• 2 Posts
  • 229 Comments
Joined 2 years ago
cake
Cake day: April 23rd, 2023

help-circle
rss
  • 3D-printed shoes could be a great idea given how different everyone’s feet can be. It could save on transportation and logistical costs, and everyone could have shoes perfect for their feet, created much more locally than Vietnam.

    However, the cynic in me says that’s not what Nike is doing, or why - they’re doing this because it lets them cut workers. Traditional shoe manufacturing involves human hands at many process steps, often with machine assistance or other tools. This lets them cut out all of those workers and all of that equipment in favor of one machine that makes an entire generic shoe for them to shove onto shelves next to all the other generic factory-made shoes. This is not the future.


  • Also, the raw material is expected to be quite rare relatively soon.

    To be fair, this wouldn’t be nearly as true if we had persisted with our original plan which was to reprocess the spent fuel, more than 90% of which is still usable material. Once we found a couple huge deposits of Uranium, it became much cheaper to simply mine more of it and dispose of the spent fuel, so the recycling plans were scrapped. Sure, we can technically still pull the spent fuel back out again and recycle it, but we spent many years building reactors without building an equal capacity of reprocessing facilities (which are almost as hard to build safely as reactors), so that ship has more or less sailed.




  • In general, Bazzite being immutable just means the core system isn’t modular to the end user to the degree that Arch is. You of course can use flatpaks or appimages like any distro, and there are still several ways to install traditional rpm/deb/aur programs (the usual Fedora method doesn’t work because dnf doesn’t exist). If it’s just an app that doesn’t require significant integration with the OS, the recommendation is to install them into a distrobox container (where dnf does exist) and then distrobox-export [program] to make them visible to the host system. VPNs need a little more integration so those are installed by layering with rpm-ostree and then enabling the systemd service(s). Layering makes updates take longer to install so it should be avoided when possible.

    One of the interesting things about Universal Blue’s images like Bazzite is if you want the benefits of atomic while also having a more custom system than they offer without having to install a bunch of things in rpm-ostree, the process to build a custom image based on one of theirs is apparently quite easy to do and automate, though I haven’t done it myself.


  • In general, yes. Most of the difficulty is due to being on Linux and running games through the Proton/WINE compatibility layer, so there can be an extra layer of jank involved, but it’s very possible.

    If modding consists of dropping files into the game directory, it will work almost exactly the same as in Windows. However, if some of those files replace the game’s DLLs, then whatever WINE runner you use might need to be told to use the DLLs in the game directory instead of its own.

    If you need to use a mod manager, that situation is still not ideal - native Linux mod managers I know of are only the Nexus Mods app (very new, there’s some talk of it being integrated directly into the Heroic launcher) and Limo. Everything else, you’ll be running whatever bespoke Windows mod manager your game uses through Proton/WINE, probably with Steam Tinker Launch, possibly Lutris.

    tl;dr There can be an extra layer of complexity over modding on Windows, but it’s otherwise comparable.


  • During boot, you’re presented with 4 snapshots you can choose between so if an update did happen to break something, it’s easy as just choosing an older snapshot after a reboot.

    Those are actually just two snapshots, there’s a bug in GRUB that displays them twice. Purely visual, and you can fix it with a ujust script, run in the terminal with ujust configure-grub. There are lots of little scripted tweaks and installations available; you can get most of the list by running ujust by itself. Incredible work by the maintainers.







  • You’re entirely correct, but in theory they can give it a pretty good go, it just requires a lot more computation, developer time, and non-LLM data structures than these companies are willing to spend money on. For any single query, they’d have to get dozens if not hundreds of separate responses from additional LLM instances spun up on the side, many of which would be customized for specific subjects, as well as specialty engines such as Wolfram Alpha for anything directly requiring math.

    LLMs in such a system would be used only as modules in a handcrafted algorithm, modules which do exactly what they’re good at in a way that is useful. To give an example, if you pass a specific context to an LLM with the right format of instructions, and then ask it a yes-or-no question, even very small and lightweight models often give the same answer a human would. Like this, human-readable text can be converted into binary switches for an algorithmic state machine with thousands of branches of pre-written logic.

    Not only would this probably use an even more insane amount of electricity than the current approach of “build a huge LLM and let it handle everything directly”, it would take much longer to generate responses to novel queries.