pet peeve of mine when people still call it Twitter (or variants like Xitter). The exact phrasing “Twitter (I refuse to call it X)” isn’t even a callout to a specific person since I’ve seen that exact text string multiple times by multiple geminauts.

      I’m not opposed to appellation activism—I try to always get #ChangeTheName in there if I have to refer to Apache or GIMP. I wish the ill-fated Glimpse project had only been an Icecat-like rename branch. That would still have been an enormous undertaking but with slightly easier merges. Glimpse would've been a great name for this image editor. I was not onboard with their UI changes. And I liked it when people on Masto took to calling Twitter “Birdsite” (in the pre-Musk era).

• Programming

  • An Elegant Arctangent Formula

    This strikes me as extremely odd. The arctangent of 1 is simply π/4, but arctan(2) and arctan(3) don't seem to be special; they don't even have closed-form expressions. What's special about these numbers, such that their arctangents sum to π, of all things?

    The answer is surprisingly simple. Let's first look at the formula for adding two arctangents together. We can derive this formula from the sums of sines and cosines.

  • a voxel engine, part 5: water without simulation

    The terrain is not just land. It also contains water. Islands need coastlines, valleys suggest rivers, and once those are present, structures like dams become relevant.

    A full simulation would require additional data and a different update model. Instead, water is part of the existing map, defined by color. There is no separate structure and no global sea level.

    The only rule is local: a cell may become water if a neighbor is water and the height allows it. This is enough to form connected regions. Low areas fill, higher areas remain dry.