AmbitiousProcess (they/them)

  • 0 Posts
  • 11 Comments
Joined 1 year ago
cake
Cake day: June 6th, 2025

help-circle



  • I would be concerned about language collapse (i.e. models incrementally forget vocabulary + fail to adapt to new terms leading to language becoming incrementally worse) before I would worry about knowledge collapse

    Why so?

    The exact same mechanism is on display here. Ask an LLM for information, and it’s more likely to condense an answer, not display additional relevant information, and slowly but surely produce less and less outputs including that given information.

    Sources then begin to be generated by the AI model that adds less and less of that information over time, LLMs cite those new sources that don’t include that information, etc, etc, and now that information is just not present in a majority of sources, and is never stated in any LLM responses.

    If anything, this is more likely than models forgetting certain vocabulary, because while the underlying tokens for a lot of that vocabulary still exist in the model, the subsequent combinations for a given piece of information are substantially less likely, and the models’ finite size can only store a subset of that actual information in statistical averages. (i.e. you can store the alphabet and a dictionary of every word in the English language, and many simple phrases, in a tiny fraction of what the sum of Wikipedia is, and Wikipedia itself will contain a subset of all information on the internet, and so on, but you can still theoretically write any English content you’ll find on Wikipedia or the internet using that alphabet and dictionary)



  • You could make a super long argument against this but I think there’s a few simple points (that you can use any number of, you don’t really need all of these at once for most people) that usually get it across to people, at least from my experience.

    There’s the moral/emotional argument:

    “You do have something to hide. Wanna unlock your phone and let me go through all your messages/photos/emails, leave the bathroom door open all the time, and never close your blinds? Even if you aren’t doing something illegal AND have no problem with me seeing all your conversations and data, privacy is still a comfort for many people”

    …and there’s the monetary arguments:

    “If a company is collecting that data, it’s because they make money off it. It’s a multi-billion dollar industry, and it’s the only reason Google, Facebook, etc can physically exist. If they know more about you, they can manipulate you into spending money with advertisers easier, allow political campaigns to manipulate your opinion more effectively, etc”

    “You might not have a problem with how your data is used now, but if the law changes, or a corporation decides to do something different, suddenly data about your health and finances that was just for ads might then become a way to increase your health insurance pricing to the max you can afford”


  • Entirely lacking disk encryption for typical users due to not having a secure element is a pretty major flaw.

    It’s also missing hardware accelerated virtualization which is necessary for much of GrapheneOS’s sandboxing, has weak security for other keys in the OS keystore, is missing hardware memory tagging which makes it much easier for apps to use overflow attacks, doesn’t have proper verified boot support once a custom alternative OS is flashed, and leaves exposed debugging APIs even when the phone is locked.

    This breaks:

    • Secure app spawning
    • Memory corruption protection
    • Integer overflow protection
    • Most of Graphene’s kernel hardening
    • Much of Graphene’s attack surface reduction abilities
    • Hardware-based attestation and security monitoring
    • Quick tile protection pre-unlock
    • Debugging access prevention
    • Verified Boot
    • The security of your PIN against any automated attack

    At that point, GrapheneOS can’t physically provide you essentially any security anymore.





  • 3D prints are almost always worse than injection moulded plastic

    Entirely depends on your use case. 99% of the time it won’t matter to most people with a 3D printer.

    If I’m making a Gridfinity setup for my desk, it doesn’t matter if it’s injection molded, it sits in my desk and doesn’t come into contact with any food or liquids and isn’t put under high stress. If I’m making attachments for my IKEA Skadis board, it doesn’t matter if they’re injection molded, nothing I’m putting on them needs highly scratch-resistant plastic or marginally heavier weight capacity. If I’m making replacement mounting for a bar on an item I own to make it more convenient to use, it doesn’t need injection molding to work just as effectively as an injection molded part, it’s holding a few grams of weight pulled by gravity and has no other stresses or contact with food or liquids.

    Injection molding is only necessary if:

    • You need a more food-safe part without layer lines
    • You need a part that is absolutely as strong as it physically can be with that plastic material, even if it will only be slightly stronger than the 3D-printed part’s layer bonds
    • You need to mass produce an item at scale very quickly, and have the upfront demand to cover the very high cost of tooling, custom molds, (thousands of dollars per individual mold) logistics, etc.

    If I wanted molds made for all the things I’ve printed to have them injection molded instead, I’d be in debt hundreds of thousands, if not millions of dollars.

    Not a single thing I’ve printed would have been cheaper, truly better quality (unless I had simply picked a badly made design in the first place), or more usable if it was injection molded, and I’ve been printing things regularly for the past 8 years.