Lettuce eat lettuce

Always eat your greens!

  • 1 Post
  • 137 Comments
Joined 3 years ago
cake
Cake day: July 12th, 2023

help-circle
  • I agree with Stallman and others on this sort of thing, it should be illegal for any public institution to use proprietary, corporate software in any capacity.

    Public institutions are supposed to serve and benefit the public good first and foremost. The idea that they can do that equally well or better by partnering with mega-corps who’s only fundamental drive is to generate as much profit as possible for their shareholders, is idiotic.






  • I once worked as a sysadmin at a Windows-only small business (a few dozen employees total.)

    I never had an issue switching to Linux on my laptop. I used Reminnia to RDP into all our Windows servers. I used the browser versions of Outlook, Teams, and Word. All our network devices and VMWare host had web interfaces that I accessed from my browser. Our Entra permissions for email and our cloud-side stuff was all in the browser too.

    I used some FOSS PDF reader/editor for PDFs, printing worked without any tweaking. I had a basic Windows VM on my system for the one Windows-only program that I occasionally needed to test.

    But of course, being the entire IT department by myself meant that I could do that without getting approval from anybody other than the owner, and he said he didn’t care what OS I used on my laptop, as long as it didn’t hurt my ability to keep everything up to date, secure, and our users happy.







  • Absolutely correct, Capitalism will incentivize innovation only as long as it’s profitable to do so. Once a company becomes a dominant player in their industry, innovation actually becomes largely unprofitable.

    We can see this play out all the time, this is why industry giants almost never try to outcompete smaller start ups by innovating their products and services, instead, they either crush them out of the market via a bunch of different methods. Or they just buy them out, kill off their IP, or if it’s way better, they just brand it as their own, and usually fire most of the old staff.

    In a Capitalist system, this actually creates a whole new incentive structure where start ups will try to grow as fast as possible with the sole intention of getting bought out by an industry titan, making the founders of the start up filthy rich.



  • If you’re using a 30+ year old computer in the modern day, you’ve already sacrificed security.

    The price and availability of a 15 year old computer is almost the same as a 30 year old computer, you can get a fully functional 1st gen i5 PC that will sprint circles around the highest end 90’s PC for $40. Honestly, most people are just e-wasting computers that old, I’m sure you could easily get them for free, hell, lots of people would pay you to take them off their hands.





  • If there’s anybody I trust to use LLMs in a responsible way, (to whatever degree that’s even possible) it would be Linus Torvalds. He has some of the highest standards for software development in the world, and the same can be said for the core Linux kernel development team.

    The general problem though, is that the people who are the least qualified to do software development are very often the people who are the most likely to use LLMs to try to slop their way to the top.

    I think that the combination of cost increases for inference, inherent limitations in model integrity, and the economic destruction that will ensue when these fraudster CEOs have finally squeezed every last drop of blood from every last stone they can find, will leave “AI” in a similar place as blockchain/crypto/NFTs.

    There will be a relatively niche market for software engineers and related tech workers; local LLM boxes, some subscription-based models that companies pay for with their licensing packages, hobbyists, and some minimal integrations into software suites for tools like background removal, lighting enhancement, etc.

    There will be no Singularity, no AGI, no all-knowing Machine God.


    1. Check your thermals on your CPU and GPU, make sure they aren’t running too hot.
    2. Turn off your computer and unplug it from the wall. Then check all your physical connections from the power supply to the motherboard and your graphics card. Everything should be fully seated in the sockets and firmly locked in. Loose connectors can cause power delivery issues that freeze or crash a system. In rare cases they can short out, causing severe damage to the components, or even a fire. So it’s important to do anyways.
    3. Check your motherboard manufacture’s website and make sure you’re using the latest BIOS version. Your BIOS screen will tell you what your exact model number is and the current BIOS version.
    4. Check to make sure you’re not running out of system RAM.
    5. Use Furmark to stress test your GPU. If it runs in stress test mode for at least 4 hours straight without any crashes, that’s a good sign that your GPU isn’t the issue.
    6. Use a CPU stress test program to do the same to your CPU. I actually like to go to Lichess.org, open their analysis board tool Lichess Analysis Tool, go into the settings and max out the move search depth, memory limit, and maximum lines calculated. Then I just make 15-20 random moves. The chess engine will sit there and calculate as long as you have that tab open on your local machine, and it will max out every core on your CPU 100% but there are dozens of other ways to stress your CPU, so feel free to search for them.
    7. You may have a lemon of a CPU or GPU chip. All chips are “binned” meaning they are tested at the factory and depending on how they perform, they are accepted or rejected within certain boundaries, but there is a range. Some chips just barely pass the minimum standards brand new. Since all electronics degrade over time, especially modern chips with billions of microscopic transistors, if your chip already was barely good enough to pass the minimum specs when it was fresh out of the box, it can start to become unstable even under normal workloads, let alone more heavy use like what you’re describing. Undervolting and underclocking can help fix this. Using your motherboard’s settings or 3rd party tools (like LACT for GPUs), you can limit the maximum voltage and/or maximum frequency of your chips. This not only will reduce the thermal output of your chips, making them run cooler and slowing the degradation process, but often will make them run much more stable.

  • Because many tech jobs have become glorified vendor management. You don’t actually support anything yourself, you have a vendor sales rep and a technical rep or maybe two.

    When something breaks or needs an update, you call/email the rep, go back and forth a few times, schedule a date for patching, and then create your 10,000th email Chain/Teams group chat with all the various people internally and externally who “need visibility” into the process. Everybody says their piece, argues about coverage or pricing, asks the same basic questions 5 different ways, so on and so forth.

    Now multiply this by all the unique software/hardware components of your infrastructure. Add to that the AI executive-led slop fest, angry sarcastic, tech-illiterate end users, corporate software enshitification, etc.